CN117933882A - Intelligent logistics system based on Internet of things - Google Patents
Intelligent logistics system based on Internet of things Download PDFInfo
- Publication number
- CN117933882A CN117933882A CN202410138126.9A CN202410138126A CN117933882A CN 117933882 A CN117933882 A CN 117933882A CN 202410138126 A CN202410138126 A CN 202410138126A CN 117933882 A CN117933882 A CN 117933882A
- Authority
- CN
- China
- Prior art keywords
- cement
- matrix
- warehouse
- data
- order
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
- G06Q10/087—Inventory or stock management, e.g. order filling, procurement or balancing against orders
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Economics (AREA)
- Finance (AREA)
- Development Economics (AREA)
- Accounting & Taxation (AREA)
- Entrepreneurship & Innovation (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the technical field of the Internet of things, and further relates to an intelligent logistics system based on the Internet of things, which comprises: the logistics data construction unit is used for constructing a cement warehouse data matrix, a cement transporter data matrix, a cement order data matrix and a cement provider data matrix; the feature extraction unit is used for respectively encoding by using the read-heat codes to obtain feature vectors of each attribute vector; a cement order classification unit for distributing cement orders to corresponding warehouses; a cement transporter determining unit for distributing cement transporters to corresponding warehouses; and the cement provider binding optimization unit is used for binding the cement provider to the corresponding warehouse. The invention improves the efficiency of cement supply chain management, reduces the cost, improves the customer satisfaction degree and enhances the competitiveness through data processing and algorithm optimization.
Description
Technical Field
The invention belongs to the technical field of the Internet of things, and particularly relates to an intelligent logistics system based on the Internet of things.
Background
Cement is an important building raw material, and its supply chain management has a key role in modern building and infrastructure construction. However, in conventional cement logistics, there are still a series of problems that directly affect the efficiency and sustainability of the supply chain.
Conventional cement logistics management typically relies on periodic inventory and static inventory management policies. This results in the inability to monitor and adjust inventory levels in cement warehouses in real time, and often problems with stock backlog or shortages. The transportation of cement typically involves multiple suppliers and carriers, and information transfer and monitoring is opaque, making it difficult to achieve real-time tracking and control of the progress of the transportation. This may lead to transportation delays and losses. Cement orders typically need to be distributed to different warehouses to meet customer needs. Traditional distribution methods are often based on experience, lack of scientific data support, and tend to lead to uneven order distribution and inconsistent transportation. In the cement supply chain, it is important to select the appropriate supplier and establish a stable binding relationship therewith. Conventional approaches often lack systematic vendor assessment and binding optimization methods, resulting in a threat to the stability of the supply chain. In conventional cement logistics management, information is often scattered in each link, and different participants have difficulty in sharing real-time data and making collaborative decisions. Such islands of information cause delays in information delivery and inaccuracy in decisions.
In summary, the conventional cement logistics management has the problems of opaque information, low efficiency, difficulty in adapting to changes, high cost and the like, and the problems directly affect the operation efficiency and customer satisfaction of the cement supply chain. Accordingly, there is a need to introduce new techniques and methods to optimize cement stream management, improving the sustainability and competitiveness of the supply chain.
Disclosure of Invention
The invention mainly aims to provide an intelligent logistics system based on the Internet of things, which improves the efficiency of cement supply chain management, reduces the cost, improves the customer satisfaction and enhances the competitiveness through data processing and algorithm optimization.
In order to solve the problems, the technical scheme of the invention is realized as follows:
An intelligent logistics system based on the internet of things, the system comprising: a logistics data construction unit, configured to acquire cement warehouse data to construct a cement warehouse data matrix, wherein each element in the matrix represents an attribute vector of a cement warehouse, acquire cement transporter data to construct a cement transporter data matrix, each element in the matrix represents an attribute vector of a cement transporter, acquire cement order data to construct a cement order data matrix, each element in the matrix represents an attribute vector of a cement order, and acquire cement provider data to construct a cement provider data matrix, and each element in the matrix represents an attribute vector of a cement provider; the feature extraction unit is used for respectively encoding the cement order data matrix, the cement warehouse data matrix, the cement transporter data matrix and the cement provider data matrix by using the read-heat codes so as to convert the non-numerical attribute vector into a numerical attribute vector, then respectively carrying out feature extraction on each attribute vector in each data matrix to obtain a feature vector of each attribute vector, and finally obtaining a feature vector matrix of the cement order data matrix, a feature vector matrix of the cement warehouse data matrix, a feature vector matrix of the cement transporter data matrix and a feature vector matrix of the cement provider data matrix; a cement order classification unit for distributing cement orders to corresponding warehouses based on the eigenvector matrix of the cement order data matrix and the eigenvector matrix of the cement warehouse data matrix; a cement transporter determining unit for assigning cement transporters to the corresponding warehouses based on the eigenvector matrix of the cement transporter data matrix and the eigenvector matrix of the cement warehouse data matrix, and the result of assigning cement orders to the corresponding warehouses; and the cement provider binding optimization unit is used for building a binding model by using a fast random tree based on the eigenvector matrix of the cement provider data matrix, the result of distributing the cement order to the corresponding warehouse by cement and the result of distributing the cement transporter to the corresponding warehouse, binding the cement provider to the corresponding warehouse, and maximizing a binding model objective function by considering a multi-layer tree structure and nonlinear factors in the process of building the binding model so as to optimize parameters of the binding model.
Further, the elements of the attribute vector of the cement order include at least: cement demand, order user location data, and order urgency; the attribute vector of the cement transporter at least comprises: upper limit of cement transportation amount, position data of transportation and transportation price; the cement provider's attribute vector includes at least: supply cement price, supplier location data and supplier supply; the attribute vector of the cement warehouse comprises: maximum capacity of the cement silo, remaining capacity of the cement silo, and cement silo position data.
Further, let the cement order data matrix be:
。
Wherein the method comprises the steps of Represent the firstAttribute vectors for individual cement orders; setting a cement warehouse data matrix as follows:
。
Wherein the method comprises the steps of Represent the firstAttribute vectors of the cement warehouses; let the cement transporter data matrix be:
。
Wherein the method comprises the steps of Represent the firstAttribute vectors of individual cement transporters; let the cement provider data matrix be:
。
Wherein the method comprises the steps of Represent the firstAttribute vectors for individual cement suppliers; setting the eigenvector matrix of the cement order data matrix as; Setting the eigenvector matrix of the cement warehouse data matrix as; Let the eigenvector matrix of the cement transporter data matrix be; Let the eigenvector matrix of the cement provider data matrix be; Wherein,For the quantity of cement orders,For the number of cement warehouses,For the number of cement carriers,For the number of cement suppliers, the following constraint is satisfied:
；
；
。
Further, the cement order classification unit trains a dual support vector machine classifier by using a dual support vector kernel function based on the feature vector matrix of the cement order data matrix and the feature vector matrix of the cement warehouse data matrix, distributes the cement order to the corresponding warehouse, and uses a convex optimization algorithm when training the dual support vector machine classifier, so that the objective function of the dual support vector machine classifier reaches the maximum value to optimize the parameters of the dual support vector machine classifier.
