Tracking method based on full-automatic flow monitoring of lithium battery positive electrode material production process

Abstract

The invention discloses a tracking method based on full-automatic flow monitoring of a lithium battery positive electrode material production process, and relates to the technical field of industrial automation control. According to the invention, by deploying the multi-source heterogeneous sensor network, the multi-dimensional data in the production process is comprehensively collected, the data blind area is eliminated, and the transparency and predictability of the production process are significantly improved; a distributed data collaborative analysis framework is constructed and combined with a federated learning algorithm, so that efficient sharing and joint modeling of multi-node data are realized, the problem of data collaborative processing is solved, and a global view angle is provided for optimization of a production process, so that the production efficiency and the product quality are greatly improved; the change trend of the key variables is monitored in real time, and deep analysis is carried out in combination with historical and real-time data, so that rapid identification and accurate intervention of the abnormal mode are realized, the stability and reliability of the production process are remarkably enhanced, and a powerful guarantee is provided for industrial upgrading and supply chain stability.

Description

Technical Field

[0001] This invention relates to the field of industrial automation control technology, specifically a tracking method for fully automated process monitoring of lithium battery cathode material production. Background Technology

[0002] In modern industrial manufacturing, especially in critical industries like cathode material production, process monitoring and optimization directly impact product quality and production efficiency, playing an irreplaceable role in driving industrial upgrading and ensuring supply chain stability. However, many current solutions suffer from significant shortcomings in technological integration and intelligence, often relying on single sensing methods and isolated data analysis. These solutions fail to comprehensively capture the complex changes in the production process, hindering real-time response and precise intervention, resulting in low production efficiency and frequent quality fluctuations.

[0003] A deeper analysis of the challenges in this field reveals that the primary problem lies in the insufficient comprehensive sensing capabilities of multi-source data. Due to the complexity of the production environment, a single sensor cannot cover all key aspects, resulting in blind spots in data collection and consequently, a lack of comprehensive basis for subsequent analysis and prediction. This problem further gives rise to the challenge of data collaborative processing: in a multi-node, multi-device environment, how to achieve efficient sharing and analysis while protecting data privacy has become a significant bottleneck hindering the construction of intelligent monitoring systems. These two issues are intertwined; the former directly affects data quality, while the latter limits the in-depth mining of data value, ultimately making it difficult to detect and effectively resolve production anomalies in a timely manner. Summary of the Invention

[0004] The purpose of this invention is to provide a tracking method for fully automated process monitoring of lithium battery cathode material production. Through multi-source data acquisition, distributed data collaborative analysis, and real-time anomaly intervention, the transparency, efficiency, stability, and product quality of the production process are comprehensively improved, providing an innovative solution for intelligent monitoring of lithium battery cathode material production.

[0005] The objective of this invention can be achieved through the following technical solutions: This application provides a tracking method for fully automated process monitoring of lithium battery cathode material production, including the following steps: By deploying a multi-source heterogeneous sensor network and arranging various types of sensing devices at key process nodes, environmental parameters, equipment status and material characteristics data are collected from multiple dimensions to obtain a comprehensive set of production process information. Data fusion technology is used to preprocess multi-source data to eliminate missing values ​​and noise interference caused by data blind spots, and to determine a unified standardized data format. A distributed data collaborative analysis framework is constructed. While protecting data privacy, it uses federated learning algorithms to achieve efficient sharing and joint modeling of multi-node data and obtain cross-device data correlation features. From the data correlation characteristics across devices, extract the changing trends of key variables in the production process, determine whether there are abnormal patterns that deviate from the preset threshold, and generate a preliminary abnormal alarm signal when an abnormal pattern is detected. By combining the correlation characteristics of historical and real-time data, a pre-established anomaly classification model is used for in-depth analysis to determine the specific type and scope of impact of the anomaly. Based on the specific type and scope of the anomaly, the preset process parameter adjustment rule library is invoked to obtain the corresponding intervention strategy configuration and generate targeted control instructions. By generating targeted control commands, the production equipment is driven to perform parameter adjustment operations, while real-time data changes after intervention are recorded to determine whether the control has achieved the expected stable state.

