Enterprise supply chain side evaluation method and system based on big data

Abstract

The invention discloses an enterprise supply chain side evaluation method and system based on big data, and provides an innovative solution for the problems of low data processing efficiency, weak analysis and prediction ability, insufficient causal relationship analysis, decision support lag and the like in supply chain management. The method comprises the steps of integrating and processing real-time data, deeply analyzing the data by using an advanced analysis and prediction module in combination with a mathematical model and a machine learning algorithm, deeply understanding a causal relationship in a supply chain through a causal relationship analysis and optimization module, and finally accelerating a decision process through a visualization and decision support module. According to the system, through modular design and an advanced algorithm, the supply chain management efficiency and scientificity are remarkably improved, and comprehensive evaluation and optimization of the supply chain are achieved. The invention aims to solve the problem of improving the efficiency, accuracy and scientificity of enterprise supply chain management under the background of big data so as to adapt to rapidly changing market demands and a complex supply chain network.

Description

Technical Field

[0001] The present invention relates to the field of big data analysis and processing, and in particular to a method and system for evaluating the enterprise supply chain based on big data. The method and system focus on using big data technology to intelligently manage and optimize the enterprise supply chain, aiming to achieve high efficiency and scientificity of supply chain management through functions such as real-time data processing, advanced analysis and prediction, causal relationship analysis and optimization, and decision support. Background Art

[0002] In modern enterprise management, efficient management and optimization of the supply chain are crucial to the competitiveness and profitability of enterprises. However, traditional supply chain management methods often rely on manual experience and static analysis, which makes it difficult to cope with the rapid changes in the market and the complexity of the supply chain network. Especially in the era of big data, enterprises are faced with massive and diverse data. How to extract valuable information from it and conduct scientific analysis and prediction has become a key challenge for supply chain management. Traditional data processing methods, such as converting data into character types for storage and analysis, are inefficient, especially when processing large-scale data. The time consumption of data conversion and storage seriously affects the overall performance. At present, common problems in supply chain management include:

[0003] (1) Inefficient data integration and processing: In a big data environment, data conversion and storage time is too long, which reduces the overall efficiency of data processing.

[0004] (2) Insufficient analysis and forecasting capabilities: Lack of in-depth analysis and accurate forecasting of supply chain data makes it difficult to make scientific decisions.

[0005] (3) Weak causal relationship analysis: In supply chain management, there is a lack of in-depth analysis of the causal relationship between variables, resulting in insufficient basis for optimization decisions.

[0006] (4) Decision support is not timely enough: Traditional methods have a slow response speed in decision support and cannot adapt to market changes. Summary of the invention

[0007] The present invention provides an enterprise supply chain side evaluation method and system based on big data to solve the problem of how to effectively improve the efficiency, accuracy and scientificity of enterprise supply chain management under the background of big data to cope with the rapidly changing market demands and complex supply chain networks.

[0008] In order to solve the above problems, the present invention provides an enterprise supply chain evaluation method based on big data, comprising: A method for evaluating the enterprise supply chain based on big data, including: S100, obtain data from multiple sources and receive data streams in real time, perform data cleaning, conversion and real-time analysis, and store the processed data and distribute it to the application system; S200, extract features from data, train models, evaluate and predict, provide decision support based on real-time data, and dynamically update models; S300, build causal relationship models, perform causal reasoning, optimize supply chain decisions, and adjust models and decision plans based on feedback; S400, display data through charts and dashboards, generate decision suggestions, provide user interface and notifications, and support user interaction and feedback; S500, monitors performance, manages security, configures the environment, supports system expansion and integration, user management and behavior analysis, and ensures the safe, stable and efficient operation of the system.

[0009] As a preferred solution, the step S100 specifically includes: S110, obtain raw data from different data sources, including ERP system, logistics system, sales system, and supplier system; S120, receiving real-time data streams from sensors and IoT devices; S130. Manage and coordinate requests from different data interfaces through the API gateway to ensure the reliability and security of data collection; S140, the data stream receiver processes the preliminary analysis of the real-time data stream and temporarily stores the data in a buffer; S150, using the real-time analysis engine to process and analyze the data in real time, executing the analysis rules defined in the real-time rule engine, performing threshold monitoring and pattern recognition, and applying the machine learning model to perform real-time prediction and anomaly detection; S160, store the processed real-time data, provide efficient query and access support, store the real-time data in a distributed file system, and distribute the processed real-time data to various application systems and user interfaces.

[0010] As a preferred solution, step S200 specifically includes: S210, extract features from raw data using mathematical and physical indicators, select the most important features through statistical and machine learning methods, and divide the data into training, validation and test sets; S220. Use mathematical models for preliminary analysis and prediction, use physical models to simulate physical processes of logistics network flow, inventory management and production technology, and integrate mathematical models with physical models for comprehensive analysis and prediction; S230, use machine learning and deep learning techniques, including feature engineering and feature selection, deep convolutional neural networks to automatically extract complex features, and automatically tune model hyperparameters through Bayesian optimization and particle swarm optimization algorithms; S240. Provide decision support based on real-time data and dynamically update the model.

[0011] As a preferred solution, step S300 specifically includes: S310, Use structural equation model, Bayesian network and Granger causality analysis to build causal relationship model between variables in supply chain; S320. Apply causal reasoning algorithms to conduct counterfactual reasoning and intervention analysis, generate easy-to-understand analysis reports, and provide decision support for supply chain optimization; S330. Combine causal analysis results and optimization algorithms to simulate and evaluate the effects of different decision-making plans and generate optimization recommendations.

[0012] As a preferred solution, the step S400 specifically includes: S410, provides interactive dashboards and dynamic visualization tools to intuitively display key supply chain indicators and analysis results; S420, automatically generate regular reports, including charts and text descriptions, to facilitate users to share and record analysis results; S430, monitor key indicators in real time, detect abnormal situations and send alarms, and perform anomaly detection in combination with mathematical models.

[0013] As a preferred solution, the step S500 specifically includes: S510, monitor the usage of system resources in real time, analyze system performance by combining mathematical models and physical models, and identify performance bottlenecks and optimization points; S520, role-based and behavior-based access control model, combined with advanced encryption standards and multi-factor authentication technology, ensures the security and privacy protection of system data; S530, supports system expansion and integration, user management and behavior analysis, ensuring the safe, stable and efficient operation of the system.

[0014] As a preferred solution, it includes: Data acquisition and preprocessing module; Feature extraction and model training module; Cause-effect analysis and decision optimization module; Dynamic visualization and real-time decision support module; Performance monitoring and security assurance module.

[0015] As a preferred solution, it also includes: Data management module, used to support efficient storage and fast query of large-scale data; Security management module, used to ensure data security and privacy protection; The system performance monitoring module is used to monitor the usage of system resources in real time to ensure the efficient operation of the system.