Further, based on the eigenvector matrix of the cement order data matrix and the eigenvector matrix of the cement warehouse data matrix, training a dual support vector machine classifier using a dual support vector kernel function, the method of assigning cement orders to corresponding warehouses comprises: using polynomial kernel functionsMatrix of eigenvectors of cement order data matrixEigenvector matrix for data matrix of cement warehouseMapping to a high-dimensional space; eigenvector matrix using cement order data matrixSum kernel functionTraining a dual support vector machine classifier comprising weight parameters of support vectors、Bias termIs calculated; decision function of classificationIs used to categorize orders ifThen the order is allocated to the corresponding warehouseOtherwise, not distributing; wherein,Eigenvector matrix being a matrix of cement order dataMiddle (f)The number of elements to be added to the composition,Index for subscript; Eigenvector matrix being a cement warehouse data matrix The first of (3)The number of elements to be added to the composition,Index for the subscript.
Further, a decision function of classificationThe expression is used as follows:
；
wherein, Is the order weight parameter, which is the first in the order weight matrixThe order weight matrix is a preset matrix, and optimization adjustment is carried out through a maximized objective function; is the warehouse weight parameter, which is the first in the warehouse weight matrix The warehouse weight matrix is a preset matrix, and optimization adjustment is carried out through a maximized objective function; Is a bias term.
Kernel functionThe expression is used as follows:
。
Further, a convex optimization algorithm is used to make the objective function of the dual support vector machine classifier reach the maximum value, and the objective function is expressed by the following formula:
。
Further, the cement transporter determining unit, based on the eigenvector matrix of the cement transporter data matrix and the eigenvector matrix of the cement warehouse data matrix, and the result of assigning the cement order to the corresponding warehouse, assigns the cement transporter to the corresponding warehouse, the method includes:
The allocation score is calculated using the following formula:
；
wherein, Representing the presentation to beAnd (3) withThe operation of matrix splicing is carried out, the matrix splicing operation is carried out,Is the Frobenius norm; is L1 norm; Is that The feature vector of the corresponding warehouse to which the corresponding cement order is allocated; if it isGreater than the set first threshold value, thenCorresponding cement transporter is bound toA corresponding warehouse.
Further, the binding model established by the cement provider binding optimization unit is expressed by using the following formula:
；
wherein, For the calculated vendor binding score, ifGreater than a set second threshold value, thenThe corresponding vendor is bound toA corresponding warehouse; is a parameter of the binding model.
Further, the binding model objective function is maximized to optimize the binding model parameters by the following formula:
；
wherein, Representation fetchIs a minimum value in (a).
The intelligent logistics system based on the Internet of things has the following beneficial effects: the dual support vector machine classifier can distribute cement orders to proper warehouses with high precision, and unreasonable order distribution is avoided. The convex optimization algorithm is adopted to optimize the parameters of the support vector machine classifier, so that the support vector machine classifier can rapidly decide order allocation, and the decision efficiency is improved. By building a binding model, the system is able to intelligently select suppliers and bind them to the appropriate warehouse, thereby improving the stability and sustainability of the supply chain. The binding model takes into account the multi-layer tree structure and non-linear factors, making it more adaptable to complex supply chain environments. The high-precision order classification reduces the occurrence of false orders, and reduces the order processing cost and customer satisfaction. The optimized order distribution reduces unnecessary transportation, reduces transportation cost and improves benefit. The logistics data construction unit can acquire data of a cement warehouse, a transporter, an order and a supplier in real time and construct a corresponding data matrix. This ensures timeliness and accuracy of the data. The feature extraction unit converts the non-numerical attribute vector into a numerical attribute vector by using techniques such as reading thermal coding and the like, and extracts important feature information. This helps to better understand and analyze the data.
Drawings
Fig. 1 is a schematic system structure diagram of an intelligent logistics system based on the internet of things according to an embodiment of the present invention.
Detailed Description
In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
Example 1: referring to fig. 1, an intelligent logistics system based on internet of things, the system comprising: a logistics data construction unit, configured to acquire cement warehouse data to construct a cement warehouse data matrix, wherein each element in the matrix represents an attribute vector of a cement warehouse, acquire cement transporter data to construct a cement transporter data matrix, each element in the matrix represents an attribute vector of a cement transporter, acquire cement order data to construct a cement order data matrix, each element in the matrix represents an attribute vector of a cement order, and acquire cement provider data to construct a cement provider data matrix, and each element in the matrix represents an attribute vector of a cement provider; the feature extraction unit is used for respectively encoding the cement order data matrix, the cement warehouse data matrix, the cement transporter data matrix and the cement provider data matrix by using the read-heat codes so as to convert the non-numerical attribute vector into a numerical attribute vector, then respectively carrying out feature extraction on each attribute vector in each data matrix to obtain a feature vector of each attribute vector, and finally obtaining a feature vector matrix of the cement order data matrix, a feature vector matrix of the cement warehouse data matrix, a feature vector matrix of the cement transporter data matrix and a feature vector matrix of the cement provider data matrix; a cement order classification unit for distributing cement orders to corresponding warehouses based on the eigenvector matrix of the cement order data matrix and the eigenvector matrix of the cement warehouse data matrix; a cement transporter determining unit for assigning cement transporters to the corresponding warehouses based on the eigenvector matrix of the cement transporter data matrix and the eigenvector matrix of the cement warehouse data matrix, and the result of assigning cement orders to the corresponding warehouses; and the cement provider binding optimization unit is used for building a binding model by using a fast random tree based on the eigenvector matrix of the cement provider data matrix, the result of distributing the cement order to the corresponding warehouse by cement and the result of distributing the cement transporter to the corresponding warehouse, binding the cement provider to the corresponding warehouse, and maximizing a binding model objective function by considering a multi-layer tree structure and nonlinear factors in the process of building the binding model so as to optimize parameters of the binding model.