[0006] Furthermore, a comprehensive set of production process information is obtained, specifically including: By deploying a multi-source heterogeneous sensor network, diverse data from the production environment are acquired. Preprocessing techniques are used to clean and unify the collected environmental parameters, equipment status, and material characteristics data to obtain a standardized dataset. When certain environmental parameters in the standardized dataset exceed the preset threshold range, the anomaly detection mechanism is triggered to determine the location and type of the abnormal data points. Through the analysis of the abnormal data points, the trend of equipment status changes related to the process node is obtained to determine whether there are potential fault risks. Based on the trend of equipment status changes, the support vector machine algorithm is used to predict the operational stability of key process nodes and obtain stability assessment results. When the stability assessment results show that the operation is unstable, correlation analysis is performed on the material characteristic data to determine the main factors affecting stability. By continuously monitoring key factors and obtaining real-time feedback data, the acquisition frequency of the sensor network can be adjusted to optimize the update efficiency of comprehensive process information.

[0007] Furthermore, a unified standardized data format is determined, specifically including: By collecting data from multiple sources, we obtain the original dataset containing missing values ​​and noise interference. When the original dataset contains missing values, we use the K-nearest neighbor algorithm to interpolate and obtain the completed dataset. The wavelet transform method is used to filter noise interference to obtain a denoised dataset. The attribute features of each data source are obtained, the standardized data format is determined, and a formatted dataset is obtained. When the attributes of a formatted dataset do not conform to a unified format, field mapping technology is used to convert them to obtain a unified dataset. Then, principal component analysis is used to extract key features to obtain the dataset for analysis preparation.

[0008] Furthermore, obtain cross-device data association characteristics, specifically including: By constructing a distributed framework, preliminary data structuring results are obtained. Federated learning algorithms are then used to perform distributed training on multi-node data to obtain cross-node model parameter updates. When the model parameters are updated within the preset threshold range, further data collaboration processing is performed on the multi-node data to determine the potential consistency characteristics between the data. Feature mining is performed on cross-device data associations to determine the distribution pattern of association features. A joint modeling method is used to integrate multi-node data and obtain the feature mapping results of the global model. When the deviation between the feature mapping result and the preset target exceeds the threshold, the data privacy protection mechanism is adjusted to obtain an optimized privacy protection strategy. By using an optimized privacy protection strategy, the efficient sharing mechanism is verified to determine the final scope of shared data.

[0009] Furthermore, the changing trends of key variables in the production process are extracted to determine whether there are any abnormal patterns deviating from preset thresholds, specifically including: This involves acquiring multi-source information from cross-device data, integrating and processing data streams from different devices, preprocessing the data using a unified formatting method, extracting correlation features, and using correlation analysis to identify key variables in the production process and determine the distribution characteristics of these key variables. Based on the distribution characteristics of key variables, the changing trends are analyzed. The historical data of the variables are modeled using time series analysis methods to obtain trend prediction results. When the trend prediction results deviate from the preset threshold, they are identified as potential abnormal patterns. The significance of the abnormal patterns is judged by combining historical data and current data in multiple dimensions. When the significance of an abnormal pattern exceeds a predetermined standard, the anomaly detection mechanism is triggered, generating an initial alarm signal. This signal is transmitted to the monitoring module through the system's internal channel, where contextual data from the production process is obtained. The real-time status of key variables is then verified to confirm the accuracy of the alarm signal. Finally, the relevant characteristics of the abnormal pattern are automatically recorded and stored in the anomaly log database.

