CN117413988A - Unconventional large lapel garment platemaking collar matching method - Google Patents

Abstract

The invention discloses an unconventional pattern-making and matching method for large-lapel clothing, which belongs to the technical field of clothing design and production. The specific steps of the collar-fitting method are as follows: (1) Determine the large-lapel clothing pattern-making drawing based on the clothing knowledge map; (2) Collect Fabric information and cutting according to the pattern making drawing; (3) Intelligent sewing of large lapel garments on the cut fabric; (4) Collect sewing data and upload to the design platform; (5) Distributed storage of production data and real-time monitoring and adjustment Platform performance; the present invention can realize the automated sewing process, improve sewing efficiency, speed up production, reduce the need for manual intervention, reduce the risk of human error, improve the quality of plate making, optimize the use of fabrics, avoid unnecessary waste, and enable Designers and practitioners can more easily access and understand knowledge related to clothing design, provide accurate information and guidance, help improve the innovation of design, and improve the quality of clothing design and pattern making.

Description

An unconventional method of pattern making and collar matching for large lapel garments

Technical field

The present invention relates to the technical field of clothing design and production, and in particular to an unconventional pattern making and collar matching method for large lapel clothing.

Background technique

Fashion design is a creative and evolving field that reflects social, cultural and personal changes. In this dynamic field, fashion design is considered an art as well as a complex project. The design and pattern making of clothing is one of the key steps in creating a fashion brand and meeting consumer needs. However, traditional garment pattern making methods are often cumbersome, time-consuming, error-prone, and may limit designers' creativity. Large lapel clothing has always been highly praised for its luxurious and chic appearance, but its pattern making and collar matching require higher technology and skills. The traditional pattern-making process usually requires a lot of manual craftsmanship and complex measurements, which not only increases the cost of pattern-making, but may also result in imperfect garments. Therefore, seeking an unconventional pattern-making and matching method for large-lapel garments to improve efficiency, reduce errors, and encourage more innovation has become the focus of the fashion industry; therefore, an unconventional large-lapel garment was invented. The plate making and matching method has become particularly important.

After searching, Chinese patent number CN110558659A discloses an unconventional pattern-making and matching method for large lapel clothing. Although this invention promotes the upgrading of teaching materials in clothing colleges, thereby improving the teaching quality and employment rate, and promoting the transformation, upgrading and prosperity of the clothing industry, it The automatic sewing process cannot be carried out, the production speed is slow, and a large amount of manual intervention is required, which increases the risk of human error; in addition, the existing pattern making and matching method of large lapel clothing is not conducive to improving the innovation of design, reducing the cost of clothing design and The quality of pattern making cannot provide accurate information and guidance; therefore, we propose an unconventional pattern making and matching method for large lapel clothing.

Contents of the invention

The purpose of the present invention is to propose an unconventional pattern making and matching method for large lapel clothing in order to solve the defects existing in the prior art.

In order to achieve the above objects, the present invention adopts the following technical solutions:

An unconventional pattern-making and matching method for large lapel clothing. The specific steps of this collar matching method are as follows:

(1) Determine the layout of large lapel clothing based on the clothing knowledge map;

(2) Collect fabric information and cut according to the pattern making drawing;

(3) Intelligent sewing of large lapel garments from the cut fabrics;

(4) Collect sewing data and upload it to the design platform;

(5) Distributed storage of production data and real-time monitoring and adjustment of platform performance.

As a further solution of the present invention, the specific steps for determining the pattern making of the large lapel garment described in step (1) are as follows:

Step 1: Collect data on fashion design, lapels, fabrics, and fashion history from different data sources, then preprocess the collected data, and then use natural language processing technology to identify entities, relationships, and attributes in the text, and use RDF or The data collected by OWL is standardized and each set of entities and attributes is assigned a unique identifier;

Step 2: Extract the corresponding attributes of each group of entities in each group of knowledge information, and establish the relationship between the entities to form a connection with the clothing knowledge graph. Use the form of triples to process the entities, attributes, and relationships into corresponding graph structures. Generate a clothing knowledge graph, and continuously update and maintain the clothing knowledge graph;

Step 3: The designer queries the detailed knowledge of different fashion elements, historical fashion trends, fabric characteristics, large lapel shapes, materials and historical designs through the clothing knowledge graph. Then the designer determines the concept and style of the clothing based on the feedback knowledge information, and then extracts The concept of clothing and the entity information in the style are matched with the clothing knowledge graph to provide information about different fabrics and tailoring methods, and help designers generate relevant clothing pattern drawings.

