CN118673316B - Productive financial service data synthesis method and system based on large model reasoning - Google Patents
Productive financial service data synthesis method and system based on large model reasoning Download PDFInfo
- Publication number
- CN118673316B CN118673316B CN202410686464.6A CN202410686464A CN118673316B CN 118673316 B CN118673316 B CN 118673316B CN 202410686464 A CN202410686464 A CN 202410686464A CN 118673316 B CN118673316 B CN 118673316B
- Authority
- CN
- China
- Prior art keywords
- product
- data
- model
- synthesis
- data set
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/29—Graphical models, e.g. Bayesian networks
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Computing Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
本发明涉及数据处理技术领域，具体公开基于大模型推理的生产性金融服务数据合成方法及系统，该方法包括：数据集收集、变换模型提取、产品质量可合成集匹配、变换模型的合成效益值分析以及数据合成服务优化，首先获取各产品在生产过程中的不同状态数据集，提取出变换模型，为后续的合成操作提供模型支撑；评估各产品的质量达标指数，匹配出各产品的质量达标可合成集，基于变换模型进行合成操作，能够更快速地处理和合成数据，提高整体质量达标可合成集的合成速度，分析变换模型的合成效益值，由此实现生产性产品质量数据的合成过程，以此提供快速、准确的数据合成服务以满足客户的需求。
The present invention relates to the field of data processing technology, and specifically discloses a method and system for synthesizing productive financial service data based on large model reasoning. The method comprises: data set collection, transformation model extraction, product quality synthesizable set matching, transformation model synthesis benefit value analysis and data synthesis service optimization. First, different state data sets of each product in the production process are obtained, and the transformation model is extracted to provide model support for subsequent synthesis operations; the quality compliance index of each product is evaluated, and the quality compliance synthesizable set of each product is matched. Synthesis operations are performed based on the transformation model, which can process and synthesize data more quickly, improve the synthesis speed of the overall quality compliance synthesizable set, analyze the synthesis benefit value of the transformation model, thereby realizing the synthesis process of productive product quality data, thereby providing fast and accurate data synthesis services to meet customer needs.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a method and a system for synthesizing productive financial service data based on large model reasoning.
Background
Currently, in order to promote productivity increase and economic development, a large-scale data synthesis operation is required when detecting the quality of a product, wherein a data synthesis service of a data set is one of important parts, a production institution needs to evaluate the quality of the product by fully utilizing data set information, formulate a quality detection policy and optimize quality risk management, and synthesize the adaptation degree of a user and each corresponding product by comparing the quality states of the product so as to provide a better quality, efficient and safe financial service, wherein complex algorithms and a large amount of computing resources may be required, the traditional data synthesis technology cannot synthesize quality data efficiently, and meanwhile, the data synthesis technology depends on a specific generation model, so that the quality and diversity of the synthesized data are limited.
For example, the invention patent with publication number CN108376069B discloses a data synthesis method and device thereof, which are operated according to the following steps of solidifying audio data into a first firmware and recompiling the first firmware, debugging electroacoustic parameters and converting the electroacoustic parameters into a data file containing the electroacoustic parameters, synthesizing the data file and the first firmware and generating a second firmware, synthesizing the first firmware generated by a compiling tool and the data file generated by a converting tool through a data synthesis tool and generating the second firmware, and saving manpower resources, reducing the trouble of recompiling the firmware, improving the working efficiency and improving the accuracy.
For example, the invention patent with publication number of CN111625523B discloses a method, a device and equipment for synthesizing data, which comprises obtaining original data, dividing the original data into key value data and residual data, generating new key value according to the key value data, processing the residual data to obtain discrete data and continuous data, carrying out chi-square distribution processing on the discrete data to obtain new discrete data, carrying out factor processing on the continuous data to obtain character type data and combined digital data, generating new continuous character data according to the character data, generating new continuous digital data according to the digital data Copula process and the operation formula data, processing the digital data obtained in the Copula process into time type data, and carrying out data synthesis on the new key value, the new discrete data, the new continuous digital data and the time type data.
However, in the process of realizing the embodiment of the application, the technology at least has the technical problems that in the process of synthesizing data, the dimension of the data to be synthesized is single, the work efficiency and the accuracy of the data synthesis cannot be improved, the reliability of the synthesis service cannot be improved, and the synthesis operation is directly finished after the data synthesis, but the accuracy of the synthesized data is still to be inspected, so that the synthesized result has a certain deviation, and the accuracy of the synthesis service cannot be improved.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a method and a system for synthesizing productive financial service data based on large model reasoning, which can effectively solve the problems related to the background art.