[0010] Furthermore, the specific type and scope of the anomaly are determined, including: The system acquires preliminary anomaly alarm signals, extracts real-time data streams from multiple sensor nodes through a data acquisition system to obtain a preliminary anomaly signal set, and uses data fusion technology to perform feature association between the preliminary anomaly signal set and historical data. The system then uses principal component analysis to extract key features to obtain a feature association dataset. When the feature values ​​in the feature association dataset exceed the preset threshold, a pre-established random forest model is used to classify anomalies, determine the anomaly type, and partition the real-time data stream using cluster analysis to determine the scope of anomaly impact and obtain the impact range dataset. By employing time series analysis, the abnormality types in the affected data set are analyzed to predict the abnormality development trend. The abnormality trend prediction results are then obtained, and the parameters of the abnormality classification model are updated to obtain an optimized classification model. By re-analyzing new abnormal signals in the real-time data stream, a more accurate abnormality type and affected range are determined.

[0011] Furthermore, targeted regulatory instructions are generated, specifically including: By extracting key features from the data on anomaly types and impact ranges, the specific classification and action boundaries of the anomalies are determined, relevant entries in the preset process parameter adjustment rule library are obtained, and a matching set of adjustment rules is obtained. A logical comparison method is used to determine the degree of fit between the rule and the current abnormal characteristics. When the degree of fit is higher than the preset threshold, the corresponding intervention strategy configuration is determined. Through the intervention strategy configuration, a preliminary draft of the control instruction is generated, and the parameter adjustment direction and magnitude information contained therein are obtained. Based on the parameter adjustment direction and magnitude information in the draft control instructions, and combined with historical data on anomaly handling, the instructions are optimized using a support vector machine model to obtain the adjusted control instructions. The system acquires real-time process parameter status data. When there is a discrepancy between the status data and the instruction content, it uses an iterative comparison method to determine the final targeted control instruction. Combined with the boundary data of the impact range, it generates an executable instruction sequence to complete the closed-loop control for anomaly handling.

[0012] Furthermore, determining whether the regulation has achieved the expected stable state specifically includes: The control system generates control commands to drive production equipment to adjust parameters, obtain equipment response data, acquire real-time data from the equipment response data, and perform structured processing using data acquisition technology to obtain a formatted dataset. A recording mechanism is used to store data changes. When the data changes exceed a preset threshold, a feedback loop is triggered to obtain an adjustment signal. The control system updates the control instructions and drives the production equipment to perform parameter adjustments again, thus obtaining new equipment response data. Extract data change trends from the response data of the new equipment, use result analysis technology to calculate stability indicators, and determine whether a stable state has been reached. If the stability indicators do not meet the preset standards, generate new control commands through feedback loops, repeat the adjustment process, and obtain the final stable data. Based on the final stable data, result analysis technology is used to generate control logs, which are stored in the recording mechanism to obtain business execution records.

[0013] Furthermore, after determining whether the regulation has reached the expected stable state, it also includes: updating the model parameters and rule base in the data collaborative analysis framework based on real-time data changes after the intervention, and obtaining the latest anomaly detection and intervention strategy basis.

[0014] Furthermore, obtain the latest information on anomaly detection and intervention strategies, specifically including: By acquiring real-time data from the monitoring system and analyzing its fluctuation characteristics, a preliminary data change trend is obtained. Combined with a pre-established rule base update mechanism, the relevant rule content is adjusted to determine the updated rule set. When there is a deviation between the updated rule set and the current model parameters, the direction and magnitude of parameter adjustment are obtained by comparing the differences between historical data and real-time data. The adjusted parameters are then used to update the core model in the data collaborative analysis framework, and the support vector machine algorithm is used to classify the data and determine the distribution of outliers. Based on the distribution of anomalies, a corresponding draft intervention strategy is generated. Combined with historical intervention records in the analysis framework, an optimized strategy plan is obtained. Finally, the basis for anomaly detection is generated, and the priority order of strategy execution is determined by combining real-time data change characteristics. Based on the priority order of strategy execution meeting the preset threshold conditions, the strategy plan is transmitted to the execution module through the information processing stage to obtain execution feedback data.