Specifically, the logistics data construction unit first collects various data from a plurality of data sources (such as cement warehouse, cement transporter, cement order, and cement supplier), which may include various information of logistics transportation such as the number of goods, transportation time, supplier information, warehouse capacity, and the like. The collected data is integrated into a unified data structure, typically a matrix, in which each element represents a vector of attributes of a particular entity (e.g., a cement warehouse, a carrier, an order, a supplier). This attribute vector contains various information about the entity, such as geographic location, shipping history, capacity, etc. The main purpose of the logistic data construction unit is to integrate data from different sources into one unified data structure. This helps to eliminate data islanding and fragmentation so that the system can process and manage all relevant data at one central location. Such an integrated view of data helps to improve the accessibility and usability of the data. After integrating the data into the form of attribute vectors, the logistics data construction unit may preprocess the data to ensure the quality and consistency of the data. This may include data cleansing, deduplication, missing value processing, etc., to improve the accuracy of subsequent data analysis.
The first step of the feature extraction unit is to encode the data. A commonly used encoding method is one-hot encoding (one-hotencoding), which converts non-numeric attribute vectors into numeric attribute vectors. This process converts each non-numeric attribute into a binary vector where only one element is a 1 and the remaining elements are all 0. This 1 position represents the value of the original attribute so that these vectors can be used in mathematical calculations. Once the data is encoded into numerical attribute vectors, the feature extraction unit uses various feature extraction methods to extract useful information from these vectors. These feature extraction methods may include statistical methods, frequency analysis, dimension reduction techniques (e.g., principal component analysis), time series analysis, and the like. The goal is to abstract the information in the raw data for subsequent analysis and decision making. Non-numerical properties are generally not directly useful for mathematical computation and machine learning model training. By means of the one-time encoding, the feature extraction unit converts these non-numeric attributes into numeric form, rendering the data processable. In this way, the system can more easily apply various analyses and algorithms. Feature extraction may also be used to reduce the dimensionality of the data, thereby reducing computational complexity and storage requirements. By preserving the most important features, the data can be simplified without losing too much information, improving the efficiency of analysis and modeling.
In the cement order classification unit, first, a feature vector of the cement order data is generated using the feature extraction unit. These feature vectors are numerical representations of the order's attributes that contain various information about the order, such as the weight of the order, the destination, the time stamp, etc. The cement order classification unit uses a classification algorithm to classify the order. These algorithms may include traditional machine learning methods (e.g., decision trees, support vector machines, random forests, etc.) or deep learning methods (e.g., neural networks). The choice of classification algorithm generally depends on the nature of the data and the complexity of the problem. The main function of the cement order classification unit is to distribute new orders to corresponding cement warehouses according to the feature vectors of the orders and the trained classification model. This is a very important step in a logistics system because proper order distribution ensures that the supplies arrive at the destination on time, reducing inventory backlog and transportation costs. By automating the order sorting process, the system can more quickly process a large number of orders, reducing the need for manual intervention. This improves the efficiency of logistics management and reduces the risk of human error.
Similar to the cement order classification unit, the cement transporter determination unit first needs to generate a feature vector of cement transporter data using the feature extraction unit. These feature vectors are numerical representations of the attributes of the transporter, including historical performance of the transporter, vehicle information, delivery routes, and the like. The cement transporter determination unit uses a correlation analysis or matching algorithm to match the cement order with the most appropriate transporter. These algorithms take into account order characteristics, characteristics of the transporter, and possible route and delivery constraints. The main function of the cement transporter determination unit is to distribute cement orders to the most suitable transporter. This helps to optimize logistics transportation, ensuring that orders are delivered to the destination in the most efficient and economical manner. By matching orders with the most appropriate carriers, shipping costs can be reduced, shipping time can be reduced, and punctuality of delivery can be maximized. The unit takes into account a number of factors such as the historical performance of the transporter, the type of vehicle, the current location and the order requirements. The comprehensive consideration of the factors enables the order distribution to be more intelligent, and the method can be suitable for optimal decisions under different conditions. Automated transporter determination processes reduce the need for human intervention. This means that the system can respond more quickly to the arrival of new orders and can match more efficiently in the case of large-scale orders.
The cement provider binding optimization unit uses a binding model to determine which cement silo each provider best matches. This binding model is typically built based on a machine learning method, possibly using algorithms such as a fast random tree (RandomForest). The goal of the model is to maximize the parameters of the binding model to optimize the matching of the vendor to the warehouse. The primary role of the cement provider binding optimization unit is to determine which cement warehouse each cement provider best matches. This helps ensure that each warehouse gets the most appropriate supplier to meet its needs and maximize the efficiency of the supply chain. By binding each vendor to the most appropriate warehouse, the system may reduce unnecessary transportation and inventory costs. This helps to increase the efficiency of the overall supply chain, reducing the overall cost of logistics management.
Example 2: the elements of the attribute vector of the cement order include at least: cement demand, order user location data, and order urgency; the attribute vector of the cement transporter at least comprises: upper limit of cement transportation amount, position data of transportation and transportation price; the cement provider's attribute vector includes at least: supply cement price, supplier location data and supplier supply; the attribute vector of the cement warehouse comprises: maximum capacity of the cement silo, remaining capacity of the cement silo, and cement silo position data.
In particular, the integration and analysis of these attribute vector elements enables a more efficient, intelligent logistics system to achieve more intelligent logistics management. The system may select the appropriate suppliers and carriers based on order demand, urgency, and user location data to minimize costs and ensure timely delivery. Meanwhile, the information of the accommodating quantity of the warehouse is beneficial to the dynamic management of the warehouse, and unnecessary stock backlog is reduced. By comprehensively considering these elements, the intelligent logistics system can provide more optimized logistics decisions, thereby improving the efficiency of the supply chain and customer satisfaction. The comprehensive information integration and analysis process enables logistics management to be intelligent and automatic, and provides competitive advantages for enterprises.