[0016] The present invention has at least the following effects: (1) Data integration and real-time processing effects. Multi-source data acquisition and integration: The present invention realizes comprehensive collection of supply chain data by integrating data from ERP systems, logistics systems, sales systems, supplier systems, sensors, and IoT devices, ensuring data diversity and integrity; Real-time data processing: Through real-time data stream receivers, data buffers, and message queues, the system can efficiently manage and distribute real-time data streams, ensuring the reliability and efficiency of data transmission, and meeting the needs of real-time monitoring of the supply chain.

[0017] (2) Improvement of data quality and consistency. Efficient data cleaning: Real-time detection and processing of missing values, duplicate data, and outliers in data, using a variety of cleaning algorithms, significantly improves data quality and ensures the accuracy of subsequent analysis and decision-making; Unified data format and standardized processing: Convert real-time data into a unified format and perform standardized processing to ensure data consistency, facilitate subsequent processing and analysis, and improve the efficiency and accuracy of data processing.

[0018] (3) Model application and optimization effect. Comprehensive model application: Through linear regression, time series analysis, support vector regression, and high-order tensor regression models, the system can handle complex relationships and nonlinear data in the supply chain and provide accurate analysis and prediction results; Physical model integration: Using complex number analysis, inventory management models, and production process physical models, the system can simulate and optimize logistics network flow, inventory levels, and production efficiency, thereby improving the scientificity and practicality of supply chain management; Model fusion and optimization: Combining mathematical models and physical models, the system achieves comprehensive analysis and prediction, improving the accuracy and reliability of supply chain evaluation.

[0019] (4) Machine learning and deep learning effects. Automatic feature extraction and selection: Through feature engineering and deep convolutional neural networks, the system can automatically extract and select important features to improve the representativeness and predictive ability of the model; Hyperparameter optimization: Using Bayesian optimization and particle swarm optimization algorithms, the system can automatically tune the hyperparameters of the model to ensure the optimization of model performance and improve prediction accuracy.

[0020] (5) Causal relationship analysis and optimization of decision-making effects. Causal relationship modeling and reasoning: Using structural equation models, Bayesian networks and Granger causality analysis, the system can build a causal relationship model between variables in the supply chain, reveal the mutual influence between variables, and provide scientific causal reasoning and decision support; Optimize decision generation: Combining causal analysis results and optimization algorithms, the system can simulate and evaluate the effects of different decision-making plans, generate optimization suggestions, help enterprises optimize supply chains, and improve decision quality.

[0021] (6) Dynamic visualization and real-time decision support. Intuitive data display: Providing interactive dashboards and dynamic visualization tools, the system can intuitively display key indicators and analysis results of the supply chain, support users to view and analyze data in real time, and improve user experience; Real-time alerts and report generation: The system can monitor key indicators in real time, detect abnormal situations and send alerts, combine mathematical models for abnormality detection, and automatically generate regular reports to ensure timely communication and recording of information.

[0022] (7) Efficient data management and security protection. Efficient data storage and query: Through the distributed file system and real-time database, the system can support efficient storage and fast query of large-scale data, ensuring data availability and reliability; Data security and privacy protection: Based on the role and behavior access control model, combined with advanced encryption standards and multi-factor authentication technology, the system can ensure data security and privacy protection; System performance monitoring and optimization: Real-time monitoring of system resource usage, combined with mathematical models and physical models to analyze system performance, identify performance bottlenecks and optimization points, and ensure efficient operation of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a flow chart of a method and system for evaluating an enterprise supply chain based on big data provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0025] In the following, only some exemplary embodiments are briefly described. As those skilled in the art will appreciate, the described embodiments may be modified in various ways without departing from the spirit or scope of the present application. Therefore, the drawings and descriptions are considered to be exemplary and non-restrictive in nature.

[0026] Reference Figure 1 , is a flow chart of an enterprise supply chain side evaluation method and system based on big data according to an embodiment of the present invention. The enterprise supply chain side evaluation method and system based on big data may at least include steps S100-S500: S100, obtain data from multiple sources and receive data streams in real time, perform data cleaning, conversion and real-time analysis, store the processed data and distribute it to the application system.

[0027] S200, extract features from data, train models, evaluate and predict, provide decision support based on real-time data, and dynamically update models.

[0028] S300. Build causal relationship models, perform causal reasoning, optimize supply chain decisions, and adjust models and decision plans based on feedback.

[0029] S400, displays data through charts and dashboards, generates decision suggestions, provides user interface and notifications, and supports user interaction and feedback.

[0030] S500, monitors performance, manages security, configures the environment, supports system expansion and integration, user management and behavior analysis, and ensures the safe, stable and efficient operation of the system.

[0031] The enterprise supply chain evaluation method and system based on big data provided in the embodiments of the present invention are mainly used to solve the technical problem solved by the present invention, which is "how to effectively improve the efficiency, accuracy and scientificity of enterprise supply chain management under the background of big data to cope with rapidly changing market demands and complex supply chain networks". Specifically, it is the problem of "how to achieve comprehensive evaluation and optimization of enterprise supply chains based on big data technology, improve data processing efficiency, enhance analysis and prediction capabilities, deepen causal analysis, and provide timely decision support".

[0032] The present invention discloses a method and system for enterprise supply chain side evaluation based on big data, and proposes innovative solutions to the problems of low data processing efficiency, weak analysis and prediction capabilities, insufficient causal analysis, and lagging decision support in supply chain management. The method includes real-time data integration and processing, using advanced analysis and prediction modules combined with mathematical models and machine learning algorithms to deeply analyze data, deepening the understanding of causal relationships in the supply chain through causal analysis and optimization modules, and finally accelerating the decision-making process through visualization and decision support modules. The system significantly improves the efficiency and scientificity of supply chain management through modular design and advanced algorithms, and realizes comprehensive evaluation and optimization of the supply chain. The present invention aims to solve the problem of improving the efficiency, accuracy and scientificity of enterprise supply chain management under the background of big data, so as to adapt to rapidly changing market demands and complex supply chain networks.

[0033] S100, obtain data from multiple sources and receive data streams in real time, perform data cleaning, conversion and real-time analysis, store the processed data and distribute it to the application system. Step S100 includes steps S110-S160:

[0034] S110, obtain raw data from different data sources (such as ERP system, logistics system, sales system, supplier system, etc.); receive real-time data streams from sensors, IoT devices, etc.; manage and coordinate requests from different data interfaces through API gateways to ensure the reliability and security of data collection. Specifically:

[0035] (1) Initialization of multi-source data interfaces: When the system starts, the configuration and environment management submodule first loads all necessary configuration files. Multi-source data interfaces are initialized, including ERP system interface, logistics system interface, sales system interface, supplier system interface, etc. API gateway is initialized to ensure that each interface can communicate normally.

