Intelligent industrial cluster management system based on big data analysis

Abstract

The invention discloses an intelligent industry cluster management system based on big data analysis, and belongs to the technical field of big data analysis and intelligent management, and the system specifically comprises the steps: collecting and preprocessing the research, project and material data of companies of the same type on a supply chain, and generating a target company data set; loading and training a regression analysis prediction model, analyzing material requirements according to a prediction result, determining the type and number of parts, and meanwhile, mining and optimizing a purchase plan through association rules; constructing a potential customer model, matching required parts and recommending potential partners or customers; and feedback data is collected, the potential customer model, part determination and potential customer matching are optimized, a closed-loop optimization system is formed, and the supply chain efficiency and the matching precision are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of big data analysis and intelligent management, and specifically to an intelligent industrial cluster management system based on big data analysis. Background Art

[0002] An industrial cluster is an industrial ecosystem formed by the gathering of many related enterprises, institutions and organizations in a certain area. There are complex relationships and interactions between these enterprises and between enterprises and institutions, involving supply chain management, production management, marketing, technology research and development, etc. There are many shortcomings in the management of traditional industrial clusters. For example, the information exchange channels within the enterprise may not be smooth enough, resulting in a delayed response to changes in the external environment and weak risk resistance. At the same time, vicious competition among enterprises within the industrial cluster may hinder the upgrading and development of the cluster.

[0003] For example, a Chinese patent with publication number CN113902599A discloses an intelligent navigation system for an industrial cluster park, including: a client terminal, an enterprise terminal, a management terminal and a server. The management terminal is used to register and review the client terminal and the enterprise terminal, and to authorize the operation of the client terminal and the enterprise terminal. This technical solution enhances the standardization of enterprise information by reviewing the customer information and enterprise information, and uniformly uploads the enterprise information to the park display page of the client terminal for display, further enhancing the standardized management of enterprises in the park. By uploading the enterprise information to the corresponding display column after the review is passed, the enterprise is promoted, which is also convenient for customers to search and understand, and the display column where the reviewed enterprise is located is marked as reviewed and passed, thereby enhancing customers' trust in the enterprise, thereby further improving the promotion of the enterprise.

[0004] For example, the Chinese patent application with publication number CN101908165A discloses a GIS-based industrial cluster information integration service system and method, including: spatial information management subsystem, attribute information management subsystem, data element entry subsystem, spatial analysis decision subsystem and system management and maintenance subsystem. The GIS-based industrial cluster information integration service system and method described in the technical solution introduces advanced GIS technology, realizes the mapping and correspondence between spatial information and various attribute information, and meets the system users' requirements for the advancement, accuracy and timeliness of geographic information and technical information; uses content-based information processing technology to conduct intelligent analysis through the network, and provides public service support for geographical location, equipment, technology, process, product and other resources for enterprises within and between clusters in a visual and convenient way.

[0005] The above existing technologies all have the following problems: lack of prediction and optimization capabilities; although they provide the display of enterprise information and customer retrieval functions, they lack in-depth mining of customer needs and personalized recommendations; insufficient data analysis depth and prediction capabilities; and at the same time, lack of comprehensive insights into the supply chain and market demand. Summary of the invention

[0006] In view of the shortcomings of the existing technology, the present invention proposes an intelligent industrial cluster management system based on big data analysis, which collects and pre-processes the research, project and material data of similar companies in the supply chain to generate a target company data set; loads and trains the regression analysis prediction model, and analyzes the material demand according to the prediction results to determine the type and quantity of parts. At the same time, it optimizes the procurement plan through association rule mining; builds a potential customer model to match the required parts and recommend potential partners or customers; collects feedback data to optimize the potential customer model, part determination and potential customer matching to form a closed-loop optimization system, thereby improving the supply chain efficiency and matching accuracy.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] Intelligent industrial cluster management system based on big data analysis, including: data processing module, model prediction module, demand analysis module, matching and recommendation module, optimization module;

[0009] The data processing module is used to collect research directions, project data, and material usage of companies of the same type from the supply chain, and perform preprocessing to generate a target company data set;

[0010] The model prediction module is used to load a pre-built regression analysis prediction model and use the target company data set for training to obtain a trained regression analysis prediction model;

[0011] The demand analysis module is used to analyze the materials required for the company's subsequent research based on the forecast results, determine the required parts types and quantities, and optimize the parts procurement plan through association rule mining strategies;

[0012] The matching and recommendation module is used to build a potential customer model using customer portraits and market segmentation methods, match the required parts with the potential customer model, find potential companies or customers, and recommend potential partners or customers through an improved collaborative filtering algorithm;

[0013] The optimization module is used to collect feedback data and optimize the regression analysis prediction model, part determination and potential customer matching process based on the feedback data to form a closed-loop optimization system.