The invention provides a productive financial service data synthesis method based on large model reasoning, which comprises the steps of collecting operation data of production equipment in a production process of each product, supply chain information of a production line and weight and volume of each synthesized product, recording the operation data as a first data set, a second data set and a third data set of each product respectively, extracting a transformation model, constructing the first data set, the second data set and the third data set of each product as a first data set model, a second data set model and a third data set model respectively, obtaining an output result of each data set model by acquiring historical data of each product as an input of each data set model, screening out a data set model with highest data synthesis adaptation degree, matching a product quality synthesizable set, namely obtaining data set information of each product, evaluating a quality standard reaching index of each product, matching a quality standard reaching synthesizable set, analyzing a synthesis benefit value of the transformation model, optimizing the synthesis benefit value according to the synthesis benefit value, optimizing the synthesis benefit of the synthesis benefit threshold value, and optimizing the synthesis benefit of the synthesis model according to the synthesis benefit optimizing the synthesis benefit value.
The data synthesis process for realizing the quality of the productive product comprises the steps of comparing the synthesis benefit value of the transformation model with a set synthesis benefit threshold value, if the synthesis benefit value of the transformation model is lower than the synthesis benefit threshold value, carrying out difference processing on the synthesis benefit value of the transformation model and the synthesis benefit threshold value to obtain a synthesis benefit deviation value of the transformation model, and matching with a data synthesis optimization plan corresponding to each synthesis benefit deviation value interval to obtain a data synthesis optimization plan of the transformation model, thereby realizing the data synthesis process of the quality of the productive product.
The method comprises the steps of obtaining the synthesis information of the transformation model by carrying out synthesis operation on the quality standard synthesizable set of each product, specifically comprising the synthesis completion degree and the synthesis matching degree of the transformation model, and comprehensively analyzing to obtain the synthesis benefit value of the transformation model.
The quality standard-reaching synthesizable set of each product is input into a transformation model, the quality standard-reaching synthesizable set of each product is synthesized through a decoding output function of the transformation model, the synthesized data is output as synthesized data of the transformation model, and the synthesized data of the transformation model is screened according to synthesis specification indexes by using a Bayesian optimization method, so that the synthesis operation of the quality standard-reaching synthesizable set of each product is realized.
The quality standard-reaching synthesizable sets of the products are matched according to the quality standard-reaching indexes of the products, difference processing is carried out on the quality standard-reaching indexes and preset quality standard-reaching reference values to obtain quality standard-reaching deviation values of the products, and the quality standard-reaching synthesizable sets corresponding to the quality standard-reaching deviation value intervals defined in a product quality management library are matched to obtain the quality standard-reaching synthesizable sets of the products.
The method comprises the steps of constructing a first data set, a second data set and a third data set of each product into a first data set model, a second data set model and a third data set model respectively, obtaining the historical data of each product as the input of each data set model, obtaining the output result of each data set model, screening the data set model with the highest data synthesis adaptation degree, and marking the data set model as a transformation model.
The second aspect of the invention provides a system for synthesizing productive financial service data based on large model reasoning, which comprises a data set collecting module, a conversion model extracting module, a synthesis benefit value analyzing module and a synthesis benefit value optimizing module, wherein the data set collecting module is used for collecting operation data of production equipment of each product in a production process, supply chain information of a production line and weight and volume of each synthesized product, the first data set, the second data set and the third data set are respectively recorded as a first data set, a second data set and a third data set of each product, the conversion model extracting module is used for respectively constructing the first data set model, the second data set model and the third data set model of each product, the historical data of each product is used as input of each data set model, output results of each data set model can be obtained, the data set model with highest data synthesis adaptation degree is selected and recorded as a conversion model, the product quality synthesizable set matching module is used for obtaining the data set information of each product, the quality standard reaching index of each product is estimated, the quality standard reaching the synthesizable set is obtained, the conversion model is used for analyzing the conversion benefit value of each product based on the conversion model, the conversion model is used for optimizing the synthesis benefit value, the synthesis benefit is better than the synthesis benefit value is achieved by the synthesis benefit optimizing the synthesis benefit value, and the synthesis benefit value is lower than the synthesis benefit optimizing the synthesis benefit value is achieved by the synthesis benefit value.