[0036] (2) Multi-source data acquisition: The data acquisition submodule obtains raw data from different data sources (such as ERP system, logistics system, sales system, supplier system, etc.) through multi-source data interfaces. The API gateway manages and coordinates requests from different data interfaces to ensure the reliability and security of data acquisition.

[0037] (3) Real-time data stream reception: The real-time data stream interface receives real-time data streams from sensors, IoT devices, etc. The data stream receiver handles the initial parsing of the real-time data stream and temporarily stores the data in a buffer to prevent data loss and processing delays.

[0038] (4) Data stream forwarding and caching: The API gateway manages and distributes real-time data streams through message queues (such as Apache Kafka) and passes data to downstream modules for processing. The message queue ensures efficient transmission and load balancing of data streams, preventing data accumulation or loss.

[0039] (5) Data collection log records: Each data collection operation will be recorded in the log system, including information such as data source, timestamp, data volume, etc. These logs are used for subsequent data audits and troubleshooting to ensure the transparency and traceability of the data collection process.

[0040] Core role: (1) Data diversity and integrity: The data acquisition submodule obtains data from multiple heterogeneous data sources, including ERP systems, logistics systems, sales systems, supplier systems, etc., to ensure data diversity and integrity. The integration of multi-source data provides a comprehensive view of the supply chain, which facilitates subsequent advanced analysis and decision support.

[0041] (2) Real-time data acquisition and processing: The real-time data stream interface ensures that the system can receive and process real-time data streams from sensors and IoT devices. Real-time data acquisition and processing are crucial for real-time monitoring and rapid response of the supply chain.

[0042] (3) Data transmission reliability: The API gateway manages and coordinates requests from different data interfaces through a unified interface to ensure the reliability and security of data collection. The message queue manages and distributes real-time data streams to ensure the efficiency and reliability of data transmission and prevent data loss.

[0043] (4) System scalability and flexibility: The design of the data acquisition submodule enables the system to flexibly integrate new data sources and interfaces to adapt to changing business needs. Through the combined use of API gateways and message queues, the system can be easily expanded to handle larger data volumes and more data sources.

[0044] (5) Data quality and security: During the data collection process, the data collection operation is recorded through the log system to ensure data quality and security. The recorded data collection log facilitates subsequent audits and troubleshooting, ensuring the transparency and traceability of the data collection process.

[0045] S120, receiving real-time data streams from various sources and temporarily storing them in a data buffer; using message queues to manage and distribute real-time data streams to ensure the reliability and efficiency of data transmission. Specifically:

[0046] (1) Initialization phase: When the system starts, the data receiving submodule is first initialized, including the configuration and startup of the real-time data stream receiver, data buffer, and message queue. The real-time data stream receiver is ready to receive real-time data streams from various sources (such as sensors, IoT devices, business systems, etc.).

[0047] (2) Real-time data stream reception: The real-time data stream receiver starts receiving real-time data streams from sensors, IoT devices, business systems, etc. The received data is first temporarily stored in a data buffer to prevent data loss and processing delay.

[0048] (3) Data buffering and processing preparation: The data buffer performs preliminary caching of the received data to ensure that the data will not be lost due to instantaneous peaks. The data buffer is also responsible for simple format checks and preprocessing of the data to ensure data consistency and integrity.

[0049] (4) Message queue management and distribution: The data in the cache is sent to a message queue (such as Apache Kafka) for unified management and distribution. The message queue transmits the data to downstream processing modules (such as data cleaning submodule, data conversion submodule, etc.) according to the preset distribution strategy (such as load balancing).

[0050] (5) Data transmission and logging: During the data transmission process, the message queue records all relevant information about the data transmission, including data source, transmission time, data volume, etc. The log system records all receiving and distributing operations to ensure the transparency and traceability of the data transmission process.

[0051] S130, real-time detection and processing of missing values, duplicate data and outliers in data to ensure data quality; apply various real-time data cleaning algorithms to efficiently clean real-time data; record the data cleaning process, provide operation logs and results, and facilitate subsequent audits and optimization. Specifically:

[0052] (1) Initialization phase: When the system starts, the data cleaning submodule is first initialized, including the configuration and startup of the real-time data quality checker, the real-time data cleaning algorithm library, and the cleaning log system.

[0053] (2) Real-time data quality check: Data enters the data cleaning submodule from the data receiving submodule.

[0054] The real-time data quality checker performs quality checks on the received data to detect missing values, duplicate data, and outliers in the data.

[0055] (3) Data cleaning algorithm selection: Based on the detection results, the system selects an appropriate cleaning algorithm from the real-time data cleaning algorithm library. For example, for missing values, you can choose interpolation or mean filling; for duplicate data, you can choose a deduplication algorithm; for outliers, you can choose a sliding window method for processing.

[0056] (4) Data cleaning process: The selected cleaning algorithm processes the data to clean missing values, duplicate data, and outliers in real time to ensure data quality. During the cleaning process, all operations and results are recorded, including the algorithm used, the amount of data processed, and the data quality indicators before and after cleaning.

[0057] (5) Cleaning result records: The cleaning log system records the entire data cleaning process, including various operations and results. The cleaning log provides a detailed reference for subsequent audits and optimizations, ensuring the transparency and traceability of the data cleaning process.

[0058] (6) Data transfer to the next module: The cleaned data is transferred to the data conversion submodule or other downstream processing modules to ensure that the data quality is improved and provide a reliable basis for subsequent data processing and analysis.

[0059] S140: Convert the real-time data into a unified format to facilitate subsequent processing and analysis; standardize the real-time data to ensure data consistency; distribute the data to the corresponding processing nodes according to the data type and processing requirements. Specifically:

[0060] (1) Initialization phase: When the system starts, the data conversion submodule is first initialized, including the configuration and startup of the real-time format converter, real-time standardization tool, and data router.

[0061] (2) Receiving cleaned data: Data is transferred from the data cleaning submodule to the data conversion submodule.

[0062] The real-time format converter receives the cleaned data and prepares it for format conversion and normalization.

[0063] (3) Data format conversion: The real-time format converter converts data into a unified format, such as JSON, Avro, Parquet, etc., according to the preset format requirements. During the format conversion process, the integrity of the data structure and content is ensured to avoid data loss or format errors.

[0064] (4) Data standardization: The converted data is standardized through real-time standardization tools to ensure data consistency. Standardization includes unit conversion, data range adjustment, data type correction, etc., to ensure that data from different sources have the same standards and formats.

[0065] (5) Data distribution: The data router distributes the processed data to the corresponding processing nodes according to the data type and processing requirements. The data can be transmitted to the advanced analysis and prediction module, causal relationship analysis and optimization module or other downstream modules for further processing and analysis.

[0066] (6) Record the conversion process: All operations and results during the data conversion process will be recorded in the system log. This log information is used for subsequent auditing and optimization to ensure the transparency and traceability of the data conversion process.