[0014] Specifically, the demand analysis module includes a procurement optimization unit, and the procurement optimization unit is configured with an association rule mining strategy, which is used to discover the association relationship between parts and optimize the parts procurement plan according to the association relationship. The specific steps of the association rule mining strategy include:

[0015] A1: Obtain the target company dataset ,in, represents the nth target company data, and n represents the number of target company data;

[0016] A2: Set a minimum support threshold based on the company's business needs and data characteristics and the minimum confidence threshold ;

[0017] A3: Based on the target company dataset , using a counter to traverse , count the frequency of occurrence of each target company's data , sort out each target company’s data and its corresponding frequency of occurrence to generate a frequency table;

[0018] like , then discard;

[0019] like , then keep The corresponding target company data forms frequent item sets;

[0020] A4: Sort the frequent itemsets in descending order and construct an FP tree in the sorted order. At the same time, insert each target company data into the FP tree in the sorted order of the frequent itemsets and update the node count. In the FP tree, the tree node represents the target company data, and the node count represents the target company data in Frequency of occurrence ,The paths of the tree represent the co-occurrence relationships of the target company data.

[0021] Specifically, the specific steps of the association rule mining strategy also include:

[0022] A5: Generate all non-empty true subsets of frequent itemsets from the FP tree, and for each non-empty true subset Z, calculate the credibility between it and the complement B ,in, Representing Item Sets exist The number of times it appears in Represents the item set Z in The number of times it appears in It implies a logical relationship. represents a union;

[0023] A6: Based on the minimum credibility threshold ,like , then the association rules corresponding to the frequent item set are retained to form the association rule set R;

[0024] A7: Output the filtered association rule set R to form association rules between parts.

[0025] Specifically, the specific steps of the association rule mining strategy also include:

[0026] A8: Interpret the association rules between the generated parts, identify the association relationship between the parts, analyze the purchase pattern based on the parts purchase records, and present the analysis results in the form of an association rule network diagram. At the same time, formulate targeted inventory management strategies based on the analysis results;

[0027] like , then consider bulk purchases or combined orders, where represents the support of the association rule, and , Represents association rules The support of represents the number of items containing association rule A and association rule D, Indicates the total number of items;

[0028] like , then optimize the inventory management strategy, where Indicates the credibility of the association rule;

[0029] A9: Implement the optimized procurement plan and monitor the execution of the procurement plan.

[0030] Specifically, the matching and recommendation module includes a model building unit and a recommendation unit. The model building unit is used to build a potential customer model using customer portraits and market segmentation methods. The specific steps of building the potential customer model include:

[0031] B1: Obtain target company dataset , and extract features from the target company data set, and build customer portraits based on the extracted features;

[0032] B2: Based on the customer portrait data, use clustering algorithms to segment the market into different customer groups, and identify the characteristics of potential customers from the segmented market groups. At the same time, extract data corresponding to the characteristics of potential customers from the customer portrait data to generate customer characteristic data;

[0033] B3: Load the pre-built decision tree model and use the customer feature data to train the pre-built decision tree model so that it can learn the features and patterns of potential customers, obtain the trained decision tree model, and save the trained decision tree model as a potential customer model;

[0034] B4: Input real-time target company data into the potential customer model and calculate the target company data based on the preset threshold. , convert the identification results of the potential customer model into a potential customer list, classify the potential customers, prioritize the potential customers based on the company's potential needs, and output the classified and prioritized potential customer list to the report;

[0035] B5: Monitor and evaluate the potential customer model regularly, and adjust the potential customer model based on the evaluation results.