Example 3: setting a cement order data matrix as follows:
。
Wherein the method comprises the steps of Represent the firstAttribute vectors for individual cement orders; setting a cement warehouse data matrix as follows:
。
Wherein the method comprises the steps of Represent the firstAttribute vectors of the cement warehouses; let the cement transporter data matrix be:
。
Wherein the method comprises the steps of Represent the firstAttribute vectors of individual cement transporters; let the cement provider data matrix be:
。
Wherein the method comprises the steps of Represent the firstAttribute vectors for individual cement suppliers; setting the eigenvector matrix of the cement order data matrix as; Setting the eigenvector matrix of the cement warehouse data matrix as; Let the eigenvector matrix of the cement transporter data matrix be; Let the eigenvector matrix of the cement provider data matrix be; Wherein,For the quantity of cement orders,For the number of cement warehouses,For the number of cement carriers,For the number of cement suppliers, the following constraint is satisfied:
；
；
。
Specifically, first, there are four key data matrices: cement order data matrix Cement warehouse data matrixCement transporter data matrixAnd cement provider data matrix. Each data matrix contains attribute vectors for the corresponding entities. For example in a matrix of cement order dataRepresent the firstAttribute vectors for each cement order, including demand for the order, user location data, and urgency. Also, the attribute vectors in the other matrices include important information of the corresponding entities. Then, a eigenvector matrix is introduced, representing eigenvectors of the cement order, cement warehouse, cement carrier, and cement supplier data, respectively. The eigenvector matrix、、And) The construction of the system can convert the original data into the characteristics with more information content by various characteristic extraction methods based on the original data matrix. This helps the system to better understand the data, make decisions, and optimize. In addition, constraints are introduced to limit the size and relationship of the data matrix. These constraints ensure reasonable scale and balance of data. For example, the number of cement warehouses is required to be not less than the number of orders, but not more than 100 times the number of orders, which helps to maintain a reasonable scale of data. Also, the number of suppliers is at least one tenth of the number of warehouses, which helps to ensure coverage of the suppliers. In general, this embodiment provides an efficient method to structure and organize critical data for cement logistics systems. Not only does this matrix representation make the data easier to manage and analyze, it also provides the system with more intelligent decision-making and optimization opportunities. Through the constraint of feature extraction and data matrix, the system can better understand the logistics environment, improve efficiency, reduce cost and ensure on-time delivery. This is an important technique and method for modern logistics management, helping to improve the sustainability and competitiveness of the supply chain.
Example 4: the cement order classification unit is used for training a dual support vector machine classifier by using a dual support vector kernel function based on the feature vector matrix of the cement order data matrix and the feature vector matrix of the cement warehouse data matrix, distributing the cement order to the corresponding warehouse, and using a convex optimization algorithm when training the dual support vector machine classifier, so that the objective function of the dual support vector machine classifier reaches the maximum value to optimize the parameters of the dual support vector machine classifier.
Specifically, the support vector machine is a supervised learning algorithm for solving the binary or multivariate classification problem. Its main objective is to find a decision boundary or hyperplane, separate the different categories of data and maximize the separation (i.e. the distance of the nearest data point from the hyperplane). The SVM searches for the optimal classification hyperplane by maximizing the interval, and the hyperplane can correctly classify the data and has good generalization performance. Dual support vector machine (D-SVM): D-SVM is an extension of SVM for multi-class classification problems. It handles multi-class situations by combining multiple SVM classifiers into one overall system. Each SVM classifier is used to distinguish one class from all other classes, so there are multiple SVM classifiers in the multi-class problem. SVMs and D-SVMs typically use a kernel function to address the nonlinear classification problem. The kernel function may map the input features into a high-dimensional space such that the non-linearity problem becomes linearly separable in the high-dimensional space. In this embodiment, the dual support vector machine uses a dual support vector kernel function to handle the non-linear nature of the data.
The main role of the D-SVM is to distribute cement orders to the corresponding warehouses. Each warehouse is considered a category and the D-SVM learns a classifier for classifying orders into these different categories, i.e. assigning orders to the corresponding warehouse. The D-SVM is a highly accurate classifier that can accurately classify orders by learning and optimization. This helps ensure that orders are assigned to the correct warehouse, thereby reducing errors and confusion. The D-SVM can intelligently handle multiple categories of questions, which take into account the characteristics of each order and the attributes of the individual warehouse to make the best decisions. This means that it can make different allocation decisions in different situations, thereby improving the intelligence of the logistics management.
Example 5: based on the eigenvector matrix of the cement order data matrix and the eigenvector matrix of the cement warehouse data matrix, training a dual support vector machine classifier by using a dual support vector kernel function, and distributing the cement order to the corresponding warehouse comprises the following steps: using polynomial kernel functionsMatrix of eigenvectors of cement order data matrixEigenvector matrix for data matrix of cement warehouseMapping to a high-dimensional space; eigenvector matrix using cement order data matrixSum kernel functionTraining a dual support vector machine classifier comprising weight parameters of support vectors、Bias termIs calculated; decision function of classificationIs used to categorize orders ifThen the order is allocated to the corresponding warehouseOtherwise, not distributing; wherein,Eigenvector matrix being a matrix of cement order dataMiddle (f)The number of elements to be added to the composition,Index for subscript; Eigenvector matrix being a cement warehouse data matrix The first of (3)The number of elements to be added to the composition,Index for the subscript.
Specifically, first, a feature vector matrix is derived from a cement order data matrixEigenvector matrix for data matrix of cement warehouseExtracting feature vectors. These feature vectors are numerical representations that contain various attribute information about the order and warehouse, such as order requirements, warehouse attributes and locations, etc. In D-SVM, a kernel function is used to process the nonlinear properties of data. In this embodiment, a polynomial kernel function is used. The function of the kernel is to map the original features to a higher dimensional feature space for better classification. Polynomial kernel functions introduce more complex features in high-dimensional space that help solve the non-linearity problem. Using mapped feature vectorsSum kernel functionTo train a dual support vector machine classifier. The training process aims to find an optimal classification hyperplane to maximize the accuracy of classification. Training of the classifier includes computing weight parameters of the support vectorAndBias term. After the classifier training is completed, a classification decision function can be usedTo categorize new cement orders. This decision function is based on learned hyperplane and support vector parameters for deciding which warehouse to assign an order to. If it isThen the order is allocated to the corresponding warehouse; Otherwise, no allocation is made.
The main role of the D-SVM is to intelligently distribute cement orders to the appropriate warehouse. By using the polynomial kernel function and the dual support vector machine classifier, complex relations between orders and warehouses can be better processed, and the accuracy of order allocation is improved. The polynomial kernel functions map the original features into a high-dimensional space, which helps to address the non-linearity problem. In high-dimensional space, data is more easily separated, thereby improving the performance of the classifier. By training the dual support vector machine classifier, an optimal classification hyperplane can be learned to maximize classification accuracy. This enables the classifier to better accommodate different orders and warehouse situations. Once classifier training is complete, new orders can be classified and assigned automatically without manual intervention. This increases the efficiency and automation level of logistics management.
Example 6: decision function of classificationThe expression is used as follows:
；
wherein, Is the order weight parameter, which is the first in the order weight matrixThe order weight matrix is a preset matrix, and optimization adjustment is carried out through a maximized objective function; is the warehouse weight parameter, which is the first in the warehouse weight matrix The warehouse weight matrix is a preset matrix, and optimization adjustment is carried out through a maximized objective function; Is a bias term.
Decision function of classificationThe principle of (a) is based on the working principle of a dual support vector machine (D-SVM) classifier. The D-SVM is a machine learning algorithm for solving classification problems, wherein input data (here, feature vectors of cement orders and warehouses) need to be assigned to different classes (warehouses). The objective of the decision function is to determine the orderWhether or not it should be allocated to a warehouse。
Specifically, this decision function is based on several key elements: And : These parameters are weight parameters of the order and warehouse, the values of which influence the outcome of the classification. Which correspond to specific elements in the weight matrix of the order and warehouse, respectively, may be adjusted by optimizing the objective function.And: These are feature vectors of orders and warehouses, containing various attribute information about them.: The bias term, which is a constant, is used to adjust the intercept of the classification decision function.Is the result of the kernel calculation used to measure the similarity between the order and the warehouse.