[0067] S150, use the real-time analysis engine to process and analyze data in real time; execute the analysis rules defined in the real-time rule engine to perform threshold monitoring and pattern recognition; apply machine learning models to perform real-time prediction and anomaly detection to improve the accuracy of analysis; integrate mathematical and physical models to evaluate supply chain key performance indicators (KPIs), such as inventory turnover, order fulfillment time, transportation costs, etc. Specifically:

[0068] ① We will obtain key data in the supply chain from the real-time data stream receiver, including inventory level I(t), order volume O(t), shipment time S(t), cost C(t), etc., where t represents the time point.

[0069] ②Set inventory level thresholds , if I(t) < , a replenishment alarm is triggered; the pattern recognition algorithm M is used to detect the order pattern, where M(O(t)) = p, p is the pattern matching probability.

[0070] ③ Use the calculus equation to calculate the inventory turnover rate R, assuming that the inventory level change rate is ,but here, and Represents the start and end time points respectively.

[0071] Predicting order fulfillment time using higher-order function partial derivative formula , let the function of order quantity changing with time be f(t), then Among them, c and b are constants obtained by fitting historical data, representing the order growth rate and base.

[0072] ④ Combine the logistics network flow model and numerical analysis technology to optimize the transportation cost C. Assume that the flow at a certain point on the transportation path is F(t) and the cost function is g(F(t)), then

[0073] Where N is the number of nodes on the transportation path.

[0074] ⑤Integrate the above calculation results into the comprehensive evaluation model E to evaluate the key performance indicators K of the supply chain, such as in, , , is the weight coefficient, determined by expert knowledge or machine learning.

[0075] ⑥ Dynamically adjust inventory thresholds based on real-time analysis results , transportation routes, and cost function parameters to optimize overall supply chain efficiency.

[0076] S160, store the processed real-time data, provide efficient query and access support; store the real-time data in a distributed file system, support large-scale data storage and processing; distribute the processed real-time data to various application systems and user interfaces, and provide data support. Specifically:

[0077] (1) Real-time data enters the system through the data receiving submodule and is temporarily stored in the data buffer. A message queue (such as Apache Kafka) is used to manage and distribute real-time data streams to ensure the reliability and efficiency of data transmission.

[0078] (2) The data cleaning submodule detects and processes missing values, duplicate data, and outliers in the data; the data conversion submodule converts real-time data into a unified format and performs standardized processing to ensure data consistency.

[0079] (3) The real-time analysis submodule uses real-time analysis engines (such as Apache Flink and Spark Streaming) to process and analyze data in real time; executes analysis rules defined in the real-time rule engine to perform threshold monitoring and pattern recognition; and applies machine learning models for real-time prediction and anomaly detection.

[0080] (4) The processed real-time data is stored in a real-time database (such as Apache Druid, InfluxDB) for efficient query and access; for large-scale real-time data, it is stored in a distributed file system (such as HDFS) to support flexible data storage and management.

[0081] (5) The data distributor distributes the processed real-time data to various application systems and user interfaces based on data type and requirements, ensuring the timeliness and accuracy of the data and providing real-time data support to users.

[0082] S200, extract features from data, train models, evaluate and predict, provide decision support based on real-time data, and dynamically update models. Step S200 includes steps S210-S250:

[0083] S210, use mathematical and physical indicators to extract features from raw data to improve data representativeness and analysis results; select the most important features through statistical and machine learning methods to reduce the dimension and complexity of model training; divide the data into training sets, validation sets, and test sets to ensure the generalization ability of the model. Specifically:

[0084] ① Extract meaningful features from the raw data to improve the representativeness and analysis effect of the data, and ensure that subsequent analysis and prediction are based on high-quality features.

[0085] Innovation model (Formula: ) is used to identify trends and periodicity of data over time and extract time-related features.

[0086] It is used to convert time domain data into frequency domain, extract frequency domain features, and help identify periodicity and frequency components in data. It is particularly useful for analyzing periodic patterns or noise processing. The formula is:

[0087] Calculating the volatility of data, such as standard deviation and variance, helps understand the degree of volatility and stability of data, which is crucial for risk assessment and anomaly detection. The formula is:

[0088] ② Select the most important features through statistical and machine learning methods, reduce the dimension and complexity of model training, and improve model training efficiency and prediction accuracy.

[0089] By introducing L1 regularization constraints, the features that have the greatest impact on the model are screened out, unimportant features are automatically removed, the model complexity is reduced, and overfitting is prevented. The formula is:

[0090] Through dimensionality reduction technology, the component with the largest variance in the data is retained, the data dimension is reduced, and the hidden structure in the data may be discovered. The formula is: Z=XW.

[0091] The new algorithm calculates feature importance, selects important features, and helps select the most influential features. The formula is:

[0092] ③ Divide the data into training set, validation set and test set to ensure that the model has good generalization ability and can perform well on unseen data to avoid overfitting.

[0093] Cross-validation or holdout method: Use cross-validation or holdout method to divide the data set into training set, validation set and test set to ensure the generalization ability of the model. The formula is:

[0094] S220: Apply mathematical models for preliminary analysis and forecasting, such as linear regression and time series analysis; use physical models to simulate physical processes of logistics network flow, inventory management and production processes; integrate mathematical models with physical models for comprehensive analysis and forecasting to provide more accurate supply chain assessments. Specifically:

[0095] ① Predict the future trends of key indicators in the supply chain (such as inventory turnover, order fulfillment time, transportation costs, etc.). Algorithm formula:

[0096] Among them, Y(t) is the key indicator of the supply chain at the predicted time point t, is the i-th supply chain indicator associated with time t, is with is the relevant regression coefficient, and ϵ is the error term.

[0097] ②Analyze time series data of supply chain indicators to identify and predict seasonal or cyclical patterns. Algorithm formula:

[0098] in, is the supply chain indicator at time point t, and are the autoregressive and moving average parameters, respectively, is the error term.

[0099] ③ Optimize logistics network flow through complex analysis technology to reduce congestion and delays. Algorithm formula:

[0100] in, is the optimal flow, F(w) is the flow distribution function in the frequency domain, Re{⋅} represents the real part, is a specific frequency point.

[0101] ④ Use physical models to simulate changes in inventory levels, predict optimal inventory levels, and reduce inventory costs. Algorithm formula:

[0102] Where I(t) is the inventory at time t, I(0) is the initial inventory, R(s) and D(s) are the purchase rate and shipment rate at time s, respectively.

[0103] ⑤ Use physical models to simulate the production process and optimize production efficiency. Algorithm formula:

[0104] Among them, E is the production efficiency, η is the efficiency coefficient, P is the production power, t is the production time, and α is the loss coefficient related to power and time.

[0105] ⑥ Integrate mathematical models with physical models to conduct comprehensive analysis and prediction, and provide more accurate supply chain evaluation. Algorithm formula:

[0106] in, is the result of a comprehensive assessment. and are the outputs of the mathematical model and the physical model, and is the weight coefficient.

[0107] ⑦ Verify the accuracy and stability of the model, optimize model parameters, and improve prediction accuracy. Algorithm formula:

[0108] in, is the actual value, is the predicted value and N is the number of data points.

[0109] S230. Apply various machine learning and deep learning algorithms to train models to improve prediction accuracy; optimize model parameters through hyperparameter tuning to improve model performance; use distributed computing platforms for distributed training to improve model training efficiency. Specifically:

[0110] ① Use high-order tensor regression to process the complex relationship of multi-source heterogeneous data in the supply chain and improve prediction accuracy. The formula is:

[0111] in, is the predicted supply chain key performance indicator (such as inventory turnover, order fulfillment time, transportation cost, etc.), T is the tensor of input data, , , is the weight tensor, B is the bias tensor, Represents a product operation along the nth mode.

[0112] ② Use SVR to process nonlinear relationships in supply chain data and enhance the model's ability to predict complex data. Formula:

[0113] Where f(x) is the prediction function, is the Lagrange multiplier, K( , x) is the kernel function, b is the bias term, is the training sample, and N is the number of samples.

[0114] ③ Use DCNN to automatically extract complex features in supply chain data and improve prediction accuracy. Formula:

[0115] in, is the prediction result, σ is the activation function, and is the weight and bias of the i-th layer, representing the convolution operation, and X is the input data.

[0116] ④ Automatically adjust the hyperparameters of the machine learning model through Bayesian optimization to achieve optimal performance. Formula:

[0117] in, is the optimized hyperparameter, E is the expected value, p(y|θ) is the predicted probability distribution given the hyperparameter θ, and u(y) is the utility function.

[0118] ⑤ Use PSO to globally search for optimal model parameters to improve model performance. Formula:

[0119] in, (t) is the velocity of particle i at time t, ω is the inertia weight, , is the acceleration constant, , is a random number, (t) is the historical optimal position of particle i, (t) is the historical optimal position of the group, (t) is the position of particle i at time t.

[0120] ⑥ Accelerate the model training process on a distributed computing platform through data parallelism. Formula:

[0121] Where ∇L is the gradient of the total loss function, m is the batch size, and ∇l( , ; θ) is a single data point ( , )The gradient of the loss function with respect to the parameters θ.

[0122] ⑦ Split the model parameters into multiple subsets and assign them to different computing nodes to calculate the gradients in parallel, further improving the model training efficiency. Formula:

[0123] in, is a parameter gradient.

[0124] S240, use various evaluation indicators to evaluate the performance of the model to ensure the reliability of the model; perform K-fold cross validation to evaluate the stability and generalization ability of the model; use explanatory methods to explain the model prediction results to increase the transparency and interpretability of the model. Specifically:

[0125] (1) According to the requirements of the supply chain evaluation system, define various evaluation model performance indicators, such as accuracy, precision, recall, F1 score, etc.; select appropriate evaluation indicators according to different evaluation scenarios and requirements.

[0126] (2) Use the K-fold cross-validation method to evaluate the model. Divide the dataset into K subsets, reuse each subset as a validation set, and use the remaining subsets as training sets for training and validation. Calculate the evaluation indicators for each validation, take the average, and evaluate the stability and generalization ability of the model.

[0127] (3) Use explanatory methods (such as LIME, SHAP, etc.) to explain the model prediction results, identify the features that have the greatest impact on the prediction results, and improve the transparency and interpretability of the model; generate explanatory reports to help users understand the model's prediction results.

[0128] (4) Generate a model evaluation report based on various evaluation indicators and explanatory reports, including model performance indicators, stability evaluation, prediction result explanation, etc.; provide visual charts to display the model evaluation results, so that users can intuitively understand the performance and reliability of the model.

[0129] S250, deploy pre-trained models for real-time data prediction and anomaly detection, provide real-time analysis capabilities; provide decision support based on prediction results and business rules to help companies optimize supply chains; dynamically update and optimize models based on real-time feedback and new data to improve model adaptability and accuracy. Specifically:

[0130] (1) Deploy the pre-trained model to the system and configure the model service interface to ensure that the model can receive data and make predictions in real time; manage and coordinate requests for the model service interface through the API gateway to ensure the reliability and security of the model service.

[0131] (2) The deployed model receives data from the real-time data stream, performs real-time predictions, and outputs prediction results. It also performs preliminary analysis on the prediction results, identifies abnormal situations, and records them.

[0132] (3) Use pre-trained anomaly detection models to perform anomaly detection on real-time data and identify outliers and abnormal patterns in the data; define anomaly thresholds and processing strategies based on business rules, and alarm and process detected anomalies.

[0133] (4) Generate decision support recommendations based on forecast results and business rules to help companies optimize their supply chains; provide a visual display of real-time data to facilitate users to view and analyze forecast results and abnormal situations.

[0134] (5) Dynamically update and optimize the model based on real-time feedback and new data to improve the adaptability and accuracy of the model; use online learning algorithms to update model parameters in real time to ensure that the model maintains high performance in a constantly changing data environment.

[0135] S300: Build a causal relationship model, perform causal reasoning, optimize supply chain decisions, and adjust the model and decision-making plan based on feedback. Step S300 includes steps S310-S350:

[0136] S310, obtain data from data sources such as ERP systems, logistics systems, sales systems, sensors, etc., and build a causal relationship data set; apply feature engineering technology to extract and select mathematical and physical indicators to ensure that the data is suitable for causal relationship modeling. Specifically:

[0137] (1) Obtain data from data sources such as ERP systems, logistics systems, sales systems, sensors, etc., manage and coordinate data interface requests through the API gateway to ensure the reliability and security of data collection; the real-time data stream interface receives real-time data streams from sensors and IoT devices, and the data stream receiver processes the initial analysis of the real-time data stream and temporarily stores the data in the buffer.

[0138] (2) Clean and convert the collected data, process missing values, duplicate data and outliers in the data to ensure data quality; convert the data into a unified format and perform standardization to ensure data consistency.

[0139] (3) Apply feature engineering techniques to extract and select mathematical and physical indicators from raw data and construct a causal data set; use time series analysis, Fourier transform, volatility analysis and other methods to extract representative features from the data.

[0140] (4) Integrate the processed data into the causal relationship dataset to provide basic data for causal relationship modeling; ensure that the causal relationship dataset contains sufficient feature information to reflect the relationship between the variables in the supply chain.

[0141] S320, use models such as structural equation model (SEM), Bayesian network and Granger causality analysis to build a causal relationship model between variables in the supply chain; train the causal relationship model to learn and reveal the causal relationship between variables in the supply chain; verify the accuracy and stability of the causal relationship model to ensure the reliability of the model. Specifically:

[0142] (1) According to the needs of supply chain evaluation, select appropriate causal modeling methods, such as structural equation model (SEM), Bayesian network and Granger causality analysis; configure model parameters and initial conditions to ensure that the model can accurately reflect the relationship between the variables in the supply chain.

[0143] (2) Use the causal relationship dataset to train the selected model to learn and reveal the causal relationship between variables in the supply chain; during the training process, adjust the model parameters to ensure that the model can accurately capture the causal relationship between variables.

[0144] (3) Use cross-validation or holdout method to validate the model and evaluate its accuracy and stability. Based on the validation results, optimize the model parameters to ensure the model's performance on new data.

[0145] (4) Use the trained model to perform causal analysis on the supply chain data to identify the causal relationship between the variables; based on the analysis results, generate a causal report to reveal the mutual influence between the variables in the supply chain.

[0146] S330. Apply causal reasoning algorithms to conduct reasoning analysis of causal relationships, such as counterfactual reasoning and intervention analysis; use the reasoning rule library to guide the reasoning process to ensure the accuracy and interpretability of the reasoning results; generate easy-to-understand analysis reports, explain the causal reasoning results, and provide decision support. Specifically:

[0147] (1) According to the needs of supply chain assessment, select appropriate causal reasoning algorithms, such as counterfactual reasoning and intervention analysis; configure the inference rule base to guide the reasoning process and ensure the accuracy and interpretability of the reasoning results.

[0148] (2) Apply the selected causal inference algorithm to perform inference analysis on the causal relationship model and generate inference results; guide the inference process based on the inference rule library to ensure the accuracy and interpretability of the inference results.

[0149] (3) Integrate causal reasoning results into analysis reports, explain the reasoning results, and provide decision support; use visualization tools to display causal reasoning results to help users intuitively understand the analysis results.

[0150] S340. Use the algorithms in the optimization algorithm library to make supply chain optimization decisions based on causal reasoning results; build decision models to simulate and evaluate the effects of different decision plans; use decision simulators to evaluate the feasibility and effectiveness of decision plans and provide optimization suggestions for enterprises. Specifically:

[0151] (1) Based on the results of causal reasoning, select a suitable optimization algorithm, such as mixed integer linear programming (MILP) and genetic algorithm (GA); configure the optimization algorithm parameters to ensure that the algorithm can effectively solve the supply chain optimization problem.

[0152] (2) Construct a supply chain optimization decision model to simulate and evaluate the effects of different decision plans; use optimization algorithms to solve the optimization decision model and generate the optimal decision plan.

[0153] (3) Through the decision simulator, different decision plans are simulated and evaluated to analyze their feasibility and effectiveness; optimization decision reports are generated, and optimization suggestions are provided to help enterprises optimize their supply chains.

[0154] (4) Implement optimized decision-making plans, collect actual execution results and user feedback, and evaluate the actual effects of the decision-making plans; based on the feedback results, adjust the causal model and optimize the decision-making model to improve the accuracy and adaptability of the model.

[0155] S350: Collect actual execution results and user feedback to evaluate the actual effect of the decision-making plan; adjust the causal model and optimize the decision-making model according to the feedback results to improve the accuracy and adaptability of the model; dynamically update the model and optimization plan to ensure the timeliness and accuracy of the decision. Specifically:

[0156] (1) Collect the actual execution results and user feedback after the implementation of the optimization decision plan and record them in the system log; based on the feedback data, evaluate the actual effect of the decision plan.

[0157] (2) Use evaluation indicators to evaluate the actual effect of the optimized decision-making plan, such as accuracy, execution efficiency, cost savings, etc.; based on the evaluation results, generate a decision-making plan evaluation report and provide decision optimization suggestions.

[0158] (3) Based on the feedback results, adjust the causal model and optimize the decision-making model to improve the accuracy and adaptability of the model; dynamically update the model and optimization plan to ensure the timeliness and accuracy of the decision.

[0159] (4) Establish a feedback and adjustment mechanism to continuously collect actual implementation results and user feedback, and continuously optimize decision-making models and plans; ensure the continuous improvement and optimization of supply chain optimization decisions.

[0160] S400: Display data through charts and dashboards, generate decision suggestions, provide user interfaces and notifications, and support user interaction and feedback. Step S400 includes steps S400-S440:

[0161] S410, generate complex charts and dynamic visualizations, combine mathematical and physical indicators to intuitively display data analysis results; provide interactive dashboards, support real-time data display and user interaction, enhance user experience; automatically generate regular reports, including charts and text descriptions, for easy sharing and recording. Specifically:

[0162] (1) Obtain processed real-time data from the data storage submodule, process and aggregate the data according to analysis requirements; apply advanced data analysis methods to generate key analysis indicators and results.

[0163] (2) Use visualization tools to generate complex charts and dynamic visualizations to display supply chain key performance indicators (KPIs) and analysis results; combine mathematical and physical indicators to generate various types of charts, such as line charts, bar charts, pie charts, heat maps, etc., to intuitively display data analysis results.

[0164] (3) Provide interactive dashboards that support real-time data display and user interaction to enhance user experience; dynamically update charts and visualization results to ensure the real-time and accuracy of data.

[0165] (4) Automatically generate regular reports, including charts and text descriptions, for easy sharing and recording; provide report export function, supporting multiple formats such as PDF, Excel, etc.

[0166] S420, based on mathematical and physical models, combined with business rules, generate scientific decision-making suggestions; simulate the implementation effects of different decision-making plans, evaluate their feasibility and effectiveness, and demonstrate the mathematical optimization process; combine causal analysis and optimization algorithms to generate optimization suggestions and improve decision-making quality. Specifically:

[0167] (1) Based on mathematical and physical models, conduct in-depth analysis of supply chain data to generate key analysis results; combine business rules and decision-making requirements to generate scientific decision-making recommendations.

[0168] (2) Simulate the implementation effects of different decision-making plans and evaluate their feasibility and effectiveness; demonstrate the mathematical optimization process and display the decision-making simulation results through visualization tools.

[0169] (3) Combine causal analysis and optimization algorithms to generate optimization suggestions and improve decision-making quality; provide decision support reports, including optimization suggestions and simulation results, to help companies optimize their supply chains.

[0170] S430, provide multiple data input methods, support manual input, file upload and real-time data stream access; display analysis results and decision suggestions, provide rich interactive functions, including zooming, filtering and dynamic updating; collect user feedback and adjustment suggestions, and continuously optimize the decision support system. Specifically:

[0171] (1) Provides multiple data input methods, supports manual input, file upload, and real-time data stream access to configure the data input interface, ensuring the convenience and flexibility of data input.

[0172] (2) Display analysis results and decision recommendations, and provide rich interactive functions, including zooming, filtering, and dynamic updating; support users to view and analyze data in multiple dimensions, and provide an intuitive data display interface.

[0173] (3) Collect user feedback and adjustment suggestions and record them in the system log; optimize data display and decision support functions based on user feedback to continuously improve user experience.