[0015] The beneficial effects of this invention are as follows: By deploying a multi-source heterogeneous sensor network, comprehensive multi-dimensional data from the production process is collected, forming a comprehensive set of production process information. This effectively eliminates the data blind spots caused by insufficient coverage of a single sensor in traditional monitoring systems, enabling production managers to grasp every detail of the production process in real time, identify potential problems in advance and optimize them, thereby significantly improving the transparency and predictability of the production process and laying a solid foundation for improving production efficiency and product quality. A distributed data collaborative analysis framework was constructed, which combined federated learning algorithms to achieve efficient sharing and joint modeling of multi-node data while protecting data privacy. This solved the problem of data collaborative processing in multi-device environments, enabling the effective integration and utilization of data between different devices. It revealed the potential relationships hidden in the data, provided a global perspective for optimizing the production process, and enabled decisions based on global data to more accurately adjust production parameters and optimize the production process, thereby significantly improving production efficiency and product quality and promoting the intelligent upgrading of the production process. By monitoring the changing trends of key variables in real time and conducting in-depth analysis of historical and real-time data, abnormal patterns can be quickly identified and alarm signals generated, enabling precise intervention. This solves the problem that traditional monitoring systems cannot respond in real time and handle anomalies accurately. This real-time response and precise intervention mechanism can promptly detect and handle abnormal situations in the production process, reduce quality fluctuations and efficiency losses, significantly improve the stability and reliability of the production process, and provide strong support for improving production efficiency and ensuring product quality. Attached Figure Description

[0016] To better understand and implement this application, the technical solution is described in detail below with reference to the accompanying drawings.

[0017] Figure 1 A flowchart illustrating the tracking method for fully automated process monitoring of lithium battery cathode material production provided in this application; Figure 2 A schematic diagram illustrating the process for obtaining cross-device data correlation features using the tracking method for fully automated process monitoring of lithium battery cathode material production provided in this application; Figure 3 This is a flowchart illustrating the process for determining abnormal modes in the tracking method for fully automated process monitoring of lithium battery cathode material production provided in this application. Detailed Implementation

[0018] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, exemplary embodiments will be described in detail below, examples of which are illustrated in the accompanying drawings. In the following description relating to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of methods and systems consistent with some aspects of this application as detailed in the appended claims.

[0019] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. As used in this application and the appended claims, the singular forms "a," "an," "the," and "the" are intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0020] The following detailed description of the specific implementation methods, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided in detail.

[0021] Please see Figures 1-3 This embodiment provides a tracking method based on fully automated process monitoring of lithium battery cathode material production, including the following steps: S1. By deploying a multi-source heterogeneous sensor network, and considering the complex characteristics of the production environment, various types of sensing devices are arranged at key process nodes to collect environmental parameters, equipment status and material characteristic data from multiple dimensions, thereby obtaining a comprehensive set of production process information. Furthermore, a comprehensive set of production process information is obtained, specifically including: By deploying a multi-source heterogeneous sensor network, diverse data from the production environment are acquired. Preprocessing techniques are used to clean and unify the collected environmental parameters, equipment status, and material characteristics data to obtain a standardized dataset. When certain environmental parameters in the standardized dataset exceed the preset threshold range, the anomaly detection mechanism is triggered to determine the location and type of the abnormal data points. Through the analysis of the abnormal data points, the trend of equipment status changes related to the process node is obtained to determine whether there are potential fault risks. Based on the trend of equipment status changes, the support vector machine algorithm is used to predict the operational stability of key process nodes and obtain stability assessment results. When the stability assessment results show that the operation is unstable, correlation analysis is performed on the material characteristic data to determine the main factors affecting stability. By continuously monitoring key factors and obtaining real-time feedback data, the acquisition frequency of the sensor network can be adjusted to optimize the update efficiency of comprehensive process information.