Decision function of classificationThe function of (a) is to determine an orderWhether or not it should be allocated to a warehouse. The working mode is as follows: inner product term: This term calculates the inner product of the order feature vector and the warehouse feature vector, by weight parametersAndWeighting is performed. The inner product reflects the similarity or correlation between them, with a larger value indicating a more relevant. Kernel function term: The kernel function is used to measure the similarity or distance between the order and the warehouse. It takes into account the inner product and distance between feature vectors. The kernel function may help address non-linearity issues because it maps data to higher dimensional feature spaces.: Bias termAnd the method is used for adjusting the intercept of the classification decision function to influence the final classification result. The final decision is based on the combined result of these terms. If it isGreater than zero, orderIs allocated to warehouse; Otherwise, no allocation is made. The decision function comprehensively considers the influence of similarity (inner product item and kernel function item) and bias item between the order and the warehouse so as to achieve accurate classification.
Kernel functionThe expression is used as follows:
。
Example 7: using a convex optimization algorithm to make the objective function of the dual support vector machine classifier reach the maximum value, and expressing the objective function by the following formula:
。
Inner product term : This part calculates order feature vectorsAnd warehouse feature vectorIs the square of the inner product of (2). The inner product reflects the similarity or correlation between them, and the square term can enhance the effect of the similarity. The larger the inner product, the more similar the two feature vectors are. Index term: This section includes an index term that considers the order feature vectorAnd warehouse feature vectorThe square of the euclidean distance between them. The parameters of the exponential term includeAndThey are used to adjust the influence of distance items. By this term, the kernel function takes into account the distance information between the feature vectors, and if the feature vectors are close enough, the value of the exponential term will be large, indicating that they are similar. Comprehensively considering the information of the inner product and the distance, the principle of the kernel function is to comprehensively evaluate the order through the two partsAnd warehouseAnd the similarity between the two is used for subsequent classification decision.
Kernel functionThe function of (1) is to provide a way for a dual support vector machine (D-SVM) classifier to measure the similarity between orders and warehouses. It plays the following roles in classification tasks: the inner product term and the index term of the kernel function measure the similarity between the order and the warehouse, respectively. The inner product term measures the correlation between them and the index term considers the distance between them. Through the two terms, the kernel function can comprehensively evaluate the similarity between the two feature vectors, including the correlation and the distance. An important role of the kernel function is to map data to a feature space of higher dimension. This helps to address the non-linearity problem because in high dimensional space, the data is more easily separated. By introducing more complex features, the kernel functions enable the classifier to better fit and classify the data. The kernel function integrates a plurality of characteristic information, including inner product, distance and weight parameters. This allows the kernel function to more fully take into account the relationship between the order and the warehouse, not just the linear correlation. The calculated kernel function value is used for the classification decision function of the D-SVM classifierTo which warehouse an order is assigned. If the kernel function value is larger, the more similar the order and the warehouse are, the greater the likelihood that the order will be assigned to the warehouse.
Example 8: the cement transporter determining unit, based on the eigenvector matrix of the cement transporter data matrix and the eigenvector matrix of the cement warehouse data matrix, and the result of assigning the cement order to the corresponding warehouse, assigns the cement transporter to the corresponding warehouse, the method includes:
The allocation score is calculated using the following formula:
；
wherein, Representing the presentation to beAnd (3) withThe operation of matrix splicing is carried out, the matrix splicing operation is carried out,Is the Frobenius norm; is L1 norm; Is that The feature vector of the corresponding warehouse to which the corresponding cement order is allocated; if it isGreater than the set first threshold value, thenCorresponding cement transporter is bound toA corresponding warehouse.
Specifically, the feature vector is spliced: first, order feature vectors are usedAnd warehouse feature vectorAnd (5) performing matrix splicing. The purpose of this operation is to incorporate the order and warehouse attribute information into a feature vector to take into account the interplay between them. Assigning a molecular moiety of a scoreThe similarity between the order and the warehouse is measured by calculating the Frobenius norm of the spliced feature vector. The Frobenius norm measures the distance of two feature vectors in a high-dimensional feature space. Denominator portion of the distribution scoreFor normalization, it is the L1 norm of the eigenvector of the cement transporter. The purpose of this operation is to take into account the weight differences of the different data objects (orders, warehouse, transporter). Final distribution scoreSimilarity and weight factors are comprehensively considered. If the score is greater than the set threshold, the cement transporter is indicatedWith warehouseThe correlation is high, and they can be bound together.
Example 9: the binding model established by the cement provider binding optimization unit is expressed using the following formula:
；
wherein, For the calculated vendor binding score, ifGreater than a set second threshold value, thenThe corresponding vendor is bound toA corresponding warehouse; is a parameter of the binding model.
In particular, the method comprises the steps of,: This section represents the firstL1 norms of eigenvectors of the cement transporter. The L1 norm is the sum of the absolute values of all elements in the vector. This section is used to take into account the information on the characteristics of the cement transporter, the larger the L1 norm, the more important the characteristics representing the transporter.: This is a product term that involves the processing of all cement order feature vectors. The L1 norm of the eigenvector of each order is converted by an exponential function, then the logarithm of 1 is added, and finally the multiplication is performed. This operation introduces characteristic information of the order and is weighted by a nonlinear transformation of the exponential function.: This section involves the processing of all warehouse feature vectors. The L1 norms of the eigenvectors of each warehouse are processed by a sine function and then added. This operation introduces characteristic information of the warehouse and is weighted by a sine function.：Is a parameter for controlling the degree of influence of the model. It can adjust the contribution of different parts to the binding score to achieve flexibility and tuning capabilities of the model. In general, the principle of this formula is to calculate the binding score of the provider by comprehensively considering the characteristic information and weights of the different data objects (cement transporter, cement order, cement warehouse). The operation of the different parts introduces non-linear factors (such as exponential and sinusoidal functions) to more fully account for the effects of the features. Parameters (parameters)May be adjusted to control the behavior of the model. Behind this formula, a method of building a binding model using a fast random tree is mentioned. A fast random tree is a machine learning algorithm that is used to process high-dimensional data and complex models. It can help the model learn the relationships between the data and introduce randomness during modeling to reduce the risk of overfitting. This means that when the vendor binding model is built, a fast random tree is used to process the data and optimize the parameters to obtain accurate binding results.