Compared with the prior art, the embodiment of the invention has at least the following advantages or beneficial effects:
(1) The invention provides a method and a system for synthesizing productive financial service data based on large model reasoning, which comprises the steps of firstly obtaining different state data sets of each product in the production process, extracting a transformation model, providing model support for subsequent synthesis operation, evaluating quality standard reaching indexes of each product, matching the quality standard reaching synthesizable set of each product, carrying out synthesis operation based on the transformation model, processing and synthesizing data more quickly, reducing synthesis waiting time, improving the synthesis speed of the whole quality standard reaching synthesizable set, analyzing the synthesis benefit value of the transformation model, realizing the data synthesis process of the quality of the productive product, and providing rapid and accurate data synthesis service to meet the market demand through optimizing the synthesis benefit of the transformation model.
(2) According to the invention, the quality standard reaching index of each product is evaluated by acquiring the data set information of each product, so that a production mechanism can more accurately predict the quality risk of the product, and a proper risk management strategy is formulated for the risk, so that the quality standard reaching synthesizable set of each product is matched, an objective and quantifiable product quality standard reaching judgment basis can be provided, and the production mechanism can be helped to make a more accurate quality decision when examining and approving the product.
(3) The invention is based on the transformation model, synthesizes the quality standard synthesizable set of each product, analyzes the synthesis benefit value of the transformation model, is beneficial to the production mechanism to improve the reliability of the data synthesis service of each product, and simultaneously enhances the management intensity of the production mechanism on the quality risk of the product and the synthesis efficiency of the quality standard synthesizable set.
Drawings
The invention will be further described with reference to the accompanying drawings, in which embodiments do not constitute any limitation of the invention, and other drawings can be obtained by one of ordinary skill in the art without inventive effort from the following drawings.
FIG. 1 is a flow chart of the method steps of the present invention.
Fig. 2 is a schematic diagram of system module connection according to the present invention.
Fig. 3 is a graph showing a moving speed according to the present invention.
And 1, a speed threshold straight line is shown in the figure 3.
Detailed Description
The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are only some embodiments of the present invention, but not all embodiments, and all other embodiments obtained by those skilled in the art without making creative efforts based on the embodiments of the present invention are included in the protection scope of the present invention.
Referring to FIG. 1, a first aspect of the present invention provides a method for synthesizing productive financial service data based on large model reasoning, comprising collecting data sets, namely, collecting operation data of production equipment of each product in a production process, supply chain information of a production line, and weight and volume of each synthesized product, which are respectively recorded as a first data set, a second data set and a third data set of each product.
Specifically, the first data set, the second data set and the third data set of each product are specifically analyzed as follows:
According to the operation data of production equipment in the production process, the operation average temperature and the vibration frequency average value of the production equipment in the production period are extracted, wherein the operation average temperature and the vibration frequency average value of the equipment can be obtained through a temperature sensor and a fluctuation sensor carried by the equipment, the operation average temperature and the vibration frequency average value are recorded as a first data set of each product, the displacement and the moving speed of each product are extracted according to the supply chain information of the production line, the displacement and the moving speed of each product can be obtained through a product positioning system in the supply chain, the second data set of each product is recorded, and meanwhile, the weight and the volume of each synthesized product can be extracted in a product production report according to the weight and the volume of each synthesized product, and the third data set of each product is recorded.
And extracting a transformation model, namely respectively constructing a first data set, a second data set and a third data set of each product into a first data set model, a second data set model and a third data set model, obtaining the output result of each data set model by acquiring the historical data of each product as the input of each data set model, and screening the data set model with the highest data synthesis adaptation degree to be recorded as the transformation model.
Further, the transformation model specifically comprises the following analysis processes:
The method comprises the steps of dividing a first data set, a second data set and a third data set of each product into a corresponding training set, a verification set and a test set respectively, constructing the first data set model, the second data set model and the third data set model, selecting a proper model type according to the characteristics of the data sets of each product, using the training set training model, using the verification set to adjust model parameters, using the test set to evaluate the accuracy and generalization capability of the model, determining the data set model, taking historical data of each product as input of the data set model, obtaining an output index set of each data set model, storing the output index set of each data set model as an output index set of each data set model, screening the data set model with the highest data synthesis adaptation degree as a transformation model, wherein the determination of the data synthesis adaptation degree is determined by an expert in the financial field according to financial data and self experience comprehensive analysis, specifically, setting a section to which the output index set of each data set model belongs, storing the output index set of each data set model into a section, distributing the data set corresponding to the data set model, and ranking the data set according to the ranking order, and ranking the data set of the data set model, and matching the data set model according to the ranking the corresponding to the data set adaptation degree.
And (3) matching the product quality synthesizable sets, namely acquiring data set information of each product, evaluating the quality standard reaching index of each product, and matching the quality standard reaching synthesizable sets of each product.