[0022] Specifically, by deploying a multi-source heterogeneous sensor network and arranging various sensing devices at key process nodes, data is collected from multiple dimensions to form a comprehensive set of production process information. This solves the problems of data blind spots and insufficient coverage of single sensors in traditional monitoring systems. Data is standardized through preprocessing techniques, and potential fault risks are promptly identified using anomaly detection mechanisms. Support vector machine algorithms are combined to predict the stability of key process nodes, identify the main factors affecting stability, and optimize information update efficiency by dynamically adjusting sensor acquisition frequency. This technology significantly improves the transparency, predictability, and operational stability of the production process, providing efficient and accurate monitoring and optimization support for lithium battery cathode material production.

[0023] S2. Based on a comprehensive set of production process information, data fusion technology is used to preprocess multi-source data, eliminate missing values ​​and noise interference caused by data blind spots, and determine a unified standardized data format for subsequent analysis.

[0024] Furthermore, a unified standardized data format is determined, specifically including: By collecting data from multiple sources, we obtain the original dataset containing missing values ​​and noise interference. When the original dataset contains missing values, we use the K-nearest neighbor algorithm to interpolate and obtain the completed dataset. The wavelet transform method is used to filter noise interference to obtain a denoised dataset. The attribute features of each data source are obtained, the standardized data format is determined, and a formatted dataset is obtained. When the attributes of a formatted dataset do not conform to a unified format, field mapping technology is used to convert them to obtain a unified dataset. Then, principal component analysis is used to extract key features to obtain the dataset for analysis preparation.

[0025] Specifically, by using data fusion technology to preprocess multi-source data, the problems of missing values ​​and noise interference caused by data blind spots in traditional monitoring systems are solved. This method not only effectively eliminates missing values ​​and noise in the data, but also ensures the consistency and availability of the data, providing a solid foundation for subsequent in-depth analysis and modeling, and significantly improving the accuracy and efficiency of production process data processing.

[0026] S3. For standardized data formats, construct a distributed data collaborative analysis framework. On the basis of protecting data privacy, achieve efficient sharing and joint modeling of multi-node data through federated learning algorithms to obtain cross-device data correlation features. Furthermore, obtain cross-device data association characteristics, specifically including: S31. By constructing a distributed framework, preliminary data structuring results are obtained. Federated learning algorithm is used to perform distributed training on multi-node data to obtain cross-node model parameter updates. S32. When the model parameters are updated within the preset threshold range, further data collaboration processing is performed on the multi-node data to determine the potential consistency characteristics between the data. S33. Perform feature mining on cross-device data association, determine the distribution pattern of association features, and use joint modeling methods to integrate multi-node data to obtain the feature mapping results of the global model. S34. When the deviation between the feature mapping result and the preset target exceeds the threshold, the data privacy protection mechanism is adjusted to obtain an optimized privacy protection strategy. S35. Through the optimized privacy protection strategy, the efficient sharing mechanism is verified for data transmission to determine the final scope of shared data.

[0027] Specifically, by constructing a distributed data collaborative analysis framework and adopting a federated learning algorithm, efficient sharing and joint modeling of multi-node data were achieved while protecting privacy. This solved the problem of multi-device data collaborative processing in traditional monitoring systems. It not only effectively explored the potential correlations in cross-device data but also optimized the data sharing mechanism, providing strong support for the global optimization of the production process and significantly improving the intelligence level and data utilization efficiency of the production monitoring system.