Example 10: the binding model objective function is maximized to optimize the parameters of the binding model by the following formula:
；
wherein, Representation fetchIs a minimum value in (a). In particular, the method comprises the steps of,: This is a maximization problem, with the goal of finding the largest parameters. The function of this term is to maximize the parametersIn order to maximize the binding model objectives.: This coefficient is used to scale the latter term in order to better control the influence of the parameters.: This is a product term that contains three index variablesThey are respectively in the range ofTo the point ofAnd (5) taking an internal value. This term is a running-on of virtually all possible combinations, taking into account various combinations of cement orders, cement warehouses and cement carriers.: This is a factor in the continuous term that represents the Frobenius norm of the feature vector. The Frobenius norm is the square root of the sum of the squares of all elements of a matrix. The purpose of this factor is to measure the importance of the different combinations by the norms of the feature vectors.: This is the parameter to be optimized, by adjusting which the value of the whole objective function can be controlled.: This part introduces the feature information of the warehouse, weighted by computing the sine function values of the warehouse feature vectors.
The objective function is to find the optimal parameter configurationTo maximize the target value of the binding model. The specific actions are as follows: the objective of the overall objective function is to maximize the parametersThis means that one is to be foundSo that the value of the whole objective function is as large as possible. This helps to optimize the binding model to make it more accurate and efficient in binding cement suppliers and warehouses. Continuous multiplying termVarious combinations of cement orders, cement warehouses, and cement carriers are contemplated. This means that the model will integrate the effects of the different combinations to find the best binding configuration. By using the Frobenius norm to measure the importance of feature vectors, it can be ensured that the model is more focused on features with larger norms. This helps to better capture important information in the data. The warehouse features are weighted by a sine function, and the model can take into account the binding effect of different features of the warehouse. This makes the model more flexible and can be optimized according to the characteristics of different warehouses.
In addition, metering device data will be collected in real time during transport by the transporter. The system first needs to collect data of various metering devices (such as wagon balance, flow balance, etc.), including weight, flow, and other relevant parameters, in real time. The data acquired by the device are represented in time series, whereinRepresent the firstDevice data of the moment. The system needs to be connected to the terminal box through the Internet of things and controls the display screen of the terminal box to realize data synchronization. The terminal box updates the display content by receiving a system instruction, whereinRepresent the firstControl command of time. The system uses an infrared grating technique to determine when a vehicle enters and exits the wagon balance area. This can be expressed as a logic functionWhereinIndicating that the vehicle is within the wagon balance zone,Indicating that the vehicle is outside the wagon balance area. The system preprocesses the collected data, including denoising, smoothing, abnormality detection and the like. This can be expressed as a data preprocessing functionWhereinRepresenting a data preprocessing operation. The system processes the data of the infrared raster using an image processing algorithm and a pattern recognition algorithm to determine the status of the vehicle (entering or exiting the wagon balance area). This can be expressed as a pattern recognition function. And synchronizing the preprocessed data and the mode identification result to a terminal box display screen by the system so as to update the display content in real time. This can be expressed as a data synchronization function. The system encodes and feature extracts the data of each metering device to obtain a numerical attribute vector. This can be expressed as a feature extraction function. The system uses the encoded data to make intelligent logistics decisions, including assigning orders to corresponding warehouses and assigning shippers to corresponding warehouses. This can be expressed as a decision function. The system uses a fast random tree algorithm to build a binding model that binds suppliers to corresponding warehouses. This can be expressed as a binding model building function:
。
the system optimizes parameters of the binding model, taking into account the multi-layer tree structure and nonlinear factors to maximize the objective function of the binding model. This can be expressed as an optimization function:
。
the real-time collection of the metering equipment data can ensure that the system always has the latest weight, flow and other key parameter information, thereby realizing the real-time update of the logistics data. Through the internet of things technology, the system can be connected with various metering devices and acquire measurement data of the metering devices at regular intervals. These data are the basis for logistics management for tracking the amount and status of supplies.
By controlling the display screen of the terminal box and synchronizing the data in real time, the system can provide immediate logistics information for drivers and operators, thereby improving the decision-making efficiency. The logistics system is connected to the terminal box through the Internet of things and sends an instruction to the terminal box to update the display content. In this way, the key data such as the state of the vehicle, order information, etc. can be immediately communicated to the relevant person.
The infrared grating technology can accurately judge when a vehicle enters and leaves a wagon balance area, so that accuracy of metering data is ensured. The infrared grating system uses an infrared sensor to detect whether the vehicle passes through the grating. When a vehicle enters or leaves, the sensor will record a time stamp, thereby determining the position of the vehicle.
Data preprocessing helps to remove noise and outliers from the data, ensuring that the system is analyzing high quality data. Data preprocessing may include techniques such as smoothing, filtering, and anomaly detection to ensure that the data acquired from the metrology device is reliable. This helps to improve the accuracy of the subsequent steps.
Image processing and pattern recognition techniques can accurately determine the status of a vehicle, including entering and exiting a wagon balance area. The system uses a pattern recognition algorithm to detect the motion pattern of the vehicle by processing the raster data. This can be achieved by analyzing the time stamp of the sensor record and the vehicle position.
The data synchronization ensures that all relevant personnel can timely acquire the update of the logistics data, thereby supporting real-time decision making. The system synchronizes the preprocessed data and pattern recognition results to the terminal box display screen to ensure that the driver and operator can immediately learn the status and related information of the vehicle.
Data encoding and feature extraction convert non-numeric attribute vectors into numeric attribute vectors for subsequent intelligent decision making. By using the appropriate encoding method, the system converts various data into a numerical representation. The feature extraction algorithm then extracts key features from these numerical attributes for use in subsequent decision models.
Based on the encoded data, the system can intelligently allocate orders to corresponding warehouses and allocate carriers to corresponding warehouses to optimize logistics flows. The system uses decision algorithms to determine the optimal allocation scheme taking into account various factors such as the number of goods, warehouse capacity, mode of transportation, etc. This helps to improve logistic efficiency and reduce costs.
Establishing a binding model can bind suppliers to corresponding warehouses, and more effective logistics management is realized.
Principle of: the system uses a fast random tree algorithm to build a binding model, taking into account multi-layer tree structures and nonlinear factors. This model may help determine which suppliers should cooperate with which warehouses to meet the order requirements. By optimizing the parameters of the binding model, the system can further improve the efficiency and accuracy of the supply chain. The optimization process considers the objective function of the binding model and maximizes this function by adjusting the model parameters. This helps ensure that the supplier to warehouse match is optimal, thereby minimizing logistic costs and improving delivery efficiency.