[0036] Specifically, the recommendation unit is configured with an improved collaborative filtering algorithm, which is used to recommend potential partners or customers. The specific steps of the improved collaborative filtering algorithm include:

[0037] C1: Obtain target company dataset ,according to , build a company-customer matrix , where rows represent companies, columns represent customers, and matrix elements represent the company's behavior records toward customers;

[0038] C2: Using an improved similarity calculation method, using the company-customer matrix Calculate the similarity between companies and store the similarity between companies in a company similarity matrix Among them, The rows and columns in represent companies, and the matrix elements represent the similarity between companies;

[0039] C3: For the target company, sort the companies from high to low according to the company similarity matrix, and select the K companies that are most similar to it as neighboring companies according to the company similarity sorting results;

[0040] C4: Analyze the transaction pattern characteristics of neighboring companies based on their behavior records, and generate a recommendation list for the target company based on the target company's business needs and customer characteristics;

[0041] C5: Use accuracy evaluation indicators to evaluate the recommendation results and optimize the recommendation process based on the evaluation results.

[0042] Specifically, the specific formula of the improved similarity calculation method in C2 is:

[0043] ;

[0044] in, express The similarity with V, and V represent two companies of the same type in the company-customer matrix The behavior records in , m represents the number of companies of the same type, represents the weight of the i-th dimension, express The i-th element of represents the i-th element of V, express The offset in the i-th dimension, represents the offset of V in the i-th dimension, represents the nonlinear term coefficient, and express and The nonlinear transformation function is and Represents a positive number.

[0045] Specifically, the specific steps of C4 include:

[0046] C4.1: Collect behavioral records of neighboring companies, as well as the business needs and customer characteristics of the target company, and pre-process the collected data;

[0047] C4.2: Extract the material characteristic information required for the neighbor company research based on the neighbor company's behavior records;

[0048] C4.3: Based on the extracted material feature information, a clustering algorithm is used to classify neighboring companies into different pattern groups;

[0049] C4.4: Use the similarity calculation formula to match the target company's customer characteristics with the pattern groups of neighboring companies, and find the neighboring company group that is most similar to the target company's customers;

[0050] C4.5: Based on the matching results, select companies whose transaction model characteristics meet the business needs of the target company from the group of neighboring companies that are similar to the target company's customers, and use the products of the companies that meet the needs as recommendations to generate a recommendation list;

[0051] C4.6: Optimize the recommendation list based on feedback from target companies.

[0052] Specifically, the recommendation list generation formula in C4.5 is:

[0053] ;

[0054] in, represents the recommendation degree of target company a for product r, represents the average rating of all products by target company a, represents the set of neighboring companies similar to the target company a, represents the similarity between target company a and neighbor company j, represents the actual rating of product r by neighbor company j, represents the average rating of all products by neighbor company j, express The absolute value of .

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] 1. The present invention proposes an intelligent industrial cluster management system based on big data analysis, and optimizes and improves the architecture, operation steps and processes. The system has the advantages of simple process, low investment and operation costs, and low production work costs.

[0057] 2. The present invention proposes an intelligent industrial cluster management system based on big data analysis. The data processing and prediction system can accurately predict the material requirements and number of parts types of the target company by collecting and analyzing the research directions and project data of similar companies in the supply chain, effectively support the company's research planning and procurement strategies, and improve the efficiency and accuracy of supply chain management.

[0058] 3. The present invention proposes an intelligent industrial cluster management system based on big data analysis. By combining customer portraits, market segmentation and advanced recommendation algorithms, the system can accurately match and recommend potential partners or customers. At the same time, through closed-loop optimization of feedback data, it continuously optimizes the prediction model and matching process, further improving the accuracy and efficiency of management. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is the architecture diagram of the intelligent industrial cluster management system based on big data analysis of the present invention;

[0060] Figure 2 This is a flowchart of the principle of implementing the intelligent industrial cluster management system based on big data analysis of the present invention;

[0061] Figure 3 This is a flow chart for implementing the association rule mining strategy of the intelligent industrial cluster management system based on big data analysis of the present invention;

[0062] Figure 4 A flow chart for constructing a potential customer model for the intelligent industrial cluster management system based on big data analysis in the present invention. DETAILED DESCRIPTION

[0063] Example 1

[0064] See also Figure 1-Figure 2 , an embodiment provided by the present invention: an intelligent industrial cluster management system based on big data analysis, comprising the following steps:

[0065] Data processing module, model prediction module, demand analysis module, matching and recommendation module, optimization module;

[0066] The data processing module is used to collect the research directions, project data, and material usage of similar companies from the supply chain, and perform preprocessing to generate the target company data set;

[0067] Model prediction module, used to load the pre-built regression analysis prediction model and train it using the target company data set to obtain a trained regression analysis prediction model for subsequent real-time prediction;

[0068] Demand analysis module, which is used to analyze the materials required for the company's next research based on the forecast results, determine the types and quantities of parts required, and optimize the parts procurement plan through association rule mining strategies, reduce inventory costs, and improve supply chain efficiency;

[0069] The matching and recommendation module is used to build a potential customer model using customer portraits and market segmentation methods, match the required parts with the potential customer model, identify potential companies or customers, and recommend potential partners or customers through an improved collaborative filtering algorithm;

[0070] The optimization module is used to collect feedback data and optimize the regression analysis prediction model, part determination and potential customer matching process based on the feedback data to form a closed-loop optimization system.

[0071] The model prediction module includes: model loading unit, model training unit, and real-time prediction unit;

[0072] A model loading unit, used to load a pre-built regression analysis prediction model;

[0073] A model training unit, used to train the regression analysis prediction model using the target company data set, and adjust the regression analysis prediction model parameters to improve the prediction accuracy;

[0074] The real-time prediction unit is used to input the real-time acquired data into the trained regression analysis prediction model to perform real-time prediction.

[0075] The demand analysis module includes: demand forecasting unit, parts determination unit, and procurement optimization unit;

[0076] Demand forecasting unit, used to analyze the materials needed for the company's next research based on the forecast results;

[0077] Furthermore, the specific process of analyzing the materials required for the company's research based on the forecast results includes:

[0078] (1) Obtain and analyze the forecast results obtained by the model forecast module, where the forecast results include accurate information on the quantity, type, and change trend of materials required for the company's research projects in different time periods. At the same time, the forecast results may be presented in the form of charts, reports, or data tables;

[0079] (2) Compare the forecast results with historical data to analyze the growth or decrease trend of material demand. At the same time, it is also necessary to combine seasonal factors, industry development trends, etc. to more comprehensively understand the reasons for changes in material demand;

[0080] (3) Further analyze the material supply situation, including evaluating factors such as supplier reliability, delivery cycle, and material quality to ensure that the company can obtain the required materials in a timely and stable manner. In addition, it is necessary to consider material substitutes and alternatives to reduce supply chain risks.

[0081] A parts determination unit that uses bill of materials and supply chain management knowledge to determine the type and quantity of parts required;

[0082] The procurement optimization unit discovers the association relationships between parts by introducing association rule mining strategies, and optimizes the parts procurement plan based on these relationships, such as batch procurement, order consolidation, etc., to reduce procurement costs and improve supply chain efficiency.

[0083] The matching and recommendation module includes: model building unit, matching unit, and recommendation unit;

[0084] Model building unit, used to build potential customer models using customer profiling and market segmentation methods;

[0085] A matching unit is used to match the required parts with the potential customer model to identify potential companies or customers;

[0086] The recommendation unit is used to introduce an improved collaborative filtering algorithm to recommend potential partners or customers.

[0087] Specifically, the overall implementation process of the intelligent industrial cluster management system based on big data analysis includes:

[0088] Step S1: Collect the research direction, project data, and material usage of similar companies through supply chain data, clean the initial data set, remove redundant and abnormal data, and perform standardization to form the target company data set;

[0089] Step S2: Load the pre-built regression analysis prediction model and train it using the target company data set to obtain a trained regression analysis prediction model;

[0090] Step S3: Input the target company data set acquired in real time into the trained regression analysis prediction model to make predictions on the company's research direction, material quantity and type;

[0091] Step S4: Based on the prediction results, the company analyzes the materials needed for the next research, and uses the bill of materials and supply chain management knowledge to determine the types and quantities of parts required. At the same time, the association rule mining strategy is introduced to discover the association relationship between parts. Based on the association relationship, the parts procurement plan is optimized to reduce inventory costs and improve supply chain efficiency.

[0092] Step S5: Use customer portraits and market segmentation methods to build a potential customer model, match the required parts with the potential customer model, find out potential companies or customers, and at the same time, introduce an improved collaborative filtering algorithm to recommend potential partners or customers to improve matching accuracy and satisfaction;

[0093] Step S6: Return the matching results to the company for reference in decision-making. At the same time, collect feedback data, including prediction accuracy and procurement efficiency. According to the feedback data, optimize the regression analysis prediction model, parts determination and potential customer matching process to form a closed-loop optimization system.

[0094] Example 2

[0095] See also Figure 3 In this embodiment, the demand analysis module includes a procurement optimization unit, and the procurement optimization unit is configured with an association rule mining strategy, which is used to discover the association relationship between parts and optimize the parts procurement plan according to the association relationship. The specific steps of the association rule mining strategy include:

[0096] A1: Obtain the target company dataset ,in, represents the nth target company data, and n represents the number of target company data;

[0097] A2: Set a minimum support threshold based on the company's business needs and data characteristics and the minimum confidence threshold ;

[0098] A3: Based on the target company dataset , using a counter to traverse , count the frequency of occurrence of each target company's data , sort out each target company’s data and its corresponding frequency of occurrence to generate a frequency table;

[0099] like , then discard;

[0100] like , then keep The corresponding target company data forms frequent item sets;

[0101] A4: Sort the frequent itemsets in descending order and construct an FP tree in the sorted order. At the same time, insert each target company data into the FP tree in the sorted order of the frequent itemsets and update the node count. In the FP tree, the tree node represents the target company data, and the node count represents the target company data in Frequency of occurrence ,The paths of the tree represent the co-occurrence relationships of the target company data;

[0102] A5: Generate all non-empty true subsets of frequent itemsets from the FP tree, and for each non-empty true subset Z, calculate the credibility between it and the complement B ,in, Representing Item Sets exist The number of times it appears in Represents the item set Z in The number of times it appears in It implies a logical relationship. represents a union;

[0103] Non-empty proper subsets refer to all subsets of a set, excluding the empty set and the set itself. For example, for the set {a, b, c}, its non-empty proper subsets include {a}, {b}, {c}, {a, b}, {a, c}, {b, c}; then each frequent item set is traversed to generate its non-empty proper subsets.

[0104] A6: Based on the minimum credibility threshold ,like , then the association rules corresponding to the frequent item set are retained to form the association rule set R;

[0105] A7: Output the filtered association rule set R to form association rules between parts;

[0106] A8: Interpret the association rules between the generated parts, identify the association relationship between the parts, analyze the purchase pattern based on the parts purchase records, and present the analysis results in the form of an association rule network diagram. At the same time, formulate targeted inventory management strategies based on the analysis results;

[0107] Among them, the association relationship between parts includes complementary relationship and substitution relationship; the purchase pattern includes which parts are often purchased together and which parts have a purchase order or time association. At the same time, when presented in the form of an association rule network diagram, the antecedent and consequent of the association rule are used as nodes, and the association relationship is used as an edge to display in the diagram.

[0108] like , then consider bulk purchases or combined orders, where represents the support of the association rule, and , Represents association rules The support of represents the number of items containing association rule A and association rule D, Indicates the total number of items;

[0109] like , then optimize the inventory management strategy, where Indicates the credibility of the association rule;

[0110] In summary, inventory management strategies include considering bundling inventory or combination management for parts that are often purchased together; for parts that have a substitution relationship, their cost-effectiveness can be evaluated and more suitable parts can be selected for inventory control; and inventory levels and replenishment strategies can be adjusted according to seasonal or cyclical changes in purchasing patterns.

[0111] A9: Implement the optimized procurement plan and monitor the execution of the procurement plan.

[0112] Example 3

[0113] See also Figure 4 In this embodiment, the matching and recommendation module includes a model building unit and a recommendation unit. The model building unit is used to build a potential customer model using customer portraits and market segmentation methods. The specific steps of building the potential customer model include:

[0114] B1: Obtain target company dataset , and extract features from the target company data set, and build customer portraits based on the extracted features;

[0115] B2: Based on the customer portrait data, a clustering algorithm is used to segment the market into different customer groups, and potential customer characteristics are identified from the segmented market groups. At the same time, data corresponding to the potential customer characteristics are extracted from the customer portrait data to generate customer characteristic data. The clustering algorithm is a prior art in this field and is not an inventive solution of the present application, and is not described in detail here.

[0116] B3: Load a pre-built decision tree model, and use customer feature data to train the pre-built decision tree model so that it learns the features and patterns of potential customers, obtain a trained decision tree model, and save the trained decision tree model as a potential customer model, wherein the decision tree model is a prior art content in this field, is not an inventive solution of the present application, and is not described in detail here;

[0117] B4: Input real-time target company data into the potential customer model and calculate the target company data based on the preset threshold. , convert the identification results of the potential customer model into a potential customer list, classify the potential customers, prioritize the potential customers based on the company's potential needs, and output the classified and prioritized potential customer list to the report;

[0118] Furthermore, the specific steps of B4 include:

[0119] (1) Obtain real-time target company data and potential customer models;

[0120] (2) Set thresholds based on business needs , used to convert the recognition results of the potential customer model into a potential customer list;

[0121] (3) Input the real-time target company data into the potential customer model to obtain the recognition result of the potential customer model;

[0122] (4) Based on the preset threshold, the identification results are converted into a list of potential customers.

[0123] (5) Use decision tree classification algorithm to classify potential customers, such as hot customers, warm customers, and cold customers. Hot customers are those who may place an order within one month, warm customers are those who may place an order within three months, and cold customers are those who may place an order after three months.

[0124] (6) Prioritize the classified potential customers based on the company's potential needs, using factors such as the customer's purchasing power and potential demand intensity to prioritize;

[0125] (7) Use the CRM system to output the classified and prioritized list of potential customers into a report. The report should include the customer's basic information, classification information, priority ranking, and possible purchase intention, etc. The CRM system is the prior art in this field and is not an inventive solution of the present application, and is not elaborated here.

[0126] B5: Monitor and evaluate the potential customer model regularly, and adjust the potential customer model based on the evaluation results.

[0127] The recommendation unit is configured with an improved collaborative filtering algorithm, which is used to recommend potential partners or customers. The specific steps of the improved collaborative filtering algorithm include:

[0128] C1: Obtain target company dataset ,according to , build a company-customer matrix , where rows represent companies, columns represent customers, and matrix elements represent the company's behavior records toward customers;

[0129] C2: Using an improved similarity calculation method, using the company-customer matrix Calculate the similarity between companies and store the similarity between companies in a company similarity matrix Among them, The rows and columns in represent companies, and the matrix elements represent the similarity between companies;

[0130] C3: For the target company, sort the companies from high to low according to the company similarity matrix, and select the K companies that are most similar to it as neighboring companies according to the company similarity sorting results;

[0131] C4: Analyze the transaction pattern characteristics of neighboring companies based on their behavior records, and generate a recommendation list for the target company based on the target company's business needs and customer characteristics;

[0132] C5: Use accuracy evaluation indicators to evaluate the recommendation results, and optimize the recommendation process based on the evaluation results. The accuracy evaluation indicators are prior art content in this field and are not the creative solution of this application, so they will not be elaborated here.

[0133] The specific formula of the improved similarity calculation method in C2 is:

[0134] ;

[0135] in, express The similarity with V, and V represent two companies of the same type in the company-customer matrix The behavior records in , m represents the number of companies of the same type, represents the weight of the i-th dimension, express The i-th element of represents the i-th element of V, express The offset in the i-th dimension, represents the offset of V in the i-th dimension, represents the nonlinear term coefficient, and express and The nonlinear transformation function is and Represents a positive number.

[0136] It should be noted that the formula in the present invention introduces the weighting factor , so that when calculating the similarity, it is possible to perform differentiated processing according to the importance of different features, which enhances the flexibility and adaptability of the formula; secondly, the formula and represents the adjustment or offset of the feature value, which helps to more accurately measure the similarity between two objects in a specific context; in addition, by adding the regularization term , which can capture the nonlinear relationship between features and further improve the accuracy and comprehensiveness of similarity calculation; finally, and As a small constant term, it effectively avoids the situation where the denominator is zero, and enhances the stability and robustness of the formula. In general, the formula in the present invention is improved in terms of flexibility, accuracy, comprehensiveness and stability.

[0137] The specific steps of C4 include:

[0138] C4.1: Collect the behavior records of neighboring companies, as well as the business needs and customer characteristics of the target company, and pre-process the collected data. The behavior records include transaction data and product usage records.

[0139] C4.2: Based on the behavior records of neighboring companies, extract the material characteristic information required for neighboring company research, including transaction frequency, transaction amount, and product category distribution;

[0140] C4.3: Based on the extracted material feature information, a clustering algorithm is used to divide neighboring companies into different pattern groups, wherein the clustering algorithm is a prior art content in this field and is not an inventive solution of this application, and will not be described in detail here;

[0141] C4.4: Use the similarity calculation formula to match the target company's customer characteristics with the pattern groups of neighboring companies, and find the neighboring company group that is most similar to the target company's customers;

[0142] C4.5: Based on the matching results, select companies whose transaction model characteristics meet the business needs of the target company from the group of neighboring companies that are similar to the target company's customers, and use the products of the companies that meet the needs as recommendations to generate a recommendation list;

[0143] C4.6: Optimize the recommendation list based on feedback from target companies.

[0144] The formula for generating the recommendation list in C4.5 is:

[0145] ;

[0146] in, represents the recommendation degree of target company a for product r, represents the average rating of all products by target company a, represents the set of neighboring companies similar to the target company a, represents the similarity between target company a and neighbor company j, represents the actual rating of product r by neighbor company j, represents the average rating of all products by neighbor company j, express The absolute value of .

[0147] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in the field may also change, modify, replace and modify the above-mentioned embodiments without departing from the purpose and scope of protection of the present invention, and all of these are within the protection of the present invention.

[0148] If the disclosed technical solution involves personal information, the product using the disclosed technical solution has clearly informed the personal information processing rules and obtained the individual's voluntary consent before processing the personal information. If the disclosed technical solution involves sensitive personal information, the product using the disclosed technical solution has obtained the individual's separate consent before processing the sensitive personal information, and at the same time meets the "explicit consent" requirement. For example, on personal information collection devices such as cameras, clear and prominent signs are set to inform that the personal information collection scope has been entered and personal information will be collected. If the individual voluntarily enters the collection scope, it is deemed that he or she agrees to the collection of his or her personal information; or on the device that processes personal information, when the personal information processing rules are notified by obvious signs / information, the individual's authorization is obtained through pop-up information or by asking the individual to upload his or her personal information; among them, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the type of personal information processed.

Claims

1. Intelligent industrial cluster management system based on big data analysis, characterized by: include: Data processing module, model prediction module, demand analysis module, matching and recommendation module, optimization module; The data processing module is used to collect research directions, project data, and material usage of companies of the same type from the supply chain, and perform preprocessing to generate a target company data set; The model prediction module is used to load a pre-built regression analysis prediction model and use the target company data set for training to obtain a trained regression analysis prediction model; The demand analysis module is used to analyze the materials required for the company's subsequent research based on the forecast results, determine the required parts types and quantities, and optimize the parts procurement plan through association rule mining strategies; The matching and recommendation module is used to build a potential customer model using customer portraits and market segmentation methods, match the required parts with the potential customer model, find potential companies or customers, and recommend potential partners or customers through an improved collaborative filtering algorithm; The optimization module is used to collect feedback data and optimize the regression analysis prediction model, part determination and potential customer matching process based on the feedback data to form a closed-loop optimization system.

2. The intelligent industrial cluster management system based on big data analysis as claimed in claim 1, characterized in that: The demand analysis module includes a procurement optimization unit, and the procurement optimization unit is configured with an association rule mining strategy, which is used to discover the association relationship between parts and optimize the parts procurement plan according to the association relationship. The specific steps of the association rule mining strategy include: A1: Obtain the target company dataset ,in, represents the nth target company data, and n represents the number of target company data; A2: Set a minimum support threshold based on the company's business needs and data characteristics and the minimum confidence threshold ; A3: Based on the target company dataset , using a counter to traverse , count the frequency of occurrence of each target company's data , sort out each target company’s data and its corresponding frequency of occurrence to generate a frequency table; like , then discard; like , then keep The corresponding target company data forms frequent item sets; A4: Sort the frequent itemsets in descending order and construct an FP tree in the sorted order. At the same time, insert each target company data into the FP tree in the sorted order of the frequent itemsets and update the node count. In the FP tree, the tree node represents the target company data, and the node count represents the target company data in Frequency of occurrence ,The paths of the tree represent the co-occurrence relationships of the target company data.

3. The intelligent industrial cluster management system based on big data analysis as claimed in claim 2, characterized in that: The specific steps of the association rule mining strategy also include: A5: Generate all non-empty true subsets of frequent itemsets from the FP tree, and for each non-empty true subset Z, calculate the credibility between it and the complement B ,in, Representing Item Sets exist The number of times it appears in Represents the item set Z in The number of times it appears in It implies a logical relationship. represents a union; A6: Based on the minimum credibility threshold ,like , then keep the association rules corresponding to the frequent item set to form the association rule set R; A7: Output the filtered association rule set R to form association rules between parts.

4. The intelligent industrial cluster management system based on big data analysis as claimed in claim 3, characterized in that: The specific steps of the association rule mining strategy also include: A8: Interpret the association rules between the generated parts, identify the association relationship between the parts, analyze the purchase pattern based on the parts purchase records, and present the analysis results in the form of an association rule network diagram. At the same time, formulate targeted inventory management strategies based on the analysis results; like , then consider bulk purchases or combined orders, where represents the support of the association rule, and , Represents association rules The support of represents the number of items containing association rule A and association rule D, Indicates the total number of items; like , then optimize the inventory management strategy, where Indicates the credibility of the association rule; A9: Implement the optimized procurement plan and monitor the execution of the procurement plan.