Specifically, the production equipment abnormal operation influence degree coefficient of each product is specifically analyzed as follows:
And extracting and obtaining production equipment operation data of each product and production line supply chain information of each product according to the data set information of each product.
And according to production equipment operation data of each product, wherein the production equipment operation data comprise the average temperature of the production equipment in the production period and the average value of the vibration frequency.
And performing difference processing on the vibration frequency mean value of the production equipment in the production period and the vibration reference frequency to obtain a vibration deviation value of the production equipment.
And extracting the operation reference temperature and the vibration reference frequency from the product quality management library.
The data stored in the product quality management library are obtained by fitting after detecting quality information of the product for a plurality of times, for example, the operation reference temperature is obtained by acquiring operation conditions of the production equipment at different temperatures for a plurality of times, the detection personnel determine the optimal operation temperature according with the production equipment, and average processing is carried out on all the temperatures according with the conditions to obtain the operation reference temperature.
The data defined in the product quality management library may be changed according to the conditions of the production equipment, quality status, production environment, etc. of the product, and the embodiment is not limited in particular.
The production equipment vibration deviation value of each product is multiplied by the correction factor corresponding to the production equipment vibration deviation value to obtain the production equipment vibration influence index of each product.
The correction factors corresponding to the temperature deviation values of the production equipment and the correction factors corresponding to the vibration deviation values of the production equipment are used for reducing negative effects caused by systematic errors of the temperature of the production equipment and the vibration of the production equipment of each product in the process of calculating the abnormal operation influence degree coefficients of the production equipment of each product, and the correction factors corresponding to the temperature deviation values of the production equipment are in the range of 0.56-0.66, and the correction factors corresponding to the vibration deviation values of the production equipment are in the range of 0.59-0.65.
And adding the production equipment temperature influence index of each product with the production equipment vibration influence index of each product to obtain a production equipment abnormal operation influence degree coefficient beta i of each product.
Further, the production line supply and demand influence degree coefficient of each product comprises the following specific analysis processes:
and according to the production line supply chain information of each product, wherein the production line supply chain information comprises the displacement of each product moving in the supply period.
And carrying out ratio processing on the displacement of each product moving in the supply period and the time length corresponding to the supply period to obtain the moving speed of each product in the supply period, wherein the determination of the supply period is obtained by comprehensively analyzing the combination of the production time, the equipment utilization rate and the worker efficiency by a supplier.
According to the moving speed of each product in the supply period, a moving speed curve of the product is constructed, and the number of products with the moving speed greater than a set speed threshold value is extracted from the moving speed curve and is recorded as the number of products moving at high speed.
The above speed threshold is determined by a supplier through comprehensive analysis in combination with the supply time, the equipment productivity, and the worker efficiency.
From the moving speeds of the respective products in the supply period, a moving speed profile of the products is constructed, and as shown in fig. 3, the abscissa of the moving speed profile of the products is the product, the ordinate is the moving speed, and the unit is cm/min.
And according to the moving speed curve of the product, locating a speed threshold straight line 1, obtaining the number of products with the moving speed higher than the speed threshold straight line 1, and recording the number of products moving at a high speed.
And extracting influence factors corresponding to the products which define the displacement and move at a single high speed from the product quality management library.
The production line supply and demand influence degree coefficients of each product are comprehensively analyzed, and in one specific embodiment, various information of a production line supply chain can be collected, including but not limited to supply chain logistics information, production inventory information and product sales information, various product quality risk assessment models can be established based on the collected data, and the production line supply and demand influence degree coefficients of each product can be obtained through the models.
In this embodiment, the coefficient of the degree of influence of supply and demand of the production line of each product is obtained by comprehensively analyzing the displacement of each product moving in the supply period and the number of products moving at high speed, so as to determine the numerical value of the degree of influence of supply and demand of the production line of each product, and in this embodiment, a more accurate calculation method is adopted to obtain the numerical value, and the specific expression is as follows:
In this embodiment, if the product moves substantially in the supply period, the displacement of the product will increase, and the moving speed of the product in unit time will also increase substantially, so that the present expression is simplified, and the quality risk status of different products can be more accurately mastered.
I is denoted as the number of each product, i=1, 2,3,..m, m is denoted as the total number of products.
FZ i is the displacement of the ith product during the supply cycle, and refers to the distance the product travels from the supply chain to the production line during the supply cycle.
FZ' is denoted as defining the displacement and refers to the maximum value of the displacement.
GS is expressed as the number of products moving at a high speed, and refers to the number of products moving at a speed higher than a speed threshold.