The above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. An intelligent logistics system based on the internet of things, the system comprising: a logistics data construction unit, configured to acquire cement warehouse data to construct a cement warehouse data matrix, wherein each element in the matrix represents an attribute vector of a cement warehouse, acquire cement transporter data to construct a cement transporter data matrix, each element in the matrix represents an attribute vector of a cement transporter, acquire cement order data to construct a cement order data matrix, each element in the matrix represents an attribute vector of a cement order, and acquire cement provider data to construct a cement provider data matrix, and each element in the matrix represents an attribute vector of a cement provider; the feature extraction unit is used for respectively encoding the cement order data matrix, the cement warehouse data matrix, the cement transporter data matrix and the cement provider data matrix by using the read-heat codes so as to convert the non-numerical attribute vector into a numerical attribute vector, then respectively carrying out feature extraction on each attribute vector in each data matrix to obtain a feature vector of each attribute vector, and finally obtaining a feature vector matrix of the cement order data matrix, a feature vector matrix of the cement warehouse data matrix, a feature vector matrix of the cement transporter data matrix and a feature vector matrix of the cement provider data matrix; a cement order classification unit for distributing cement orders to corresponding warehouses based on the eigenvector matrix of the cement order data matrix and the eigenvector matrix of the cement warehouse data matrix; a cement transporter determining unit for assigning cement transporters to the corresponding warehouses based on the eigenvector matrix of the cement transporter data matrix and the eigenvector matrix of the cement warehouse data matrix, and the result of assigning cement orders to the corresponding warehouses; and the cement provider binding optimization unit is used for building a binding model by using a fast random tree based on the eigenvector matrix of the cement provider data matrix, the result of distributing the cement order to the corresponding warehouse by cement and the result of distributing the cement transporter to the corresponding warehouse, binding the cement provider to the corresponding warehouse, and maximizing a binding model objective function by considering a multi-layer tree structure and nonlinear factors in the process of building the binding model so as to optimize parameters of the binding model.
2. The internet of things-based intelligent logistics system of claim 1, wherein the elements of the attribute vector of the cement order comprise at least: cement demand, order user location data, and order urgency; the attribute vector of the cement transporter at least comprises: upper limit of cement transportation amount, position data of transportation and transportation price; the cement provider's attribute vector includes at least: supply cement price, supplier location data and supplier supply; the attribute vector of the cement warehouse comprises: maximum capacity of the cement silo, remaining capacity of the cement silo, and cement silo position data.
3. The intelligent logistics system of claim 2, wherein the cement order data matrix is set as:
，
Wherein the method comprises the steps of Represents the/>Attribute vectors for individual cement orders; setting a cement warehouse data matrix as follows:
，
Wherein the method comprises the steps of Represents the/>Attribute vectors of the cement warehouses; let the cement transporter data matrix be:
，
Wherein the method comprises the steps of Represents the/>Attribute vectors of individual cement transporters; let the cement provider data matrix be:
，
Wherein the method comprises the steps of Represents the/>Attribute vectors for individual cement suppliers; let the eigenvector matrix of the cement order data matrix be/>; Let the eigenvector matrix of the cement warehouse data matrix be/>; Let the eigenvector matrix of the cement transporter data matrix be; Let the eigenvector matrix of the cement provider data matrix be/>; Wherein/>For cement order quantity,/>For the number of cement warehouses,/>For the number of cement transporters,/>For the number of cement suppliers, the following constraint is satisfied:
；
；
。
4. The internet of things-based intelligent logistics system of claim 3, wherein the cement order classification unit trains a dual support vector machine classifier using a dual support vector kernel function based on a feature vector matrix of the cement order data matrix and a feature vector matrix of the cement warehouse data matrix, distributes the cement order to the corresponding warehouse, and uses a convex optimization algorithm when training the dual support vector machine classifier so that an objective function of the dual support vector machine classifier reaches a maximum value to optimize parameters of the dual support vector machine classifier.
5. The internet of things-based intelligent logistics system of claim 4, wherein the method for assigning the cement order to the corresponding warehouse using a dual support vector machine classifier trained using a dual support vector kernel function based on the eigenvector matrix of the cement order data matrix and the eigenvector matrix of the cement warehouse data matrix comprises: using polynomial kernel functionsEigenvector matrix/>, of cement order data matrixAnd eigenvector matrix/>, of a cement warehouse data matrixMapping to a high-dimensional space; eigenvector matrix/>, using cement order data matrixSum kernel functionTraining a dual support vector machine classifier comprising the weight parameters/>, of the support vector、/>Bias term/>Is calculated; decision function of classification/>Is used to categorize orders ifThen the order is assigned to the corresponding warehouse/>Otherwise, not distributing; wherein/>Eigenvector matrix/>, which is a matrix of cement order dataMiddle/>Element,/>Index for subscript; /(I)Eigenvector matrix/>, which is a cement warehouse data matrix/>Element,/>Index for the subscript.
6. The internet of things-based intelligent logistics system of claim 5 wherein the decision function of the classificationThe expression is used as follows:
；
wherein, Is the order weight parameter, which is the/>, of the order weight matrixThe order weight matrix is a preset matrix, and optimization adjustment is carried out through a maximized objective function; /(I)Is a warehouse weight parameter, which is the/>, in the warehouse weight matrixThe warehouse weight matrix is a preset matrix, and optimization adjustment is carried out through a maximized objective function; /(I)Is a bias term;
Kernel function The expression is used as follows:
。
7. The internet of things-based intelligent logistics system of claim 6, wherein the convex optimization algorithm is used to maximize the objective function of the dual support vector machine classifier, which is expressed by the following formula:
。
8. the internet of things-based intelligent logistics system of claim 7, wherein the method of assigning the cement transporter to the corresponding warehouse based on the eigenvector matrix of the cement transporter data matrix and the eigenvector matrix of the cement warehouse data matrix, and the result of assigning the cement order to the corresponding warehouse comprises:
The allocation score is calculated using the following formula:
；
wherein, Representation will/>And/>Matrix stitching operation is carried out,/>Is the Frobenius norm; /(I)Is L1 norm; /(I)For/>The feature vector of the corresponding warehouse to which the corresponding cement order is allocated; if it isGreater than a set first threshold, will/>Corresponding cement transporter is bound to/>A corresponding warehouse.
9. The intelligent logistics system of claim 8, wherein the binding model established by the cement provider binding optimization unit is expressed using the following formula:
；
wherein, For the calculated vendor binding score, if/>Greater than a set second threshold, will/>Corresponding vendor is bound to/>A corresponding warehouse; /(I)Is a parameter of the binding model.