[0174] S440, send regular notifications of analysis results, decision-making recommendations and key indicators to users to ensure timely communication of information; monitor key indicators in real time, detect abnormal situations and send alerts, and perform anomaly detection in combination with mathematical models; record the sending and response of all notifications and alerts to facilitate auditing and optimization. Specifically:

[0175] (1) Configure the notification sending interface to support multiple notification methods, such as email, SMS, push notification, etc.; define notification rules to regularly send analysis results, decision recommendations, and key indicators according to user needs.

[0176] (2) Monitor key indicators in real time, detect anomalies and send alerts, and use mathematical models to detect anomalies; record the sending and response of all notifications and alerts to facilitate auditing and optimization.

[0177] (3) Send regular notifications of analysis results, decision recommendations, and key indicators to users to ensure timely communication of information; provide notification history so that users can view and manage past notifications.

[0178] S500, monitor performance, security management, configure environment, support system expansion and integration, user management and behavior analysis, and ensure the safe, stable and efficient operation of the system. S500 includes steps S510-S550:

[0179] S510, real-time monitoring of system resource usage, providing detailed performance indicators and health status information; combining mathematical models and physical models to analyze system performance, identify performance bottlenecks and optimization points; collecting and managing system logs, and performing fault diagnosis through big data analysis and physical models. Specifically:

[0180] (1) Real-time monitoring of system resource usage, including CPU, memory, disk, and network resources; providing detailed performance indicators and health status information, and displaying system resource usage through a dashboard.

[0181] (2) Analyze system performance by combining mathematical models and physical models, identify performance bottlenecks and optimization points, generate performance analysis reports, and provide optimization suggestions to help system administrators improve system performance.

[0182] (3) Collect and manage system logs, perform fault diagnosis through big data analysis and physical models; identify the causes of system failures, provide fault repair suggestions, and ensure stable system operation.

[0183] S520, role-based and behavior-based access control model to ensure the security of system data; use advanced encryption standards to encrypt storage and transmission of sensitive data to protect data privacy; combine physical models and machine learning algorithms to detect and respond to potential security threats and attacks to ensure system security. Specifically:

[0184] (1) Configure a role-based and behavior-based access control model, define user roles and permissions, and ensure the security of system data; use advanced encryption standards to encrypt storage and transmission of sensitive data to protect data privacy.

[0185] (2) Combine physical models and machine learning algorithms to detect and respond to potential security threats and attacks; monitor the system security status in real time, and record security incidents and response measures.

[0186] (3) Dynamically adjust access control policies based on security threats and user behaviors to improve system security; provide a security policy management interface to support administrators in viewing and managing security policies and events.

[0187] S530. Use version control systems to manage configuration files, support automatic loading, modification and version control of configurations; use containerization technology to achieve automated deployment and upgrades of the system to ensure configuration consistency and reliability; manage configuration and resource allocation for different operating environments, and support development, testing and production environments. Specifically:

[0188] (1) Use a version control system (such as Git) to manage configuration files and record the change history and version information of the configuration files; support automatic loading, modification, and version control of configuration files to ensure configuration consistency and reliability.

[0189] (2) Use containerization technology (such as Docker) to achieve automated deployment and upgrade of the system, ensure configuration consistency and reliability, manage configuration and resource allocation of different operating environments, and support development, testing, and production environments.

[0190] (3) Record all changes to configuration files and provide detailed change logs to facilitate auditing and troubleshooting; optimize the configuration management process based on the audit results to improve the efficiency and security of system configuration management.

[0191] S540, provides a unified API entry, combines load balancing algorithms and traffic control strategies to manage and protect API access; supports the development, installation and management of plug-ins, combines mathematical models to evaluate the performance and impact of plug-ins; uses data fusion technology and physical models to provide data integration interfaces with external systems, supports multiple data formats and protocols, and enhances system interoperability. Specifically:

[0192] (1) Provide a unified API entry, manage and protect API access, and ensure the security and stability of the API; combine load balancing algorithms and traffic control strategies to optimize API access performance and prevent overload and denial of service attacks.

[0193] (2) Support the development, installation, and management of plug-ins, and use mathematical models to evaluate the performance and impact of plug-ins; provide a plug-in management interface to support administrators to view and manage installed plug-ins.

[0194] (3) Use data fusion technology and physical models to provide data integration interfaces with external systems and support multiple data formats and protocols; ensure the interoperability and scalability of the system and enhance data integration and exchange capabilities.

[0195] S550, use multi-factor authentication and behavior analysis technology to ensure the uniqueness and security of user identity; manage user permissions and roles based on role and behavior access control models, and control user access to system resources; combine data mining and physical models to monitor and analyze user behavior, and provide audit and optimization suggestions. Specifically:

[0196] (1) Use multi-factor authentication technology (such as passwords, SMS verification codes, fingerprints, etc.) to ensure the uniqueness and security of user identities; combine behavioral analysis technology to detect abnormal user behavior and prevent identity theft and illegal access.

[0197] (2) The role- and behavior-based access control model manages user permissions and roles, controls user access to system resources, and dynamically adjusts user permissions to ensure the security and compliance of system data.

[0198] (3) Combine data mining and physical models to monitor and analyze user behavior and identify potential security threats; provide behavior analysis reports and record abnormal user behavior to facilitate auditing and optimization.

[0199] The key innovations of the present invention are concentrated in the following aspects: (1) Multi-source data integration and real-time analysis. Multi-source data acquisition: The system obtains data from multiple data sources such as ERP systems, logistics systems, sales systems, supplier systems, sensors, and IoT devices to ensure the comprehensiveness and diversity of the data; real-time data stream processing: Through API gateways and message queues, the real-time data stream is managed and distributed to ensure the reliability and efficiency of data transmission; data quality assurance: Real-time detection and processing of missing values, duplicate data, and outliers in the data, using a variety of cleaning algorithms to ensure high data quality.

[0200] (2) Application of advanced mathematical and physical models. Multiple mathematical models: The system integrates models such as linear regression, time series analysis, support vector regression (SVR), and high-order tensor regression to handle complex relationships and nonlinear data in the supply chain. Physical model integration: Through complex number analysis, inventory management models, and production process physical models, the logistics network flow, inventory level, and production efficiency in the supply chain are simulated and optimized. Model fusion and optimization: Combine mathematical models and physical models to conduct comprehensive analysis and prediction to improve the accuracy and reliability of supply chain evaluation.

[0201] (3) Machine learning and deep learning technologies. Feature engineering and feature selection: systematically apply feature engineering technology to extract and select important features from multidimensional data to improve the representativeness and predictive ability of the model; deep learning model: use deep convolutional neural network (DCNN) to automatically extract complex features and improve prediction accuracy; model optimization: automatically tune model hyperparameters through Bayesian optimization and particle swarm optimization (PSO) algorithms to ensure optimal model performance.