As a further solution of the present invention, the specific steps for cutting the fabric described in step (2) are as follows:

Step 1: Collect fabric information in real time through surveillance cameras, and extract a certain number of image frames from the video data every second to cover the entire video, obtain each group of pixel values in each group of image frames, and then use the pixel value for each group of pixels. The average value of the surrounding pixels replaces the value of the central pixel, and the above process is repeated to process the entire image to remove any existing noise, watermarks, or distortions, and adjust the image size and resolution so that each set of images has the same size;

Step 2: Count the pixel distribution of different color channels in the image, and then use a window with specified pixels to move among each set of image information. Calculate the gray-level co-occurrence matrix under the window at each move, and calculate the relevant image information from the gray-level co-occurrence matrix. For the texture features in the image pyramid, the optimized image information is scaled and normalized, and the features of each group of image information are extracted, and then feature fusion is performed through the bidirectional feature pyramid to obtain the target detection frame;

Step 3: Expand and crop each image information according to the target detection frame to obtain the target image, and then store the obtained texture features in the form of an array to the corresponding pixel position. When the texture features meet the preset conditions, the current pixel is judged The area is a wrinkled area. If it is not satisfied, the current pixel area is judged to be a normal area, and the wrinkled area is smoothed based on the judgment result;

Step 4: Use the Harris corner detection method to identify the fabric corners in the image, then use the SIFT algorithm to identify key points in the image and extract feature descriptors, and then use the Hough transform to detect straight lines or curves in the image to find clipping line position;

Step 5: Detect each group of clipping point candidates based on the extracted feature information and the position of the clipping line, cluster the detected candidate points to determine the clipping line, and then determine the positions of each group of clipping points based on the determined clipping lines, and add them in the image Make a mark.

As a further solution of the present invention, the specific calculation formula of the Harris corner point detection method described in step 4 is as follows:

R＝det(M)-k*(trace(M)) ²

(1)

In the formula, M represents the matrix including image gradient information; det(M) represents the determinant; trace(M) represents the trace; k represents a constant;

The specific calculation formula of Hough transform mentioned in step 4 is as follows:

ρ＝x*cos(θ)+y*sin(θ)

(2)

In the formula, (ρ, θ) represents the straight line parameters in polar coordinate space; (x, y) represents the point coordinates in the image.

As a further solution of the present invention, the specific steps for intelligent sewing of large lapel clothing described in step (4) are as follows:

Step 1: Collect the setting parameters of the sewing machine, different types of fabrics, sewing thread characteristics, and detailed information about different sewing parts, label each set of data collected, and then normalize each set of data to remove noise. And after processing missing values, divide them into training sets and test sets;

Step ②: Create an intelligent sewing model and define the model loss function, then select the stochastic gradient descent algorithm to adjust the model parameters to minimize the loss function, divide the training set into small batches, and use each batch to update the weight of the model, Repeatedly perform forward propagation to calculate the output of the model, and then perform backpropagation to calculate the gradient and update the parameters of the model, until all training sets have been used;

Step ③: Use an independent test set to evaluate the performance of the trained model, and calculate the loss value of the model based on the model evaluation results, then replace the test set with another subset, and then take the remaining subset as the training set, and again Calculate the loss value until all data are predicted once, and select the combination parameter corresponding to the minimum loss value as the optimal parameter in the data interval and replace the original parameters of the intelligent sewing model;

Step ④: Collect fabric sewing information in real time and input it into the intelligent sewing model. After each hidden layer processes the input data separately, it passes it layer by layer through the weights and activation functions between each layer, and then outputs the layer decoder. The processed data is decoded to obtain fabric sewing prediction data, and real-time sewing control is performed based on the prediction data.

As a further solution of the present invention, the specific steps for creating distributed data storage as described in step (5) are as follows:

Step Ⅰ: Divide each group of sewing data according to the preset time interval to form multiple groups of data blocks, and then use a hash algorithm to generate the identification of each group of data blocks and collect the node information of each group;

Step II: Obtain the data block division rules and node load conditions, and select appropriate nodes to store each group of data blocks through the load balancing algorithm. After the data block storage is completed, configure and copy the specified number of data blocks according to the system requirements and available resources. to multiple groups of nodes;

Step III: When the data stored in the node changes, the data update is propagated from one node to other nodes through the data synchronization algorithm, and then the node operation status is automatically detected, and the faulty node is migrated or repaired.

As a further solution of the present invention, the specific steps for platform performance adjustment described in step (5) are as follows:

Step 1: Determine the accessed data in the system and the data with large computational overhead and the corresponding pointer structure based on the administrator's preset information, determine the linked list node structure based on the data object and pointer structure, create an empty linked list, and at the same time, according to the system memory Resource and performance requirements set the maximum capacity of the linked list;

Step 2: When you need to access data, look up the data in the cache linked list. If the data exists in the linked list, move it to the head of the linked list, indicating that it has been used recently. If the data is not in the linked list, retrieve it from the blockchain or Other data sources obtain data and add it to the head of the linked list, and regularly monitor the length of the linked list, cache hit rate and performance indicators;

Step 3: When the cache capacity reaches the upper limit, determine the data that has not been accessed for the longest time in the linked list based on the most recent access time, remove the corresponding data node from the tail of the linked list and release the resources, and at the same time update the head pointer of the linked list to The new head node records the cache hit rate and the number of eviction operations, and regularly monitors platform performance.

Compared with the existing technology, the beneficial effects of the present invention are:

1. This unconventional large lapel garment pattern making and collar method uses image recognition technology to obtain the position of the fabric cutting point, and at the same time processes the fabric folds, and then collects the setting parameters of the sewing machine, different types of fabrics, sewing thread characteristics and sewing Detailed information of different parts, and label each set of data collected, and then preprocess each set of data and divide it into a training set and a test set, create an intelligent sewing model and train it with the training set, and then Use an independent test set to evaluate the performance of the trained model, collect fabric sewing information in real time and input it into the intelligent sewing model. After each hidden layer processes the input data separately, it passes it through the weights and activations between each layer. The function is passed layer by layer, and then the output layer decoder decodes the processed data to obtain fabric sewing prediction data, and performs real-time sewing control based on the prediction data, which can realize the automated sewing process, improve sewing efficiency, and speed up production. Speed, reducing the need for manual intervention, reducing the risk of human error, improving pattern making quality, optimizing fabric use and avoiding unnecessary waste.

2. This unconventional large lapel clothing pattern making and matching method collects data on fashion design, large lapels, fabrics, and fashion history from different data sources, and then preprocesses the collected data, and then uses natural language processing technology to identify text The entities, relationships and attributes in the data are standardized through RDF or OWL, and a unique identifier is assigned to each group of entities and attributes. The corresponding attributes of each group of entities in each group of knowledge information are extracted, and the relationships between entities are established. The relationship forms the connection of the clothing knowledge graph. Entities, attributes and relationships are processed into corresponding graph structures in the form of triples to generate the clothing knowledge graph, and the clothing knowledge graph is continuously updated and maintained. Designers use clothing The knowledge graph queries the detailed knowledge of different fashion elements, historical fashion trends, fabric characteristics, lapel shapes, materials and historical designs. Then the designer determines the concept and style of the clothing based on the feedback knowledge information, and then extracts the concept and style of the clothing. Entity information and matching it with the clothing knowledge graph to provide information about different fabrics and cutting methods, and help designers generate relevant clothing layouts, making it easier for designers and practitioners to access and understand clothing design-related Knowledge, providing accurate information and guidance, helps to improve the innovation of design and improves the quality of clothing design and pattern making.

Description of the drawings

The drawings are used to provide a further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

Figure 1 is a flow chart of an unconventional large lapel garment pattern making and collar matching method proposed by the present invention.

Detailed ways

Example 1

Referring to Figure 1, an unconventional pattern-making and collar-fitting method for large lapel clothing is shown. The specific steps of this collar-fitting method are as follows:

Determine the layout of large lapel clothing based on the clothing knowledge map.

Specifically, data about fashion design, lapels, fabrics, and fashion history are collected from different data sources, and then the collected data are preprocessed, and then natural language processing technology is used to identify entities, relationships, and attributes in the text, through RDF or The data collected by OWL is standardized, and a unique identifier is assigned to each group of entities and attributes. The corresponding attributes of each group of entities in each group of knowledge information are extracted, and the relationship between entities is established to form a connection to the clothing knowledge graph. Using In the form of triples, entities, attributes and relationships are processed into corresponding graph structures to generate clothing knowledge graphs, and the clothing knowledge graphs are continuously updated and maintained. Designers can query different fashion elements, historical fashion trends, etc. through the clothing knowledge graphs. Detailed knowledge of fabric characteristics, large lapel shape, material and historical design. Then the designer determines the concept and style of the clothing based on the feedback knowledge information, and then extracts the entity information in the clothing concept and style, and compares it with the clothing knowledge map Match to provide information about different fabrics and cuts, and help designers generate relevant garment patterns.

Collect fabric information and cut according to pattern drawing.

Specifically, the fabric information is collected in real time through the surveillance camera, and a certain number of image frames are extracted from the video data every second to cover the entire video, and each group of pixel values in each group of image frames is obtained. Then for each group of pixels, the pixels are used. The average value of the surrounding pixels replaces the value of the central pixel. Repeat the above process to process the entire image to remove existing noise, watermarks or distortions, and adjust the image size and resolution so that each group of images has the same size. Statistics of different color channels in the image Pixel distribution, and then use the specified pixel window to move in each group of image information. Each time it moves, the gray level co-occurrence matrix under the window is calculated. The texture features in the relevant image information are calculated from the gray level co-occurrence matrix, and the image pyramid is optimized. The resulting image information is scaled and normalized, and the features of each group of image information are extracted. Then feature fusion is performed through a two-way feature pyramid to obtain the target detection frame. Each image information is enlarged and cropped based on the target detection frame to obtain the target image. , and then store the obtained texture features in the form of an array to the corresponding pixel location. When the texture features meet the preset conditions, the current pixel area is judged to be a wrinkle area. If not, the current pixel area is judged to be a normal area, and According to the judgment results, the wrinkle area is smoothed, and the fabric corners in the image are identified through the Harris corner detection method. Then the SIFT algorithm is used to identify the key points in the image, and the feature descriptors are extracted, and then the Hough transform is used to detect the fabric corners in the image. Use straight lines or curves to find the position of the cropping line, detect each group of cropping point candidates based on the extracted feature information and the position of the cropping line, cluster the detected candidate points to determine the cropping line, and then determine each group of cropping based on the determined cropping line. Click the location and mark it in the image.

In this embodiment, the specific calculation formula of the Harris corner point detection method is as follows:

R＝det(M)-k*(trace(M)) ²

(1)

In the formula, M represents the matrix including image gradient information; det(M) represents the determinant; trace(M) represents the trace; k represents a constant;

The specific calculation formula of Hough transform is as follows:

ρ＝x*cos(θ)+y*sin(θ)

(2)

In the formula, (ρ, θ) represents the straight line parameters in polar coordinate space; (x, y) represents the point coordinates in the image.

Example 2

Referring to Figure 1, an unconventional pattern-making and collar-fitting method for large lapel clothing is shown. The specific steps of this collar-fitting method are as follows:

Intelligently sew large lapel garments from the cut fabric.

Specifically, the setting parameters of the sewing machine, different types of fabrics, the characteristics of the sewing thread, and the detailed information of different parts of the sewing are collected, and each set of data collected is marked, and then each set of data is normalized and noise is removed. And after processing the missing values, divide them into training sets and test sets, create an intelligent sewing model and define the loss function of the model, then select the stochastic gradient descent algorithm to adjust the model parameters to minimize the loss function, and divide the training set into small batches , and use each batch to update the weights of the model, repeatedly perform forward propagation to calculate the output of the model, and then perform back propagation to calculate the gradient and update the parameters of the model, until all training sets have been used, stop using independent The test set is used to evaluate the performance of the trained model, and based on the model evaluation results, the loss value of the model is statistically calculated, and then the test set is replaced with another subset, and then the remaining subset is taken as the training set, and the loss value is calculated again until the All data are predicted once. By selecting the combination parameters corresponding to the minimum loss value as the optimal parameters in the data interval and replacing the original parameters of the intelligent sewing model, the fabric sewing information is collected in real time and input into the intelligent sewing model. After the hidden layer processes the input data respectively, it is passed layer by layer through the weights and activation functions between each layer. Then the output layer decoder decodes the processed data to obtain fabric sewing prediction data, and based on the prediction Data for real-time sewing control.

Collect sewing data and upload it to the design platform.

Distributed storage of production data and real-time monitoring and adjustment of platform performance.

Specifically, each group of sewing data is divided according to the preset time interval to form multiple groups of data blocks, and then the identification of each group of data blocks is generated through a hash algorithm, each group of node information is collected, and the data block division rules are obtained. Node load conditions, and select appropriate nodes to store each group of data blocks through the load balancing algorithm. After the data block storage is completed, the specified number of data blocks are copied to multiple groups of nodes according to the system requirements and available resources. When the node stores When the data changes, the data update is propagated from one node to other nodes through the data synchronization algorithm, and then the node running status is automatically detected, and the faulty node is migrated or repaired.

Specifically, the accessed data in the system, data with large computational overhead, and corresponding pointer structures are determined based on the administrator's preset information, and the linked list node structure is determined based on the data objects and pointer structures, and an empty linked list is created. At the same time, based on the system memory resources Set the maximum capacity of the linked list and performance requirements. When data needs to be accessed, the data is searched in the cache linked list. If the data exists in the linked list, it is moved to the head of the linked list, indicating that it has been used recently. If the data is not in the linked list, then Obtain data from the blockchain or other data sources and add it to the head of the linked list. Regularly monitor the length of the linked list, cache hit rate and performance indicators. When the cache capacity reaches the upper limit, determine the most recent access time in the linked list. Data that has not been accessed for a long time, and the corresponding data node is removed from the tail of the linked list and the resources are released. At the same time, the head pointer of the linked list is updated to the new head node, the cache hit rate and the number of elimination operations are recorded, and the platform performance is regularly monitored. .

Claims (7)

1. The unconventional large lapel clothing platemaking collar matching method is characterized by comprising the following specific steps of:

(1) Determining a large lapel garment layout according to the garment knowledge graph;

(2) Collecting fabric information and cutting according to the layout;

(3) Intelligently sewing large lapel clothing on the cut fabric;

(4) Collecting sewing data and uploading the sewing data to a design platform;

(5) And storing the production data in a distributed manner and monitoring and adjusting the performance of the platform in real time.

2. The non-conventional large lapel garment platemaking and collar matching method according to claim 1, wherein the large lapel garment platemaking and collar matching method in step (1) comprises the following specific steps:

step one: collecting related fashion design, lapel, fabric and fashion history data from different data sources, preprocessing the collected data, identifying entities, relationships and attributes in the text by using natural language processing technology, carrying out standardized processing on the data collected by RDF or OWL, and distributing a unique identifier for each group of entities and attributes;

step two: extracting corresponding attributes of each group of entities in each group of knowledge information, establishing a relation among the entities to form connection of clothing knowledge graphs, processing the entities, the attributes and the relation into corresponding graph structures in a form of triples to generate the clothing knowledge graphs, and continuously updating and maintaining the clothing knowledge graphs;

step three: the designer inquires detailed knowledge of different simultaneous elements, historic fashion trends, fabric characteristics, lapel shapes, materials and historic designs through the clothing knowledge graph, then the designer determines the concept and style of the clothing according to the feedback knowledge information, extracts entity information in the concept and style of the clothing, matches the entity information with the clothing knowledge graph to provide information about different fabrics and cutting modes, and helps the designer generate relevant clothing making patterns.

3. The non-conventional large lapel garment platemaking collar matching method according to claim 2, wherein the specific fabric tailoring step in step (2) is as follows:

step 1: acquiring fabric information in real time through a monitoring camera, extracting a certain number of image frames from video data every second to cover the whole video, acquiring each group of pixel values in each group of image frames, replacing the value of a central pixel with the average value of pixels around each group of pixels, repeating the process to process the whole image to remove noise, watermark or distortion, and adjusting the size and resolution of the image to enable each group of images to have the same size;

step 2: counting pixel distribution of different color channels in an image, then moving in each group of image information by using a window for defining pixels, calculating a gray level co-occurrence matrix under the window every time when the window is moved, calculating texture features in related image information from the gray level co-occurrence matrix, performing scale normalization processing on the optimized image information through an image pyramid, extracting features of each group of image information, and performing feature fusion through a bidirectional feature pyramid to obtain a target detection frame;

step 3: enlarging and cutting each image information according to a target detection frame to obtain a target image, storing the obtained texture features in corresponding pixel positions in an array mode, judging the current pixel area as a wrinkle area when the texture features meet preset conditions, judging the current pixel area as a normal area if the texture features do not meet the preset conditions, and finishing the wrinkle area to be smooth according to a judgment result;

step 4: identifying fabric corner points in the image by a Harris corner point detection method, identifying key points in the image by a SIFT algorithm, extracting feature descriptors, and detecting straight lines or curves in the image according to Hough transformation to find the positions of cutting lines;

step 5: and detecting each group of clipping point candidates according to the extracted characteristic information and clipping line positions, clustering the detected candidate points to determine clipping lines, determining each group of clipping point positions according to the determined clipping lines, and marking in the image.

4. The method for producing and matching irregular large lapel clothing according to claim 3, wherein the specific calculation formula of the Harris corner detection method in step 4 is as follows:

R＝det(M)-k*(trace(M)) ²

(1)

wherein M represents a matrix including image gradient information; det (M) represents a determinant; trace (M) represents trace; k represents a constant;

the specific calculation formula of Hough transformation in the step 4 is as follows:

ρ＝x*cos(θ)+y*sin(θ)

(2)

wherein (ρ, θ) represents a straight line parameter in the polar coordinate space; (x, y) represents the coordinates of points in the image.

5. A non-conventional large lapel garment platemaking collar matching method according to claim 3, wherein the large lapel garment intelligent sewing specific steps in the step (4) are as follows:

step (1): collecting the setting parameters of a sewing machine, different types of fabrics, characteristics of sewing threads and detailed information of sewing different parts, marking each group of collected data, carrying out normalization processing, noise removal and missing value processing on each group of data, and dividing the data into a training set and a testing set;

step (2): creating an intelligent sewing model, defining a model loss function, then selecting a random gradient descent algorithm to adjust model parameters to minimize the loss function, dividing a training set into small batches, updating the weight of the model by using each group of batches, repeatedly executing the output of a forward propagation calculation model, then executing the backward propagation to calculate the gradient and update the parameters of the model until all the training sets are stopped after being used;

step (3): evaluating the performance of the trained model by using an independent test set, counting the loss value of the model according to the model evaluation result, replacing the test set with another subset, taking the rest subset as the training set, calculating the loss value again until all data are predicted once, and selecting the corresponding combined parameter with the minimum loss value as the optimal parameter in the data interval and replacing the original parameter of the intelligent sewing model;

step (4): the fabric sewing information is collected in real time and is input into an intelligent sewing model, after the input data are processed by each hidden layer, the input data are transmitted layer by layer through weights and activation functions among the layers, and then the processed data are decoded by an output layer decoder to obtain fabric sewing prediction data, and real-time sewing control is carried out according to the prediction data.

6. The method for making and matching a non-conventional large lapel garment according to claim 5, wherein the step (5) of storing the production data in a distributed manner comprises the following steps:

step I: dividing each group of sewing data according to a preset time interval to obtain a plurality of groups of data blocks, generating the identification of each group of data blocks through a hash algorithm, and collecting each group of node information;

step II: acquiring a data block dividing rule and a node load condition, selecting a proper node to store each group of data blocks through a load balancing algorithm, and after the data blocks are stored, configuring and copying a specified number of data blocks to a plurality of groups of nodes according to the requirements of a system and available resources;

step III: when the data stored by the nodes changes, the data update is transmitted from one node to other nodes through a data synchronization algorithm, then the node operation condition is automatically detected, and the data migration or repair is carried out on the fault node.

7. The non-conventional large lapel garment platemaking collar matching method as claimed in claim 1, wherein said platform performance adjustment in step (5) is specifically as follows:

the first step: determining accessed data, data with high calculation cost and corresponding pointer structures in a system according to preset information of an administrator, determining a linked list node structure according to data objects and the pointer structures, creating an empty linked list, and setting the maximum capacity of the linked list according to the memory resources and performance requirements of the system;

and a second step of: when the data is required to be accessed, searching the data in a cache chain table, if the data exists in the chain table, moving the data to the head of the chain table to indicate that the data is used recently, and if the data is not in the chain table, acquiring the data from a blockchain or other data sources, adding the data to the head of the chain table, and periodically monitoring the length, the cache hit rate and the performance index of the chain table;

and a third step of: when the cache capacity reaches the upper limit, the data which is not accessed for the longest time in the linked list is judged based on the latest access time, the corresponding data node is removed from the tail of the linked list, resources are released, meanwhile, the head pointer of the linked list is updated to a new head node, the cache hit rate and the times of elimination operation are recorded, and the performance of the platform is monitored periodically.