10. The internet of things-based intelligent logistics system of claim 9 wherein the binding model objective function is maximized to optimize parameters of the binding model by:
；
wherein, Representation of the fetch/>Is a minimum value in (a).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410138126.9A CN117933882B (en) | 2024-02-01 | 2024-02-01 | Intelligent logistics system based on Internet of things |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410138126.9A CN117933882B (en) | 2024-02-01 | 2024-02-01 | Intelligent logistics system based on Internet of things |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117933882A true CN117933882A (en) | 2024-04-26 |
| CN117933882B CN117933882B (en) | 2024-10-01 |
Family
ID=90770087
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410138126.9A Active CN117933882B (en) | 2024-02-01 | 2024-02-01 | Intelligent logistics system based on Internet of things |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117933882B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118709859A (en) * | 2024-08-27 | 2024-09-27 | 华侨大学 | Supply chain optimization method, device and readable medium based on IJS-SVM model |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5450317A (en) * | 1993-11-24 | 1995-09-12 | U S West Advanced Technologies, Inc. | Method and system for optimized logistics planning |
| CN110516985A (en) * | 2018-05-21 | 2019-11-29 | 北京京东尚科信息技术有限公司 | Warehouse selection method, system, computer system and computer readable storage medium storing program for executing |
| US11205152B1 (en) * | 2016-03-18 | 2021-12-21 | Amazon Technologies, Inc. | Virtual logistical network overlay for physical logistical network |
| CN114169602A (en) * | 2021-12-01 | 2022-03-11 | 深圳市丰巢网络技术有限公司 | Combination optimization method, device, computer equipment and storage medium for asset delivery |
| WO2022262469A1 (en) * | 2021-06-18 | 2022-12-22 | 重庆邮电大学工业互联网研究院 | Industrial park logistics scheduling method and system based on game theory |
| CN116629577A (en) * | 2023-06-20 | 2023-08-22 | 深圳市携客互联科技有限公司 | Intelligent supply chain management system based on big data |
| CN117114550A (en) * | 2023-10-17 | 2023-11-24 | 米烁网络科技(广州)有限公司 | Commodity supply chain intelligent supervision system based on internet |
| CN117172665A (en) * | 2023-09-01 | 2023-12-05 | 深圳市富润德供应链管理有限公司 | Enterprise logistics supply chain management system |
| CN117408596A (en) * | 2023-11-01 | 2024-01-16 | 广东享通速递有限公司 | Intelligent logistics distribution management system and method thereof |
-
2024
- 2024-02-01 CN CN202410138126.9A patent/CN117933882B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5450317A (en) * | 1993-11-24 | 1995-09-12 | U S West Advanced Technologies, Inc. | Method and system for optimized logistics planning |
| US11205152B1 (en) * | 2016-03-18 | 2021-12-21 | Amazon Technologies, Inc. | Virtual logistical network overlay for physical logistical network |
| CN110516985A (en) * | 2018-05-21 | 2019-11-29 | 北京京东尚科信息技术有限公司 | Warehouse selection method, system, computer system and computer readable storage medium storing program for executing |
| WO2022262469A1 (en) * | 2021-06-18 | 2022-12-22 | 重庆邮电大学工业互联网研究院 | Industrial park logistics scheduling method and system based on game theory |
| CN114169602A (en) * | 2021-12-01 | 2022-03-11 | 深圳市丰巢网络技术有限公司 | Combination optimization method, device, computer equipment and storage medium for asset delivery |
| CN116629577A (en) * | 2023-06-20 | 2023-08-22 | 深圳市携客互联科技有限公司 | Intelligent supply chain management system based on big data |
| CN117172665A (en) * | 2023-09-01 | 2023-12-05 | 深圳市富润德供应链管理有限公司 | Enterprise logistics supply chain management system |
| CN117114550A (en) * | 2023-10-17 | 2023-11-24 | 米烁网络科技(广州)有限公司 | Commodity supply chain intelligent supervision system based on internet |
| CN117408596A (en) * | 2023-11-01 | 2024-01-16 | 广东享通速递有限公司 | Intelligent logistics distribution management system and method thereof |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118709859A (en) * | 2024-08-27 | 2024-09-27 | 华侨大学 | Supply chain optimization method, device and readable medium based on IJS-SVM model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117933882B (en) | 2024-10-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Efat et al. | Deep-learning model using hybrid adaptive trend estimated series for modelling and forecasting sales | |
| Zowid et al. | Multi-criteria inventory ABC classification using Gaussian Mixture Model | |
| CN117853072B (en) | Supply chain management system and method | |
| CN117933882B (en) | Intelligent logistics system based on Internet of things | |
| CN117852980B (en) | Intelligent logistics service management system and method based on MoE model fusion technology | |
| Hasan et al. | Model development for predicting the crude oil price: Comparative evaluation of ensemble and machine learning methods | |
| CN119228273B (en) | Consignment system management system and method thereof | |
| Naik et al. | Machine learning based food sales prediction using random forest regression | |
| Kamal et al. | Machine Learning Fusion and Data Analytics Models for Demand Forecasting in the Automotive Industry: A Comparative Study. | |
| Pai et al. | Forecasting output of integrated circuit industry by support vector regression models with marriage honey-bees optimization algorithms | |
| CN119338384A (en) | A method and device for intelligent control of storage and picking optimization groups based on neural network algorithm | |
| Gunjal et al. | Fusing clustering and machine learning techniques for Big-Mart sales predication | |
| CN117592904A (en) | Methods for identification and quantity analysis of inbound and outbound materials | |
| Mouthami et al. | Analysis of SARIMA-BiLSTM-BiGRU in Furniture Time Series Forecasting | |
| Abushaega et al. | Multi-objective sustainability optimization in modern supply chain networks: A hybrid approach with federated learning and graph neural networks | |
| Baral et al. | Fuzzy TOPSIS approaches for multi-criteria decision-making problems in triangular fuzzy numbers | |
| Pandey et al. | Fuzzy based investment portfolio management | |
| Harale et al. | Supplier prediction in fashion industry using data mining technology | |
| Oztemel et al. | A Constrained Clustering and Deep Learning Integrated Solution with Density Based Anomaly Detectors for Artisan-Courier Model in E-Commerce Logistics | |
| US20250165918A1 (en) | Product distribution system and method | |
| Sogukkuyu et al. | A Supervised Learning Algorithms for Consumer Product Returns Case Study for FLO Offline Stores | |
| Pavithra et al. | Optimizing Inventory Management using Demand Forecasting based Ensemble Stacking Model with Hyperparameter Tuning | |
| Wang et al. | Platform Merchant Demand Prediction Based on Decision Tree and Multi-Layer Perceptron Models | |
| Güler et al. | Forecasting Damaged Containers with Machine Learning Methods | |
| Steenbergen | New Product Forecasting with Analogous Products: Applying Random Forest and Quantile Regression Forest to forecasting and inventory management |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |