Model-based resource matching system for enterprise digital transformation

Abstract

The invention relates to the technical field of enterprise digital management, and provides a model-based resource matching system for enterprise digital transformation, which realizes efficient resource matching through cooperative work of a server side, a background control side, a demand side and a supply side. A server side collects multi-dimensional resource data through an acquisition unit, a feature matching unit screens out key features for preliminary matching, a model building unit builds a small model through a learning method and a model knowledge base, and a model optimization unit continuously optimizes the model through user feedback and real-time data. The resource report generation unit obtains a continuously updated matching small model by establishing a supply chain simulation model and carrying out use scene training, and finally generates a visual resource matching report, so that the limitation of the small model in processing a complex supply chain relationship is solved, the matching accuracy and efficiency are improved, and the resource matching efficiency is improved. And powerful support is provided for digital transformation of small and medium-sized enterprises.

Description

1.1 Technical Field The present invention relates to the technical field of enterprise digital management, and in particular to a model-based resource matching system for enterprise digital transformation. 1.2 Background Technology With the rapid development of information technology, enterprise supplier resource matching has become a key link in improving supply chain efficiency and optimizing resource allocation. At present, the integrated application of technologies such as big data, cloud computing and artificial intelligence has brought new opportunities for enterprise supplier management. In particular, small model technology has attracted widespread attention for its lightweight, high efficiency and flexibility. This technology can accurately identify the best match between suppliers and enterprise needs through deep learning of large amounts of data, providing strong support for enterprise procurement decisions.

[0003] In the matching of enterprise supplier resources, small model technology mainly builds an accurate prediction model by analyzing historical procurement data, supplier performance, market price dynamics and other multi-dimensional information. The prediction model can quickly screen out high-quality suppliers that meet the needs of the enterprise, and predict future market trends and supplier performance. Through automated algorithms, enterprises can adjust procurement strategies in real time, optimize inventory management, reduce procurement costs, and improve the responsiveness and flexibility of the supply chain. In addition, small model technology can also help enterprises identify potential risk points and enhance the robustness of the supply chain.

[0004] Although small model technology has shown significant advantages in matching enterprise supplier resources, there are also some technical pain points. First, the accuracy and completeness of data are crucial to the model effect, but in actual operations, data is often missing or erroneous. The reasons that affect the prediction accuracy of the model include that as the market environment changes, small model technology has limitations in dealing with complex and nonlinear supply chain relationships, and it is difficult to capture all influencing factors, resulting in inaccurate supplier resource matching systems. Nonlinear supply chain relationships refer to complex, non-directly proportional relationships between various elements in the supply chain. This relationship is not a simple cause-and-effect relationship or a linear addition relationship, but includes a variety of interactions and feedback mechanisms. 1.3 Summary of the invention The purpose of the present invention is to provide a model-based resource matching system for enterprise digital transformation, so as to solve the problem that small model technology has limitations in dealing with complex and nonlinear supply chain relationships, and it is difficult to capture all influencing factors, resulting in inaccurate supplier resource matching systems.

[0006] The present invention provides a model-based resource matching system for enterprise digital transformation, including a server end, a background control end, a demand side end and a supply side end, the server end establishes a communication connection with the background control end, the demand side end and the supply side end, the server end builds an enterprise digital supplier resource matching small model according to data transmitted by the background control end and the supply side end, the server end performs resource matching on the demand data of the demand side end through the enterprise digital supplier resource matching small model, and the server end outputs a visual resource matching report, including a server end, a background control end, the demand side end and the supply side end, the server end establishes a communication connection with the background control end, the demand side end and the supply side end, the server end builds an enterprise digital supplier resource matching small model according to data transmitted by the background control end and the supply side end, the server end performs resource matching on the demand data of the demand side end through the enterprise digital supplier resource matching small model, and the server end outputs a visual resource matching report, and the server end includes: An acquisition unit acquires resource data, which includes enterprise information, demand data, market dynamics, industry trends, and historical transaction data; The feature matching unit receives the demand data, analyzes the demand data using a feature selection algorithm, selects the features that have the greatest impact on the matching results, matches the features that have the greatest impact on the matching results in the resource matching database, and obtains the supplier resource matching results; The model building unit establishes a learning method and model knowledge base, and sequentially learns the data generated in the supplier resource matching process and the resource data obtained through the learning method and model in the learning method and model knowledge base to obtain a learning method and model data set, and builds a small model of enterprise digital supplier resource matching based on the learning method and model data set; The model optimization unit collects user feedback information, uses the user feedback information and real-time resource data as new training data, and uses the stochastic gradient descent algorithm to continuously update and optimize the enterprise digital supplier resource matching small model. The enterprise digital supplier resource matching small model uses the reinforcement learning algorithm to learn matching decisions in the interaction with the environment; The resource report generation unit establishes a supply chain simulation model, simulates the usage scenarios in the supply chain, performs usage scenario training on the enterprise digital supplier resource matching small model, obtains usage scenario training data, uses the usage scenario training data as new training data, and continuously trains the enterprise digital supplier resource matching small model using the new training data to obtain a continuously updated enterprise digital supplier resource matching small model. The continuously updated enterprise digital supplier resource matching small model is used to perform data processing on user needs to obtain a visual resource matching report.