[0028] S4. Extract the changing trends of key variables in the production process from the cross-device data association features, determine whether there are abnormal patterns that deviate from the preset threshold, and generate a preliminary abnormal alarm signal when an abnormal pattern is detected. Furthermore, the changing trends of key variables in the production process are extracted to determine whether there are any abnormal patterns deviating from preset thresholds, specifically including: S41. Obtain multi-source information from cross-device data, integrate and process data streams from different devices, preprocess the data using a unified formatting method to obtain a structured dataset, extract correlation features, use correlation analysis methods to identify key variables in the production process, and determine the distribution characteristics of key variables. S42. Based on the distribution characteristics of key variables, analyze the changing trends, model the historical data of the variables using time series analysis methods, and obtain trend prediction results. When the trend prediction results deviate from the preset threshold, they are judged as potential abnormal patterns. Combine historical data and current data for multi-dimensional comparison to determine the significance of the abnormal patterns. S43. When the significance of an abnormal pattern exceeds a predetermined standard, the anomaly detection mechanism is triggered, generating a preliminary alarm signal. This signal is transmitted to the monitoring module through the system's internal channel. Based on the preliminary alarm signal, the module obtains contextual data from the production process, performs secondary verification on the real-time status of key variables, determines the accuracy of the alarm signal, and then automatically records the relevant characteristics of the abnormal pattern through the second-verified alarm signal. This data is then stored in the anomaly log database for subsequent analysis modules to access.

[0029] Specifically, by extracting the changing trends of key variables from cross-device data and determining whether they deviate from preset thresholds, the problem of difficulty in timely detection and accurate identification of abnormal patterns in the production process is effectively solved. This enables real-time monitoring and precise early warning of the production process, significantly improving the timeliness and accuracy of anomaly handling and enhancing the stability and reliability of production.

[0030] S5. For the initial abnormal alarm signal, combine the correlation characteristics of historical data and real-time data, and use the pre-established abnormal classification model to conduct in-depth analysis to determine the specific type and scope of impact of the abnormality.

[0031] Furthermore, the specific type and scope of the anomaly are determined, including: The system acquires preliminary anomaly alarm signals, extracts real-time data streams from multiple sensor nodes through a data acquisition system to obtain a preliminary anomaly signal set, and uses data fusion technology to perform feature association between the preliminary anomaly signal set and historical data. The system then uses principal component analysis to extract key features to obtain a feature association dataset. When the feature values ​​in the feature association dataset exceed the preset threshold, a pre-established random forest model is used to classify anomalies, determine the anomaly type, and partition the real-time data stream using cluster analysis to determine the scope of anomaly impact and obtain the impact range dataset. By employing time series analysis, the anomaly development trend is predicted for the anomaly types in the affected data set. The anomaly trend prediction results are then obtained, and the parameters of the anomaly classification model are updated to obtain an optimized classification model. By re-analyzing new anomaly signals in the real-time data stream, a more accurate anomaly type and affected range are determined.

[0032] Specifically, by combining preliminary anomaly alarm signals, historical data, and real-time data correlation features, and using anomaly classification models for in-depth analysis, the system effectively solves the problems of inaccurate anomaly type identification and impact range assessment in traditional monitoring systems. Through data fusion technology, key features are extracted and correlated with historical and real-time data. Anomalies are then classified using a random forest model to determine their types, and cluster analysis is used to assess their impact range. Furthermore, time series analysis is combined to predict anomaly development trends, and the classification model parameters are dynamically updated. This enables accurate anomaly identification and precise impact range assessment, providing strong support for precise intervention and optimization of the production process and significantly improving the reliability and efficiency of the production system.

[0033] S6. Based on the specific type and scope of the anomaly, call the preset process parameter adjustment rule library, obtain the corresponding intervention strategy configuration, and generate targeted control instructions.

[0034] Furthermore, targeted regulatory instructions are generated, specifically including: By extracting key features from the data on anomaly types and impact ranges, the specific classification and action boundaries of the anomalies are determined, relevant entries in the preset process parameter adjustment rule library are obtained, and a matching set of adjustment rules is obtained. A logical comparison method is used to determine the degree of fit between the rule and the current abnormal characteristics. When the degree of fit is higher than the preset threshold, the corresponding intervention strategy configuration is determined. Through the intervention strategy configuration, a preliminary draft of the control instruction is generated, and the parameter adjustment direction and magnitude information contained therein are obtained. Based on the parameter adjustment direction and magnitude information in the draft control instructions, and combined with historical data on anomaly handling, the instructions are optimized using a support vector machine model to obtain the adjusted control instructions. The system acquires real-time process parameter status data. When there is a discrepancy between the status data and the instruction content, it uses an iterative comparison method to determine the final targeted control instruction. Combined with the boundary data of the impact range, it generates an executable instruction sequence to complete the closed-loop control for anomaly handling.