5. The intelligent industrial cluster management system based on big data analysis as claimed in claim 4, characterized in that: The matching and recommendation module includes a model building unit and a recommendation unit. The model building unit is used to build a potential customer model using customer portraits and market segmentation methods. The specific steps of building the potential customer model include: B1: Obtain target company dataset , and extract features from the target company data set, and build customer portraits based on the extracted features; B2: Based on the customer portrait data, use clustering algorithms to segment the market into different customer groups, and identify potential customer characteristics from the segmented market groups. At the same time, extract data corresponding to potential customer characteristics from the customer portrait data to generate customer characteristic data; B3: Load the pre-built decision tree model and use the customer feature data to train the pre-built decision tree model so that it can learn the features and patterns of potential customers, obtain the trained decision tree model, and save the trained decision tree model as a potential customer model; B4: Input real-time target company data into the potential customer model and calculate the target company data based on the preset threshold. , convert the identification results of the potential customer model into a potential customer list, classify the potential customers, prioritize the potential customers based on the company's potential needs, and output the classified and prioritized potential customer list to the report; B5: Monitor and evaluate the potential customer model regularly, and adjust the potential customer model based on the evaluation results.

6. The intelligent industrial cluster management system based on big data analysis as claimed in claim 5, characterized in that: The recommendation unit is configured with an improved collaborative filtering algorithm, which is used to recommend potential partners or customers. The specific steps of the improved collaborative filtering algorithm include: C1: Obtain target company dataset ,according to , build a company-customer matrix , where rows represent companies, columns represent customers, and matrix elements represent the company's behavior records toward customers; C2: Using an improved similarity calculation method, using the company-customer matrix Calculate the similarity between companies and store the similarity between companies in a company similarity matrix Among them, The rows and columns in represent companies, and the matrix elements represent the similarity between companies; C3: For the target company, sort the companies from high to low according to the company similarity matrix, and select the K companies that are most similar to it as neighboring companies according to the company similarity sorting results; C4: Analyze the transaction pattern characteristics of neighboring companies based on their behavior records, and generate a recommendation list for the target company based on the target company's business needs and customer characteristics; C5: Use accuracy evaluation indicators to evaluate the recommendation results and optimize the recommendation process based on the evaluation results.

7. The intelligent industrial cluster management system based on big data analysis as claimed in claim 6, characterized in that: The specific formula of the improved similarity calculation method in C2 is: ； in, express The similarity with V, and V represent two companies of the same type in the company-customer matrix The behavior records in , m represents the number of companies of the same type, represents the weight of the i-th dimension, express The i-th element of represents the i-th element of V, express The offset in the i-th dimension, represents the offset of V in the i-th dimension, represents the nonlinear term coefficient, and express and The nonlinear transformation function is and Represents a positive number.

8. The intelligent industrial cluster management system based on big data analysis as claimed in claim 7, characterized in that: The specific steps of C4 include: C4.1: Collect behavioral records of neighboring companies, as well as the business needs and customer characteristics of the target company, and pre-process the collected data; C4.2: Extract the material characteristic information required for the neighbor company research based on the neighbor company's behavior records; C4.3: Based on the extracted material feature information, a clustering algorithm is used to classify neighboring companies into different pattern groups; C4.4: Use the similarity calculation formula to match the target company's customer characteristics with the pattern groups of neighboring companies, and find the neighboring company group that is most similar to the target company's customers; C4.5: Based on the matching results, select companies whose transaction model characteristics meet the business needs of the target company from the group of neighboring companies that are similar to the target company's customers, and use the products of the companies that meet the needs as recommendations to generate a recommendation list; C4.6: Optimize the recommendation list based on feedback from target companies.

9. The intelligent industrial cluster management system based on big data analysis as claimed in claim 8, characterized in that: The formula for generating the recommendation list in C4.5 is: ； in, represents the recommendation degree of target company a for product r, represents the average rating of all products by target company a, represents the set of neighboring companies similar to the target company a, represents the similarity between target company a and neighbor company j, represents the actual rating of product r by neighbor company j, represents the average rating of all products by neighbor company j, express The absolute value of .

Images