[0007] Furthermore, the model-based resource matching system for enterprise digital transformation described in the present invention includes an acquisition unit, comprising: Establish relational database and non-relational database, and store the acquired resource data in the corresponding database according to the data type; Before data storage, the sensitive data in the multi-source data is encrypted using an encryption algorithm. After the data transmission is completed, the server performs integrity verification on the received data. The server records the database operation log, including data addition, deletion, modification and query operations, and regularly conducts security audits and monitoring of the database to check for abnormal access or data tampering.

[0008] Furthermore, the model-based resource matching system for enterprise digital transformation described in the present invention includes a feature matching unit, including: According to the characteristics of the required data, select the applicable feature selection algorithm, input the extracted features into the selected feature selection algorithm, and analyze the importance of each feature to the matching result through the feature selection algorithm; According to the output results of the feature selection algorithm, filter out the feature set that affects the matching results; Using each feature to match the result and the features that affect the result, a matching model is constructed to search for qualified supplier resources in the resource matching database; Furthermore, the model-based resource matching system for enterprise digital transformation described in the present invention, the feature matching unit, also includes: Taking the characteristic values ​​in the demand data as input parameters, searching and comparing them in the resource matching database through the matching model; According to the output results of the matching model, select the supplier resources that meet the requirements. The effectiveness of the supplier resources that meet the requirements are tested, and the valid supplier resources that meet the requirements are obtained after the test, and the results are returned to the demand side.

[0009] Furthermore, the model-based resource matching system for enterprise digital transformation described in the present invention includes a model building unit, including: The server collects learning methods and models suitable for supplier resource matching, including decision trees, random forests, support vector machines, and deep learning models, and builds a knowledge base of learning methods and models; The learning method and model knowledge base is used to store and manage the knowledge base architecture of learning methods and models, and deploy the built learning method and model knowledge base on the server; Integrate and preprocess the data generated during the supplier resource matching process and the resource data obtained previously to form a data set for model training. Use the data set for model training to train the learning methods and models in the learning method and model knowledge base in turn. During the training process, record the performance of each method and model. The performance of each method and model includes indicators such as accuracy and recall rate, which are used to evaluate the model effect. Furthermore, the model-based resource matching system for enterprise digital transformation described in the present invention, the model building unit, also includes: Receive enterprise-specific business requirements, including the model algorithms that need to be used; Deploy the enterprise-specific business needs, including the model algorithms that need to be used, to the server and conduct tests. Continuously monitor the performance of the enterprise-specific business needs, including the model algorithms that need to be used, during the testing process. The performance of each model involves periodic updates of the learning method and the model in the model knowledge base.

[0010] Furthermore, the model-based resource matching system for enterprise digital transformation described in the present invention includes a model optimization unit, including: Set up a user feedback system on the server to obtain users' evaluation and feedback on supplier resource allocation results through online surveys, rating systems or user feedback forms; Collect and record user feedback on the satisfaction, accuracy, timeliness, etc. of the matching results in real time, organize and analyze the collected user feedback information, and extract information points, including the most satisfactory or unsatisfactory matching results, as well as improvement suggestions; Integrate user feedback information and real-time resource data to form a new training data set. Use the stochastic gradient descent algorithm in combination with the new training data set to update and optimize the parameters of the enterprise's digital supplier resource matching model.

[0011] Furthermore, the model-based resource matching system for enterprise digital transformation described in the present invention, the model optimization unit, also includes Define a reinforcement learning environment for supplier resource matching. The reinforcement learning environment includes state space, action space, and reward function. Use the reinforcement learning algorithm to train the small model of enterprise digital supplier resource matching. The reinforcement learning algorithm will enable the small model of enterprise digital supplier resource matching to adjust the matching strategy according to the rewards obtained after obtaining different matching decisions. The enterprise digital supplier resource matching model interacts with data in the reinforcement learning environment.

[0012] Furthermore, the model-based resource matching system for enterprise digital transformation described in the present invention includes a resource report generating unit, comprising: Based on the operation process and rules of the real-time supply chain, a supply chain simulation model is built on the server. The supply chain simulation model is used to simulate the usage scenarios in the supply chain, including procurement, production, logistics, and sales. Set different usage scenario parameters in the simulation model, including market demand changes, supplier capacity fluctuations, and logistics delays; By running the simulation model, the enterprise digital supplier resource matching small model is trained under different usage scenarios, and the matching results and performance data under each scenario are recorded to form a usage scenario training data set.