[0035] Specifically, by calling a preset process parameter adjustment rule library, targeted control instructions are generated based on the specific type and scope of impact of the anomaly. This solves the problem of the lack of precision and adaptability of anomaly intervention strategies in traditional monitoring systems. By extracting key features of the anomaly and matching them with adjustment rules, intervention strategies are determined through logical comparison and model optimization, generating precise control instructions. Combined with real-time data, closed-loop control is achieved, enabling rapid and precise intervention in the production process and significantly improving the stability and responsiveness of the production system.

[0036] S7. By generating targeted control commands, the production equipment is driven to perform parameter adjustment operations, while real-time data changes after intervention are recorded to determine whether the control has achieved the expected stable state.

[0037] Furthermore, determining whether the regulation has achieved the expected stable state specifically includes: The control system generates control commands to drive production equipment to adjust parameters, obtain equipment response data, acquire real-time data from the equipment response data, and perform structured processing using data acquisition technology to obtain a formatted dataset. A recording mechanism is used to store data changes. When the data changes exceed a preset threshold, a feedback loop is triggered to obtain an adjustment signal. The control system updates the control instructions and drives the production equipment to perform parameter adjustments again, thus obtaining new equipment response data. Extract data change trends from the response data of the new equipment, use result analysis technology to calculate stability indicators, and determine whether a stable state has been reached. If the stability indicators do not meet the preset standards, generate new control commands through feedback loops, repeat the adjustment process, and obtain the final stable data. Based on the final stable data, result analysis technology is used to generate control logs, which are stored in the recording mechanism to obtain business execution records.

[0038] Specifically, by generating targeted control commands to drive production equipment to adjust parameters and recording data changes after intervention in real time, the system uses feedback mechanisms and stability indicators to determine whether the control has reached the expected stable state. This process solves the problem of difficulty in real-time verification and dynamic adjustment of control effects in traditional production systems, realizes closed-loop control and refined management of the production process, ensures rapid response and stable operation of the production system, and significantly improves production efficiency and product quality.

[0039] Furthermore, after determining whether the regulation has reached the expected stable state, it also includes: updating the model parameters and rule base in the data collaborative analysis framework based on real-time data changes after the intervention, and obtaining the latest anomaly detection and intervention strategy basis.

[0040] Furthermore, obtain the latest information on anomaly detection and intervention strategies, specifically including: By acquiring real-time data from the monitoring system and analyzing its fluctuation characteristics, a preliminary data change trend is obtained. Combined with a pre-established rule base update mechanism, the relevant rule content is adjusted to determine the updated rule set. When there is a deviation between the updated rule set and the current model parameters, the direction and magnitude of parameter adjustment are obtained by comparing the differences between historical data and real-time data. The adjusted parameters are then used to update the core model in the data collaborative analysis framework, and the support vector machine algorithm is used to classify the data and determine the distribution of outliers. Based on the distribution of anomalies, a corresponding draft intervention strategy is generated. Combined with historical intervention records in the analysis framework, an optimized strategy plan is obtained. Finally, the basis for anomaly detection is generated, and the priority order of strategy execution is determined by combining real-time data change characteristics. Based on the priority order of strategy execution meeting the preset threshold conditions, the strategy plan is transmitted to the execution module through the information processing stage to obtain execution feedback data.