In this embodiment, the number of products moving at high speed is set to 13, and the corresponding influence factor of a single product moving at high speed is set to 0.53, so that the expression is givenThe partial operation result was 2.53.
A 1 is a predefined correction factor corresponding to displacement, a 2 is an influence factor corresponding to a single high-speed moving product, wherein the value of the correction factor corresponding to the displacement and the influence factor corresponding to the single high-speed moving product are obtained by inviting experts in the related field, the experts finally determine the correction factor and the value of the influence factor according to the displacement of the product, the relation between the high-speed moving product and the product quality risk and own experience, the value range of the correction factor corresponding to the displacement is 0.38-0.52 in the embodiment, the value range of the influence factor corresponding to the single high-speed moving product is 0.47-0.59, and e is a natural constant.
Specifically, in a specific embodiment, the quality standard indexes of the products can be obtained by collecting detailed information of clients, performing quality standard risk analysis, constructing a quality standard assessment model according to the result of the quality standard risk analysis, inputting quality data of the products into the assessment model, and outputting a quality standard index by the model.
The quality standard index of each product is obtained by comprehensively analyzing the abnormal operation influence degree coefficient of the production equipment of each product and the supply and demand influence degree coefficient of the production line of each product, so as to judge the value of the quality standard index of each product, and in the embodiment, the quality standard index is obtained by adopting a more accurate calculation method, and the specific analysis process is as follows:
In this embodiment, if the degree of influence coefficient of abnormal operation of production equipment of a product is higher, it is indicated that an abnormal condition occurs in the equipment of the product, so that the risk of quality standard reaching of the product is increased, and then the degree of influence coefficient of supply and demand of a production line of the product is also increased.
I is denoted as the number of each product, i=1, 2,3,..m, m is denoted as the total number of products.
Beta i is expressed as a production equipment abnormal operation influence degree coefficient of the ith product, and is a numerical value of the production equipment abnormal operation influence degree of each product obtained through comprehensive analysis of the production equipment operation temperature and the vibration frequency of each product.
Mu i is expressed as a line supply and demand influence degree coefficient of the ith product, and is expressed as a value of line supply and demand influence degree of each product obtained by comprehensively analyzing the displacement of each product moving in the supply period and the number of products moving at high speed.
In this embodiment, the production line supply and demand influence degree coefficient of the ith product is set to 33%, and the value range of the weight factor corresponding to the production line supply and demand influence degree coefficient is set to 0.7, so that the operation result of the μ i*b2 part in the expression is 0.231.
B 1 is a weight factor corresponding to a predefined production equipment abnormal operation influence degree coefficient, b 2 is a weight factor corresponding to a predefined production line supply and demand influence degree coefficient, wherein the weight factor corresponding to the production equipment abnormal operation influence degree coefficient and the weight factor corresponding to the production line supply and demand influence degree coefficient are data of past product production cases through analysis, including production equipment information and supply chain information, the actual influence of different production equipment information and supply chain information on quality standard reaching of each product is known, the influence degree of the different production equipment information and the supply chain information is calculated through regression analysis, the regression analysis is a statistical method, in the embodiment, a linear regression analysis method is used, and is mainly used for researching the relation between the independent variable different production equipment information and the supply chain information and the quality standard reaching index of the dependent variable, so that the weight factor corresponding to the production equipment abnormal operation influence degree coefficient and the weight factor corresponding to the production line demand influence degree coefficient are obtained, the weight factor corresponding to the production equipment abnormal operation influence degree coefficient is obtained, the weight factor corresponding to the production line quality standard reaching the weight factor is 0.58 to 0.66, and the weight factor corresponding to the production line supply and demand influence degree coefficient is 0.66.
Further, the quality of each product is matched to reach the standard of the synthesizable set, and the specific matching process is as follows:
and carrying out difference processing according to the quality standard reaching indexes of the products and a preset quality standard reaching reference value, wherein the quality standard reaching reference value is obtained by comprehensively evaluating the quality standard reaching application of the products by a production mechanism according to the product quality risk management policy and an internal evaluation model of the production mechanism, and then giving the quality standard reaching reference value to obtain the quality standard reaching deviation value of each product, and matching the quality standard reaching synthesizable set corresponding to each quality standard reaching deviation value interval defined in a product quality management library, wherein the matching process is to set each quality standard reaching deviation value interval, determine the interval to which the quality standard reaching deviation value of each product belongs, and allocate the quality standard reaching synthesizable set corresponding to the interval to a client corresponding to the quality standard reaching deviation value, so as to obtain the quality standard reaching synthesizable set of each product.
And analyzing the synthetic benefit value of the transformation model, namely carrying out synthesis operation on the quality standard synthesizable set of each product based on the transformation model, and analyzing the synthetic benefit value of the transformation model.
The quality standard synthesizable set of each product is input into a transformation model, wherein the transformation model has advantages in terms of processing sequence data and complex relationships, i.e. the transformation model processes a variable length sequence of the quality standard synthesizable set by means of a self-attention mechanism, divides the quality standard synthesizable set into representation subspaces by means of a multi-headed self-attention mechanism, and performs self-attention calculations on a plurality of different representation subspaces in parallel.
The quality standard-reaching synthesizable sets of the products are synthesized through the decoding output function of the transformation model, wherein the decoding output function is realized according to a decoder in the transformation model, the main function of the decoder is to receive synthesizable set vectors processed by an encoder in the transformation model as input, namely, output the synthesizable set vectors as quality standard-reaching synthesizable data of the products, and the quality standard-reaching synthesizable data of the products are screened according to synthesis specification indexes by using a Bayesian optimization method, wherein the Bayesian optimization is to guide the synthesis process by establishing a probability model of the synthesizable set, so that data synthesis configuration for enabling the synthesizable set to obtain optimal values is found, and the quality standard-reaching synthesizable data of the products meet the synthesis specification indexes, thereby realizing the synthesis operation of the quality standard-reaching synthesizable sets of the products.
Further, in a specific embodiment, the composite benefit value of the transformation model may be obtained by assigning different weights to each factor by using factors such as a composite score, a composite condition, and a composite error of the transformation model, multiplying the score of each factor by the weight, and then summing the weights.
In this embodiment, the synthesis benefit value of the transformation model is obtained by performing synthesis operation on the quality standard synthesizable set of each product, and specifically includes the synthesis completion degree and the synthesis matching degree of the transformation model, and the synthesis benefit value of the transformation model is obtained by comprehensive analysis and is imported into the following analysis formula:
In this embodiment, when the synthesis completion degree of the transformation model and the synthesis matching degree of the transformation model are increased, the synthesis benefit value of the transformation model is also increased, which is helpful for a production mechanism to more accurately evaluate the influence of the synthesis benefit of the transformation model on the product quality reaching the standard, so that a more reasonable quality reaching synthesis policy and decision can be formulated, and meanwhile, a more compact algorithm operation is adopted for the expression, so that the synthesis benefit condition of the transformation model can be known more clearly and accurately, and the synthesis benefit reaching the product quality reaching the standard is in a high-efficiency state.
W is the synthesis completion of the transformation model, and represents the synthesis completion of the transformation model obtained by comprehensively analyzing the data item completion rate and the data synthesis time efficiency of the transformation model.
In this embodiment, the synthesis completion of the transformation model is set to 93%, and the weight factor corresponding to the synthesis completion is set to 0.62, so that the calculation result of the W (1-g 1) portion in the expression is 0.35.
P is the synthetic matching degree of the transformation model, and represents the synthetic matching degree of the transformation model obtained through the comprehensive analysis of the data synthesis consistency and the data integrity of the transformation model.
G 1 is a weight factor corresponding to a predefined synthesis completion degree, g 2 is a weight factor corresponding to a predefined synthesis matching degree, wherein the weight factors corresponding to the synthesis completion degree and the weight factor corresponding to the synthesis matching degree are the influence of different synthesis completion degrees and the synthesis matching degree on the final result of the synthesis benefit of the transformation model through simulation test, the influence of different synthesis completion degrees and the synthesis matching degree on the synthesis completion degree is determined according to the result of sensitivity analysis, and the weight factors are distributed according to the influence, in the embodiment, the weight factor corresponding to the synthesis completion degree has a value range of 0.58 to 0.65, the weight factor corresponding to the synthesis matching degree has a value range of 0.49 to 0.55, and e is a natural constant.
The numerical values of the synthesis completion degree of the transformation model are shown in table 1:
TABLE 1 completion of synthesis of transformation models
In this embodiment, according to the table of the synthesis completion degree of the transformation model, when the data item completion rate and the data synthesis time efficiency increase, the synthesis completion degree of the transformation model also increases, that is, there is a proportional relationship between the data item completion rate and the data synthesis time efficiency and the synthesis completion degree of the transformation model, so if the synthesis completion degree of the transformation model is to be improved, it is necessary to optimize the synthesis and the synthesis time of the data item, so as to improve the data item completion rate and the data synthesis time efficiency, and ensure that the synthesis completion degree of the transformation model is in a better state.
The numerical values of the synthetic matching degree of the transformation model are shown in table 2:
Table 2 synthetic match of transformation model
In this embodiment, according to the table of the synthetic matching degree of the transformation model, when the data synthetic matching degree and the data integrity are increased, the synthetic matching degree of the transformation model is also increased, that is, there is a proportional relationship between the data synthetic matching degree and the data integrity and the synthetic matching degree of the transformation model, so if the synthetic matching degree of the transformation model is to be improved, the data item completion rate and the data synthetic time efficiency need to be improved, and the synthetic matching degree of the transformation model is ensured to reach an ideal state.
And optimizing the data synthesis service, namely comparing the synthesis benefit value of the transformation model with a set synthesis benefit threshold value, and matching a data synthesis optimization plan if the synthesis benefit value of the transformation model is lower than the synthesis benefit threshold value, so as to realize the data synthesis process of the quality of the productive products.
Specifically, the data synthesis process for realizing the quality of the productive product comprises the following specific matching process:
Comparing the synthesis benefit value of the transformation model with a predefined synthesis benefit threshold, wherein the synthesis benefit threshold can be directly extracted from a product quality management library, if the synthesis benefit value of the transformation model is lower than the synthesis benefit threshold, carrying out difference processing on the synthesis benefit value of the transformation model and the synthesis benefit threshold to obtain a synthesis benefit deviation value of the transformation model, and matching with a data synthesis optimization plan corresponding to each synthesis benefit deviation value interval.
The data synthesis optimization plan specifically comprises the steps of screening data sources required by product quality data synthesis, introducing a transformation model to simplify a data synthesis flow, establishing a data synthesis real-time monitoring system, carrying out real-time monitoring and adjustment on a synthesis process, timely finding problems and carrying out adjustment according to a real-time monitoring result, and ensuring the stability and reliability of data synthesis.
The financial institution can also evaluate the risk and opportunity of the whole product production line supply chain by utilizing the benefit value synthesized by the transformation model data, and provide financial service for each link in the product production line supply chain, and meanwhile, the quality evaluation result can also help the financial institution to identify potential market risks and promote innovation and development of financial markets.
Referring to fig. 2, a second aspect of the present invention provides a system for a productive financial service data synthesis method based on large model reasoning, which comprises a data set collection module, a transformation model extraction module, a product quality synthesizable set matching module, a synthesis benefit value analysis module of a transformation model, and a data synthesis service optimization module.
The second aspect of the invention provides a system for synthesizing productive financial service data based on large model reasoning, which further comprises a product quality management library, wherein the product quality management library is used for storing quality standard-reaching synthesizable sets corresponding to all quality standard-reaching deviation value intervals.
The data set collection module is connected with the transformation model extraction module, the transformation model extraction module and the product quality synthesizable set matching module are both connected with the synthesis benefit value analysis module of the transformation model, the synthesis benefit value analysis module of the transformation model is connected with the data synthesis service optimization module, and the product quality synthesizable set matching module and the synthesis benefit value analysis module of the transformation model are both connected with the product quality management library.
The data set collecting module is used for collecting operation data of production equipment of each product in the production process, supply chain information of a production line and weight and volume of each synthesized product, and the operation data, the supply chain information and the weight and the volume are respectively recorded as a first data set, a second data set and a third data set of each product, and therefore the operation data, the supply chain information and the weight and the volume of each synthesized product enter the transformation model extracting module.
The transformation model extraction module is used for respectively constructing a first data set, a second data set and a third data set of each product into a first data set model, a second data set model and a third data set model, obtaining the output result of each data set model by obtaining the historical data of each product as the input of each data set model, screening out the data set model with the highest data synthesis adaptation degree, marking the data set model as a transformation model, and entering the synthesis benefit value analysis module of the transformation model.
The product quality synthesizable set matching module is used for acquiring data set information of each product, evaluating quality standard indexes of each product, and matching out quality standard synthesizable sets of each product, so that the product quality standard indexes enter the synthesis benefit value analysis module of the transformation model.
The synthesis benefit value analysis module of the transformation model is used for carrying out synthesis operation on the quality standard synthesizable set of each product based on the transformation model, analyzing the synthesis benefit value of the transformation model and entering the data synthesis service optimization module.
The data synthesis service optimization module is used for comparing the synthesis benefit value of the transformation model with a set synthesis benefit threshold value, and matching a data synthesis optimization plan if the synthesis benefit value of the transformation model is lower than the synthesis benefit threshold value, so that a data synthesis process of the quality of the productive product is realized.