[0013] Furthermore, the model-based resource matching system for enterprise digital transformation described in the present invention, the resource report generating unit, also includes: Furthermore, the present invention provides a model-based resource matching system for enterprise digital transformation, and a resource report generating unit, further comprising: Determine the goal of the simulation model and analyze the actual needs of the enterprise's digital supplier resource matching, which includes supplier information, product requirements, and procurement history; Select Simulink simulation tool to build a simulation model of supplier resource matching according to the actual operation of the enterprise, including supplier library, product demand library, and matching algorithm module.

[0014] Set up different usage scenarios in the simulation model, including changes in the number of suppliers, changes in product demand, and changes in the market environment; The supplier data model includes the basic information of the supplier, including name, address, contact information, product list, price, quality, and delivery time.

[0015] The beneficial effects of the present invention are as follows: Improve matching accuracy: Through the comprehensive application of small model technology and multiple learning algorithms, combined with continuous training of reinforcement learning algorithms, it is possible to accurately identify and match supplier resources that best meet enterprise needs.

[0016] Optimize resource allocation: The system analyzes market dynamics, industry trends and historical transaction data in real time, automatically screens and optimizes supplier resources, achieves efficient resource allocation, reduces procurement costs and improves supply chain efficiency.

[0017] Enhance system adaptability: Through continuous updating and learning mechanisms, the system can respond to changes in the market environment and complex supply chain relationships and adjust matching strategies in a timely manner.

[0018] Improve user experience: The introduction of the user feedback system enables user satisfaction, accuracy and timeliness feedback information to be collected in a timely manner and used for model optimization, thereby improving user experience.

[0019] Generate visual resource matching reports: The system can generate clear resource matching reports to facilitate enterprise decision-making analysis.

[0020] Promoting digital transformation of SMEs: The present invention provides a lightweight, efficient and flexible digital transformation solution for SMEs, which helps to improve their competitiveness. 1.4 Description of the Figures In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0022] Figure 1 A schematic diagram of a system architecture provided by an embodiment of the present invention.

[0023] Figure 2 A schematic diagram of a server-side device provided in an embodiment of the present invention. 1.5 Specific implementation method In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described in conjunction with the specific embodiments of the present invention and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. The technical solutions provided by the embodiments of the present invention are described in detail below in conjunction with the drawings.

[0025] See also Figure 1 to Figure 2 ,The present invention provides a model-based resource matching system for enterprise digital transformation, including; The present invention provides a model-based resource matching system for enterprise digital transformation, including a server end, a background control end, a demand side end and a supply side end, the server end establishes a communication connection with the background control end, the demand side end and the supply side end, the server end builds an enterprise digital supplier resource matching small model according to the data transmitted by the background control end and the supply side end, the server end performs resource matching on the demand side demand data through the enterprise digital supplier resource matching small model, and the server end outputs a visual resource matching report, and the server end includes: An acquisition unit acquires resource data, which includes enterprise information, demand data, market dynamics, industry trends, and historical transaction data; API interface docking establishes cooperation with various data providers, obtains demand data, historical transaction data, etc. in real time through API interface, and connects with various data through API interface to ensure the accuracy and timeliness of data.

[0026] Clean the collected raw data, remove duplicate, invalid or erroneous data, and ensure the quality and accuracy of the data. Integrate data from different data sources to form a unified and complete data set to facilitate subsequent data analysis and mining. Use data mining algorithms, such as association analysis and cluster analysis, to extract valuable information and patterns from massive data to provide support for corporate decision-making. Visualize data through charts, dashboards, etc., intuitively display corporate information, market dynamics, industry trends, etc., to help decision makers better understand data and make decisions.

[0027] The feature matching unit receives the demand data, analyzes the demand data using a feature selection algorithm, selects the features that have the greatest impact on the matching results, matches the features that have the greatest impact on the matching results in the resource matching database, and obtains the supplier resource matching results; The feature matching unit first uses the feature selection algorithm to analyze the received demand data, evaluate the importance of each feature to the matching result, and then select the features that have the greatest impact on the matching result. Common feature selection algorithms include: Filtering method: Select features based on statistical tests, such as using correlation coefficient, mutual information or chi-square test to evaluate the correlation between features and target variables. By setting a threshold, features that are highly correlated with the target variable can be screened out.

[0028] Wrapping method: The wrapping method selects features by evaluating model performance. It usually uses a machine learning algorithm to search for the optimal feature subset in the feature space. This method can take into account the interaction between features, but the computational cost is relatively high.

[0029] Embedding: The embedding method combines the feature selection process with the model training process. For example, when training models such as decision trees or random forests, features can be selected based on their importance scores. This method can naturally complete feature selection during the model training process.

[0030] Database matching technology, after screening out key features, the feature matching unit will use these features to search and match in the resource matching database. This involves the following technologies: Indexing technology: In order to improve matching efficiency, databases usually create indexes for key features. In this way, when performing matching queries, the database can quickly locate data records containing target features, thereby speeding up the matching process.

[0031] SQL query optimization: The feature matching unit needs to construct efficient SQL query statements to retrieve matching supplier resources from the database. By optimizing the structure of the query statement and using appropriate query conditions, the query speed and accuracy can be improved.

[0032] Fuzzy matching technology: When demand data contains a certain degree of fuzziness or uncertainty, fuzzy matching technology can be used to find supplier resources that are similar or close to the demand data. For example, an algorithm based on text similarity can be used to calculate the similarity between demand data and supplier resources, and return results with a higher similarity.

[0033] The model building unit establishes a learning method and model knowledge base, and sequentially learns the data generated in the supplier resource matching process and the resource data obtained through the learning method and model in the learning method and model knowledge base to obtain a learning method and model data set, and builds a small model of enterprise digital supplier resource matching based on the learning method and model data set; The model building unit uses machine learning and deep learning technologies to learn potential patterns and rules from large amounts of data, thereby enabling prediction and decision-making of unknown data. In the supplier resource matching scenario, machine learning and deep learning models can learn complex relationships in multi-dimensional data such as historical transaction data, corporate information, and market dynamics, and then predict the degree of matching between new suppliers and demanders.

[0034] Establish a knowledge base of learning methods and models, which contains a variety of learning algorithms and pre-trained models suitable for supplier resource matching scenarios. The algorithms and models may include but are not limited to linear regression, support vector machines, neural networks, decision trees, etc. By applying these algorithms and models to the data generated and the resource data obtained during the supplier resource matching process, a more accurate and efficient matching model can be trained.

[0035] During the model training process, the model building unit uses cross-validation, grid search and other technologies to tune the model to find the best model parameters and structure, which can ensure that the model has a good fit on the training set and has good generalization ability on the test set. In addition, in order to prevent the model from overfitting or underfitting, regularization, ensemble learning and other methods are used to optimize the performance of the model.

[0036] After obtaining multiple learning methods and model data sets through learning methods and model knowledge bases, the model building unit uses model integration techniques, such as voting, stacking or mixing, to combine the prediction results of these individual models to improve the accuracy and stability of the overall prediction. Finally, the trained and optimized matching small model will be deployed to the system for use in subsequent feature matching and resource report generation.

[0037] The model optimization unit collects user feedback information, uses the user feedback information and real-time resource data as new training data, and uses the stochastic gradient descent algorithm to continuously update and optimize the enterprise digital supplier resource matching small model. The enterprise digital supplier resource matching small model uses the reinforcement learning algorithm to learn matching decisions in the interaction with the environment; In order to continuously optimize the matching model, we first need to collect user feedback information, which can be achieved by designing user feedback interfaces and questionnaires, or using technologies that track user behavior (such as clickstream data tracking). The collected user feedback information may include satisfaction scores, accuracy evaluations of matching results, and user opinions and suggestions on the matching process, which provide a reference for model optimization.

[0038] The collected user feedback information and real-time resource data need to be effectively integrated and processed so that they can be used as new training data to update the optimization model. Data fusion technology involves integrating data from different sources and in different formats to ensure data consistency and availability. Data processing technology includes steps such as data cleaning, conversion, and standardization to eliminate outliers, missing values, and duplicate values, making the data more suitable for model training.

[0039] The stochastic gradient descent algorithm is an optimization algorithm used to adjust the parameters of the model during the training process to minimize the loss function. Compared with the traditional batch gradient descent algorithm, SGD uses only one sample or a small batch of samples to update the model parameters in each iteration, which makes the training process more efficient and can handle large-scale data sets. In the optimization of the small model for supplier resource matching, the SGD algorithm can help the model converge to the optimal solution faster and improve the accuracy of matching.

[0040] Reinforcement learning is a machine learning method that learns the optimal decision-making strategy by interacting with the environment. In the supplier resource matching scenario, the reinforcement learning algorithm enables the matching model to continuously learn and improve matching decisions in the interaction with the environment. Specifically, the model will try different matching strategies and adjust its decision-making method based on the reward signal fed back by the environment. In this way, the model can gradually learn the optimal matching strategy to adapt to the ever-changing market demand and supplier resource conditions.

[0041] The resource report generation unit establishes a supply chain simulation model, simulates the usage scenarios in the supply chain, performs usage scenario training on the enterprise digital supplier resource matching small model, obtains usage scenario training data, uses the usage scenario training data as new training data, and continuously trains the enterprise digital supplier resource matching small model using the new training data to obtain a continuously updated enterprise digital supplier resource matching small model. The continuously updated enterprise digital supplier resource matching small model is used to perform data processing on user needs to obtain a visual resource matching report.

[0042] In order to simulate the real supply chain environment, the resource report generation unit first establishes a supply chain simulation model. The supply chain simulation model can simulate various usage scenarios in the supply chain, such as demand changes, supplier responses, logistics distribution, etc., thereby providing the enterprise's digital supplier resource matching model with close-to-actual usage scenario training data. Through the simulation model, the matching model can be tested and optimized in a relatively safe and controllable environment.

[0043] Use the usage scenario training data generated by the supply chain simulation model to train the matching small model. In this way, new and diverse training data can be continuously provided to the model to help the model better learn and adapt to various complex supply chain environments. The training method can not only improve the generalization ability of the model, but also make the model more robust and accurate when facing new scenarios.

[0044] The enterprise digital supplier resource matching small model is continuously trained using new training data, thereby achieving dynamic updating of the model. This continuous updating technology ensures that the model can keep up with market changes and user needs, and always maintain its advanced nature and accuracy. Through continuous learning and optimization, the model can better adapt to new data and environments, thereby providing more accurate matching results.

[0045] The resource report generation unit also uses advanced data processing technology to analyze user needs to ensure the quality and accuracy of the data. At the same time, with the help of visualization technology such as charts, images, animations, etc., complex matching results are presented in an intuitive and easy-to-understand way. This not only improves the readability and comprehension of the report, but also helps users grasp the matching results and supply chain status more quickly.

[0046] Specifically, the present invention provides a model-based resource matching system for enterprise digital transformation, an acquisition unit, including: Establish relational database and non-relational database, and store the acquired resource data in the corresponding database according to the data type; Before data storage, the sensitive data in the multi-source data is encrypted using an encryption algorithm. After the data transmission is completed, the server performs integrity verification on the received data. The server records the database operation log, including data addition, deletion, modification and query operations, and regularly conducts security audits and monitoring of the database to check for abnormal access or data tampering.

[0047] Specifically, the present invention provides a model-based resource matching system for enterprise digital transformation, a feature matching unit, including: According to the characteristics of the required data, select the applicable feature selection algorithm, input the extracted features into the selected feature selection algorithm, and analyze the importance of each feature to the matching result through the feature selection algorithm; According to the output results of the feature selection algorithm, the feature set that affects the matching results is screened out, and one or more matching models are constructed by using the importance of each feature to the matching results and the feature set that affects the matching results, so as to search for qualified supplier resources in the resource matching database; The key feature values ​​in the demand data are used as input parameters, and searched and compared in the resource matching database through the matching model; Based on the output results of the matching model, the supplier resources that meet the requirements are screened out and the results are returned to the demand side.

[0048] Specifically, the present invention provides a model-based resource matching system for enterprise digital transformation, and a model building unit, including: The server collects and organizes various learning methods and models suitable for supplier resource matching, including decision trees, random forests, support vector machines, and deep learning models; The learning method and model knowledge base is used to store and manage the knowledge base architecture of these learning methods and models. The built learning method and model knowledge base is deployed on the server. The learning method and model knowledge base includes the functions of adding, deleting, modifying and checking test data, and the calling and updating mechanism of the model. Integrate and preprocess the data generated during the supplier resource matching process and the resource data obtained previously to form a data set suitable for model training. Use the prepared data set to train the learning methods and models in the learning method and model knowledge base in turn. During the training process, record the performance of each model. The performance of each model includes indicators such as accuracy and recall rate, which are used to evaluate the model effect. Based on the model evaluation results, optimize and adjust the models with poor performance, and select one or more models with the best performance from many models as the basis for subsequent construction of small models; Integrate the best selected models, combine them with the enterprise's specific business needs and rules, and output a small digital supplier resource matching model suitable for the enterprise; The constructed small model is deployed to the server and tested in the actual environment. During the operation of the small model, its performance is continuously monitored, and necessary optimization and adjustment are performed according to the actual situation. The performance of each model includes regular updates of the learning method and the model in the model knowledge base, and adjustment of the parameters and algorithms of the small model to maintain its matching accuracy and efficiency.

[0049] Specifically, the present invention provides a model-based resource matching system for enterprise digital transformation, a model optimization unit, including: Set up a user feedback system on the server to obtain user evaluation and feedback on supplier resource allocation results through online surveys, rating systems or direct user feedback forms; Collect and record user feedback on the satisfaction, accuracy, timeliness, etc. of the matching results in real time, organize and analyze the collected user feedback information, and extract information points, including the most satisfactory or unsatisfactory matching results, as well as improvement suggestions; Integrate user feedback information and real-time resource data to form a new training data set. Use the stochastic gradient descent algorithm in combination with the new training data set to update and optimize the parameters of the enterprise's digital supplier resource matching model.

[0050] Specifically, the present invention provides a model-based resource matching system for enterprise digital transformation, a model optimization unit, and further includes: Define a reinforcement learning environment for supplier resource matching. The reinforcement learning environment includes state space, action space, and reward function. Use the reinforcement learning algorithm to train the small model of enterprise digital supplier resource matching. The reinforcement learning algorithm will enable the small model of enterprise digital supplier resource matching to adjust the matching strategy according to the rewards obtained after obtaining different matching decisions. The enterprise digital supplier resource matching model continuously learns and improves its matching decision-making capabilities through continuous interaction with the reinforcement learning environment.

[0051] Specifically, the present invention provides a model-based resource matching system for enterprise digital transformation, a resource report generation unit, including: Based on the operation process and rules of the real-time supply chain, a supply chain simulation model is built on the server. The supply chain simulation model is used to simulate the usage scenarios in the supply chain, including procurement, production, logistics, and sales. Set different usage scenario parameters in the simulation model, including market demand changes, supplier capacity fluctuations, and logistics delays; By running the simulation model, the enterprise digital supplier resource matching small model is trained under different usage scenarios, and the matching results and performance data under each scenario are recorded to form a usage scenario training data set.

[0052] Specifically, the present invention provides a model-based resource matching system for enterprise digital transformation, a resource report generation unit, and further includes: Determine the goal of the simulation model and analyze the actual needs of the enterprise's digital supplier resource matching, which includes supplier information, product requirements, and procurement history; Select Simulink simulation tool to build a simulation model of supplier resource matching according to the actual operation of the enterprise, including supplier library, product demand library, and matching algorithm module.

[0053] Set up different usage scenarios in the simulation model, including changes in the number of suppliers, changes in product demand, and changes in the market environment; The supplier data model includes the basic information of the supplier, including name, address, contact information, product list, price, quality and delivery time.

[0054] The technical solution of the present invention aims at the limitation of small model technology in dealing with complex and nonlinear supply chain relationships, and adopts the following measures to improve the accuracy of the supplier resource matching system: Multi-dimensional data acquisition and analysis: Through the acquisition unit, multi-dimensional resource data such as enterprise information, demand data, market dynamics, industry trends and historical transaction data are collected to provide comprehensive data support for model construction.

[0055] Feature selection algorithm: The feature matching unit uses the feature selection algorithm to analyze the required data and select the features that have the greatest impact on the matching results. This step ensures that the model can focus on the most critical information points during the matching process.

[0056] Learning methods and model knowledge base: The model building unit has established a knowledge base containing a variety of learning methods and models (such as decision trees, random forests, support vector machines, and deep learning models). Through learning and evaluating a variety of models, the model with the best performance is selected as the basis for building small models. This multi-model integration method improves the system's ability to handle complex problems.

[0057] Continuous model optimization: The model optimization unit continuously updates and optimizes the enterprise digital supplier resource matching model by collecting user feedback information and combining it with real-time resource data. In addition, a reinforcement learning algorithm is introduced to enable the model to continuously learn and improve its matching decision-making capabilities in the interaction with the environment.

[0058] Supply chain simulation model: The resource report generation unit has established a supply chain simulation model to simulate the usage scenarios in the supply chain (such as procurement, production, logistics, and sales). Through simulation training, the model can adapt to different supply chain scenarios and improve its ability to deal with complex and nonlinear relationships.

[0059] Data security and integrity assurance: During the data acquisition and storage process, encryption algorithms are used to encrypt sensitive data, integrity checks are performed on received data, and database security audits and monitoring are performed regularly to ensure data accuracy and system robustness.

[0060] Visual resource matching report: The server outputs a visual resource matching report to make the matching results intuitive and easy to understand, which facilitates decision-making analysis for the demand side.

[0061] To sum up, the technical solution of the present invention effectively solves the limitations of small model technology in dealing with complex and nonlinear supply chain relationships through measures such as multidimensional data acquisition and analysis, feature selection algorithm, multi-model integration, continuous model optimization, supply chain simulation model, data security and integrity assurance, and visual report generation, thereby improving the accuracy and efficiency of the supplier resource matching system.

[0062] The present invention provides a model-based resource matching system for enterprise digital transformation, including a server side, a backend control side, a demand side, and a supply side. The server side establishes communication connections with the backend control side, the demand side, and the supply side.

[0063] Server-side functional unit The acquisition unit acquires resource data, including enterprise information, demand data, market dynamics, industry trends, and historical transaction data.

[0064] Establish relational databases and non-relational databases, and store the acquired resource data in the corresponding database according to the data type. Before data storage, the sensitive data in the multi-source data is encrypted using an encryption algorithm. After the data transmission is completed, the server performs integrity verification on the received data. The server records the operation log of the database, including the addition, deletion, modification and query operations of the data, and regularly conducts security audits and monitoring of the database to check whether there is abnormal access or data tampering.

[0065] The feature matching unit receives the demand data, analyzes the demand data using the feature selection algorithm, selects the features that have the greatest impact on the matching results, and matches them in the resource matching database to obtain the supplier resource matching results.

[0066] According to the characteristics of the demand data, select the applicable feature selection algorithm. Input the extracted features into the selected feature selection algorithm to analyze the importance of each feature to the matching results. According to the output results of the feature selection algorithm, build a matching model, and search and compare in the resource matching database. Return the selected supplier resources that meet the requirements to the demand side.

[0067] The model building unit establishes a learning method and model knowledge base, and uses the data generated in the supplier resource matching process and the resource data obtained to build a small model of enterprise digital supplier resource matching.

[0068] Collect and organize various learning methods and models suitable for supplier resource matching, including decision trees, random forests, support vector machines, and deep learning models. Deploy a knowledge base of learning methods and models, and include the functions of adding, deleting, modifying, and checking test data, as well as the calling and updating mechanism of models.

[0069] Integrate and preprocess the data to form a data set suitable for model training. Train through a variety of learning methods and models, record and evaluate the performance of each model, select the best performing model for integration, and output a small model suitable for digital supplier resource matching for the enterprise. Deploy the built small model to the server and test and optimize it in the actual environment.

[0070] The model optimization unit collects user feedback information and combines it with real-time resource data to continuously update and optimize the enterprise's digital supplier resource matching model.

[0071] Set up a user feedback system on the server and collect feedback information through online surveys, rating systems or direct user feedback forms. Collect and record user feedback information on the satisfaction, accuracy, timeliness and other aspects of the matching results in real time. Integrate user feedback information and real-time resource data to form a new training data set, and use the stochastic gradient descent algorithm to update and optimize the parameters of the small model. Introduce a reinforcement learning algorithm and define a reinforcement learning environment (including state space, action space, and reward function) so that the model can continuously learn and improve its matching decision-making ability in the interaction with the environment.

[0072] The resource report generation unit establishes a supply chain simulation model, simulates usage scenarios in the supply chain, trains the enterprise's digital supplier resource matching model, and generates a visual resource matching report.

[0073] Build a supply chain simulation model on the server to simulate usage scenarios in procurement, production, logistics, sales and other links.

[0074] Set different usage scenario parameters, such as market demand changes, supplier capacity fluctuations, logistics delays, etc. By running the simulation model, train the small model in different scenarios and record the matching results and performance data. Use the continuously updated enterprise digital supplier resource matching small model to process user demand data and obtain a visual resource matching report.

[0075] The above is a specific implementation of the present invention, which describes in detail the technical implementation steps and measures of each functional unit to solve the limitations of small model technology in dealing with complex supply chain relationships. Through multi-dimensional data acquisition, accurate feature matching, powerful model construction, continuous model optimization, and scenario training based on simulation models, the accuracy and efficiency of supplier resource matching are effectively improved, helping small and medium-sized enterprises achieve digital transformation.

[0076] Those skilled in the art can clearly understand that the technology in the embodiments of the present invention can be implemented by means of software plus a necessary general hardware platform. Based on this understanding, the technical solution in the embodiments of the present invention can be essentially or partly contributed to the prior art in the form of a software product, which can be stored in a storage medium such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present invention or certain parts of the embodiments.

[0077] The above-described embodiments of the present invention do not limit the protection scope of the present invention.

Claims

1. A model-based resource matching system for enterprise digital transformation, characterized in that: It includes the server side, the backend control side, the demand side and the supply side. The server side establishes communication connections with the backend control side, the demand side and the supply side. The server side builds a small model of enterprise digital supplier resource matching based on the data transmitted by the backend control side and the supply side. The server side matches the demand data of the demand side through the small model of enterprise digital supplier resource matching. The server side outputs a visual resource matching report. The server side includes: An acquisition unit acquires resource data, which includes enterprise information, demand data, market dynamics, industry trends, and historical transaction data; The feature matching unit receives the demand data, analyzes the demand data using a feature selection algorithm, selects the features that have the greatest impact on the matching results, matches the features that have the greatest impact on the matching results in the resource matching database, and obtains the supplier resource matching results; The model building unit establishes a learning method and model knowledge base, and sequentially learns the data generated in the supplier resource matching process and the resource data obtained through the learning method and model in the learning method and model knowledge base to obtain a learning method and model data set, and builds a small model of enterprise digital supplier resource matching based on the learning method and model data set; The model optimization unit collects user feedback information, uses the user feedback information and real-time resource data as new training data, and uses the stochastic gradient descent algorithm to continuously update and optimize the enterprise digital supplier resource matching small model. The enterprise digital supplier resource matching small model uses the reinforcement learning algorithm to learn matching decisions in the interaction with the environment; The resource report generation unit establishes a supply chain simulation model, simulates the usage scenarios in the supply chain, performs usage scenario training on the enterprise digital supplier resource matching small model, obtains usage scenario training data, uses the usage scenario training data as new training data, and continuously trains the enterprise digital supplier resource matching small model using the new training data to obtain a continuously updated enterprise digital supplier resource matching small model. The continuously updated enterprise digital supplier resource matching small model is used to perform data processing on user needs to obtain a visual resource matching report.

2. The model-based resource matching system for enterprise digital transformation according to claim 1, wherein the acquisition unit comprises: Establish relational database and non-relational database, and store the acquired resource data in the corresponding database according to the data type; Before data storage, the sensitive data in the multi-source data is encrypted using an encryption algorithm. After the data transmission is completed, the server performs integrity verification on the received data. The server records the database operation log, including data addition, deletion, modification and query operations, and regularly conducts security audits and monitoring of the database to check for abnormal access or data tampering.

3. The model-based resource matching system for enterprise digital transformation according to claim 1, wherein the feature matching unit comprises: According to the characteristics of the required data, select the applicable feature selection algorithm, input the extracted features into the selected feature selection algorithm, and analyze the importance of each feature to the matching result through the feature selection algorithm; According to the output results of the feature selection algorithm, filter out the feature set that affects the matching results; A matching model is constructed by utilizing the impact of each feature on the matching result and the features that affect the matching result, and is used to search for qualified supplier resources in the resource matching database.

4. The model-based resource matching system for enterprise digital transformation according to claim 3, wherein the feature matching unit further comprises: Taking the characteristic values ​​in the demand data as input parameters, searching and comparing them in the resource matching database through the matching model; According to the output results of the matching model, select the supplier resources that meet the requirements. The effectiveness of the supplier resources that meet the requirements are tested, and the effective supplier resources that meet the requirements are obtained after the test, and the results are returned to the demand side.

5. The model-based resource matching system for enterprise digital transformation according to claim 1, wherein the model building unit comprises: The server collects learning methods and models suitable for supplier resource matching, including decision trees, random forests, support vector machines, and deep learning models, and builds a knowledge base of learning methods and models; The learning method and model knowledge base is used to store and manage the knowledge base architecture of learning methods and models, and deploy the built learning method and model knowledge base on the server; The data generated during the supplier resource matching process and the resource data obtained previously are integrated and preprocessed to form a data set for model training. The data set for model training is used to train the learning methods and models in the learning method and model knowledge base in turn. During the training process, the performance of each method and model is recorded. The performance of each method and model includes indicators such as accuracy and recall rate, which are used to evaluate the model effect.

6. The model-based resource matching system for enterprise digital transformation according to claim 5, wherein the model building unit further comprises: Receive enterprise-specific business requirements, including the model algorithms that need to be used; Deploy the enterprise-specific business needs, including the model algorithms that need to be used, to the server and conduct tests. Continuously monitor the performance of the enterprise-specific business needs, including the model algorithms that need to be used, during the testing process. The performance of each model involves periodic updates of the learning method and the model in the model knowledge base.

7. The model-based resource matching system for enterprise digital transformation according to claim 1, wherein the model optimization unit comprises: Set up a user feedback system on the server to obtain users' evaluation and feedback on supplier resource allocation results through online surveys, rating systems or user feedback forms; Collect and record user feedback on the satisfaction, accuracy, timeliness, etc. of the matching results in real time, organize and analyze the collected user feedback information, and extract information points, including the most satisfactory or unsatisfactory matching results, as well as improvement suggestions; Integrate user feedback information and real-time resource data to form a new training data set. Use the stochastic gradient descent algorithm in combination with the new training data set to update and optimize the parameters of the enterprise's digital supplier resource matching model.

8. The model-based resource matching system for enterprise digital transformation according to claim 1, the model optimization unit further comprising: Define a reinforcement learning environment for supplier resource matching. The reinforcement learning environment includes state space, action space, and reward function. Use the reinforcement learning algorithm to train the enterprise digital supplier resource matching model. The reinforcement learning algorithm will enable the enterprise digital supplier resource matching model to adjust the matching strategy according to the rewards obtained after obtaining different matching decisions. The enterprise digital supplier resource matching model interacts with data in the reinforcement learning environment.

9. The model-based resource matching system for enterprise digital transformation according to claim 1, wherein the resource report generating unit comprises: Based on the operation process and rules of the real-time supply chain, a supply chain simulation model is built on the server. The supply chain simulation model is used to simulate the usage scenarios in the supply chain, including procurement, production, logistics, and sales. Set different usage scenario parameters in the simulation model, including market demand changes, supplier capacity fluctuations, and logistics delays; By running the simulation model, the enterprise digital supplier resource matching small model is trained under different usage scenarios, and the matching results and performance data under each scenario are recorded to form a usage scenario training data set.

10. The model-based resource matching system for enterprise digital transformation according to claim 9, the resource report generating unit further comprising: Determine the goal of the simulation model and analyze the actual needs of the enterprise's digital supplier resource matching, which includes supplier information, product requirements, and procurement history; Select Simulink simulation tool to build a simulation model of supplier resource matching based on the actual operation of the enterprise, including supplier library, product demand library, and matching algorithm module; Set up different usage scenarios in the simulation model, including changes in the number of suppliers, changes in product demand, and changes in the market environment; The supplier data model includes the basic information of the supplier, including name, address, contact information, product list, price, quality and delivery time.

Images