[0202] (4) Causal analysis and decision optimization. Causal modeling: Use structural equation model (SEM), Bayesian network and Granger causality analysis to build a causal model between variables in the supply chain and reveal the mutual influence between the variables; Causal reasoning: Apply causal reasoning algorithms to perform counterfactual reasoning and intervention analysis, generate easy-to-understand analysis reports, and provide decision support for supply chain optimization; Optimal decision-making: Combine causal analysis results with optimization algorithms to simulate and evaluate the effects of different decision-making plans, generate optimization suggestions, and help companies optimize their supply chains.

[0203] (5) Dynamic visualization and real-time decision support. Real-time data display: Provide interactive dashboards and dynamic visualization tools to intuitively display key supply chain indicators and analysis results, and support users to view and analyze data in real time; Automatic report generation: The system can automatically generate regular reports, including charts and text descriptions, to facilitate users to share and record analysis results; Real-time monitoring and alarms: Real-time monitoring of key indicators, detection of abnormal situations and sending alarms, combined with mathematical models for anomaly detection, to ensure the stable operation of the supply chain.

[0204] (6) Efficient data management and security assurance. Distributed storage and query: The system uses a distributed file system and real-time database to support efficient storage and fast query of large-scale data; Data security and access control: The role-based and behavior-based access control model, combined with advanced encryption standards and multi-factor authentication technology, ensures the security and privacy protection of system data; System performance monitoring and optimization: Real-time monitoring of system resource usage, combined with mathematical models and physical models to analyze system performance, identify performance bottlenecks and optimization points, and ensure efficient operation of the system.

[0205] These innovations together constitute the core advantages of this invention. By comprehensively applying technologies such as multi-source data integration, real-time analysis, mathematical and physical models, machine learning and deep learning, causal analysis and optimized decision-making, it provides a scientific, reliable and efficient solution for enterprise supply chain evaluation.

[0206] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for evaluating the enterprise supply chain based on big data, characterized in that: The following steps are involved: S100, obtain data from multiple sources and receive data streams in real time, perform data cleaning, conversion and real-time analysis, and store the processed data and distribute it to the application system; S200, extract features from data, train models, evaluate and predict, provide decision support based on real-time data, and dynamically update models; S300, build causal relationship models, perform causal reasoning, optimize supply chain decisions, and adjust models and decision plans based on feedback; S400, display data through charts and dashboards, generate decision suggestions, provide user interface and notifications, and support user interaction and feedback; S500, monitors performance, security management, configures environment, supports system expansion and integration, user management and behavior analysis, and ensures safe, stable and efficient operation of the system.

2. According to the big data-based enterprise supply chain evaluation method of claim 1, it is characterized in that: The step S100 specifically includes: S110, obtain raw data from different data sources, including ERP system, logistics system, sales system, and supplier system; S120, receiving real-time data streams from sensors and IoT devices; S130. Manage and coordinate requests from different data interfaces through the API gateway to ensure the reliability and security of data collection; S140, the data stream receiver processes the preliminary analysis of the real-time data stream and temporarily stores the data in a buffer; S150, using the real-time analysis engine to process and analyze the data in real time, executing the analysis rules defined in the real-time rule engine, performing threshold monitoring and pattern recognition, and applying the machine learning model to perform real-time prediction and anomaly detection; S160, store the processed real-time data, provide efficient query and access support, store the real-time data in a distributed file system, and distribute the processed real-time data to various application systems and user interfaces.

3. According to the big data-based enterprise supply chain evaluation method of claim 2, it is characterized in that: The step S200 specifically includes: S210, extract features from raw data using mathematical and physical indicators, select the most important features through statistical and machine learning methods, and divide the data into training sets, validation sets, and test sets; S220. Use mathematical models for preliminary analysis and prediction, use physical models to simulate physical processes of logistics network flow, inventory management and production technology, and integrate mathematical models with physical models for comprehensive analysis and prediction; S230, use machine learning and deep learning techniques, including feature engineering and feature selection, deep convolutional neural networks to automatically extract complex features, and automatically tune model hyperparameters through Bayesian optimization and particle swarm optimization algorithms; S240. Provide decision support based on real-time data and dynamically update the model.

4. According to the big data-based enterprise supply chain evaluation method of claim 3, it is characterized in that: The step S300 specifically includes: S310, Use structural equation model, Bayesian network and Granger causality analysis to build causal relationship model between variables in supply chain; S320. Apply causal reasoning algorithms to conduct counterfactual reasoning and intervention analysis, generate easy-to-understand analysis reports, and provide decision support for supply chain optimization; S330. Combine causal analysis results and optimization algorithms to simulate and evaluate the effects of different decision-making plans and generate optimization recommendations.

5. According to the big data-based enterprise supply chain evaluation method of claim 4, it is characterized in that: The step S400 specifically includes: S410, provides interactive dashboards and dynamic visualization tools to intuitively display key supply chain indicators and analysis results; S420, automatically generate regular reports, including charts and text descriptions, to facilitate users to share and record analysis results; S430, monitor key indicators in real time, detect abnormal situations and send alarms, and perform anomaly detection in combination with mathematical models.

6. The enterprise supply chain evaluation method based on big data according to claim 5 is characterized in that: The step S500 specifically includes: S510, monitor the usage of system resources in real time, analyze system performance by combining mathematical models and physical models, and identify performance bottlenecks and optimization points; S520, role-based and behavior-based access control model, combined with advanced encryption standards and multi-factor authentication technology, ensures the security and privacy protection of system data; S530, supports system expansion and integration, user management and behavior analysis, ensuring the safe, stable and efficient operation of the system.

7. An enterprise supply chain evaluation system based on big data, the system implements the method described in any one of claims 1 to 6, characterized in that: include: Data acquisition and preprocessing module, used to obtain data from multiple sources and receive data streams in real time, perform data cleaning, conversion and real-time analysis, store the processed data and distribute it to the application system; Feature extraction and model training module, used to extract features from data, train models, evaluate and predict, provide decision support based on real-time data, and dynamically update models; Causality analysis and decision optimization module, used to build causal relationship models, perform causal reasoning, optimize supply chain decisions, and adjust models and decision plans based on feedback; Dynamic visualization and real-time decision support module, used to display data through charts and dashboards, generate decision suggestions, provide user interface and notifications, and support user interaction and feedback; Performance monitoring and security assurance module, used to monitor performance, security management, configuration environment, support system expansion and integration, user management and behavior analysis, to ensure the safe, stable and efficient operation of the system.

8. The enterprise supply chain evaluation system based on big data according to claim 7 is characterized in that: Also includes: Data management module, used to support efficient storage and fast query of large-scale data; Security management module, used to ensure data security and privacy protection; The system performance monitoring module is used to monitor the usage of system resources in real time to ensure the efficient operation of the system.

Images