The foregoing is merely illustrative of the structures of this invention and various modifications, additions and substitutions for those skilled in the art can be made to the described embodiments without departing from the scope of the invention or from the scope of the invention as defined in the accompanying claims.
Claims (6)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410686464.6A CN118673316B (en) | 2024-05-30 | 2024-05-30 | Productive financial service data synthesis method and system based on large model reasoning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410686464.6A CN118673316B (en) | 2024-05-30 | 2024-05-30 | Productive financial service data synthesis method and system based on large model reasoning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118673316A CN118673316A (en) | 2024-09-20 |
| CN118673316B true CN118673316B (en) | 2025-01-28 |
Family
ID=92720639
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410686464.6A Active CN118673316B (en) | 2024-05-30 | 2024-05-30 | Productive financial service data synthesis method and system based on large model reasoning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118673316B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112101552A (en) * | 2020-09-25 | 2020-12-18 | 北京百度网讯科技有限公司 | Method, apparatus, device and storage medium for training a model |
| CN115410014A (en) * | 2022-08-25 | 2022-11-29 | 同济大学 | Self-supervision characteristic point matching method of fisheye image and storage medium thereof |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2177898A1 (en) * | 2008-10-14 | 2010-04-21 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Method for selecting an optimized evaluation feature subset for an inspection of free-form surfaces and method for inspecting a free-form surface |
| CN116976682B (en) * | 2023-09-22 | 2023-12-26 | 安徽融兆智能有限公司 | A method for evaluating the operating status of electricity consumption information collection system based on fuzzy algorithm |
| CN117455242B (en) * | 2023-11-27 | 2025-02-07 | 南京华控创为信息技术有限公司 | A water conservancy management system based on digital twin |
| CN117911392A (en) * | 2024-01-30 | 2024-04-19 | 中国人民解放军火箭军工程大学 | Reconstruction method, reconstruction device, computer equipment and medium of nuclear magnetic resonance image |
-
2024
- 2024-05-30 CN CN202410686464.6A patent/CN118673316B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112101552A (en) * | 2020-09-25 | 2020-12-18 | 北京百度网讯科技有限公司 | Method, apparatus, device and storage medium for training a model |
| CN115410014A (en) * | 2022-08-25 | 2022-11-29 | 同济大学 | Self-supervision characteristic point matching method of fisheye image and storage medium thereof |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118673316A (en) | 2024-09-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6817426B2 (en) | Yield prediction system and method for machine learning-based semiconductor manufacturing | |
| JP2021068435A (en) | Supplier evaluation device and supplier evaluation method | |
| CN115796372B (en) | A supply chain management optimization method and system based on SCOR | |
| CN107664690B (en) | A method of prediction gas dissolved in oil of power trans-formers | |
| CN119323333B (en) | Supply chain multiparty sharing management system based on digital transaction platform | |
| CN117541057A (en) | Enterprise operation early warning monitoring method and system based on data analysis | |
| CN108663501A (en) | A kind of predicting model for dissolved gas in transformer oil method and system | |
| CN117951192A (en) | A method and system for managing carbon emission monitoring data based on data mining | |
| Liao et al. | Reliability proactive control approach based on product key reliability characteristics in manufacturing process | |
| CN118863520A (en) | Risk assessment method and system for entrepreneurial incubation projects based on market demand | |
| CN118839142A (en) | Method and device for detecting center hole of wire drawing die | |
| CN118586765A (en) | A quality management risk control method and system based on big data | |
| CN118735413A (en) | Warehouse management model determination method and system | |
| CN118469094A (en) | A canvas system based on supply chain | |
| CN118916535A (en) | Production site management method and system based on three-dimensional visualization technology | |
| CN118886589A (en) | A land use planning system based on soil organic carbon monitoring | |
| CN118586683A (en) | Duck tongue production control method, device and equipment | |
| CN117391626A (en) | An engineering cost data processing method based on big data | |
| CN118673316B (en) | Productive financial service data synthesis method and system based on large model reasoning | |
| CN119226832B (en) | Cement kiln system heat consumption analysis method based on data analysis | |
| CN119599502A (en) | Enterprise digital supply chain capability evaluation method and system | |
| CN118333552A (en) | Multi-dimensional intelligent analysis and control system for opinion work orders based on natural semantic recognition | |
| CN117520910A (en) | A method and system for predicting user satisfaction within a business management system | |
| CN119067301B (en) | An intelligent monitoring method and system for computer hardware production and assembly | |
| CN120851679A (en) | A system and method for diagnosing enterprise digitalization level based on rule engine |
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 |