[0041] Specifically, by updating the model parameters and rule base in the data collaborative analysis framework with real-time data, the latest anomaly detection and intervention strategies are obtained. This process solves the problem that anomaly detection and intervention strategies in traditional systems are difficult to dynamically adjust based on real-time data. By analyzing the fluctuation characteristics of real-time data and adjusting the rule base content, the model parameters are updated in combination with the differences between historical and real-time data to generate optimized intervention strategies, which are then executed according to priority. This achieves dynamic optimization of anomaly detection and intervention strategies, improves the system's adaptability and intelligence level, and ensures the continuous, stable, and efficient operation of the production process.

[0042] The basis for the latest anomaly detection and intervention strategies includes: continuously optimizing the acquisition strategy of multi-source data perception, adjusting the monitoring focus of sensor networks, and obtaining a more accurate set of production process information.

[0043] The more accurate set of production process information obtained specifically includes: Anomaly detection algorithms are used to extract anomaly features from multi-source data. Support vector machine algorithms are used to determine the anomaly type and obtain anomaly classification results. The acquisition strategy is adjusted based on the anomaly classification results. When anomaly features appear in the data of a specific sensor, the acquisition frequency of that sensor is increased first to determine the optimized acquisition strategy. The sensor network configuration is updated using an optimized acquisition strategy. High-precision production process data is obtained by dynamically adjusting the monitoring focus, resulting in a real-time sensing data set. Key production process indicators are extracted from the real-time sensing data set, and cluster analysis algorithms are used to classify data features, determine the production process status, and obtain status classification results. Based on the state classification results, it is determined whether the production process state deviates from the preset threshold. If it exceeds the threshold, an intervention strategy is triggered. By adjusting the sensor network parameters to optimize data acquisition, an updated information set is obtained. Then, the production process trend is analyzed, and a time series analysis algorithm is used to predict the probability of anomalies and identify potential risk points. The sensor network monitoring focus is adjusted based on potential risk points. When the predicted probability of an anomaly exceeds a preset threshold, the weight of relevant sensor data acquisition is increased to obtain a high-precision set of production process information.

[0044] By continuously optimizing the acquisition strategy of multi-source data sensing and dynamically adjusting the monitoring focus of the sensor network, a more accurate set of production process information is obtained. This technical approach solves the problem of fixed data acquisition strategies and difficulty in dynamically adjusting to anomalies in traditional monitoring systems. By extracting anomaly features from multi-source data and determining the anomaly type, the system can dynamically adjust the acquisition strategy based on the anomaly features, prioritizing the acquisition frequency of key sensors. Furthermore, through methods such as cluster analysis and time series analysis, the system monitors the production process status in real time and predicts potential risk points, dynamically optimizing sensor network parameters to ensure high-precision acquisition of production process data. Ultimately, the system can more accurately capture key information in the production process, identify and address potential risks in advance, significantly improve the monitoring accuracy and risk warning capabilities of the production process, and provide strong support for improving production efficiency and product quality.

[0045] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A tracking method based on fully automated process monitoring of lithium battery cathode material production process, characterized in that: Includes the following steps: By deploying a multi-source heterogeneous sensor network and arranging various types of sensing devices at key process nodes, environmental parameters, equipment status and material characteristics data are collected from multiple dimensions to obtain a comprehensive set of production process information. Data fusion technology is used to preprocess multi-source data to eliminate missing values ​​and noise interference caused by data blind spots, and to determine a unified standardized data format. A distributed data collaborative analysis framework is constructed. While protecting data privacy, it uses federated learning algorithms to achieve efficient sharing and joint modeling of multi-node data and obtain cross-device data correlation features. From the data correlation characteristics across devices, extract the changing trends of key variables in the production process, determine whether there are abnormal patterns that deviate from the preset threshold, and generate a preliminary abnormal alarm signal when an abnormal pattern is detected. By combining the correlation characteristics of historical and real-time data, a pre-established anomaly classification model is used for in-depth analysis to determine the specific type and scope of impact of the anomaly. Based on the specific type and scope of the anomaly, the preset process parameter adjustment rule library is invoked to obtain the corresponding intervention strategy configuration and generate targeted control instructions. By generating targeted control commands, the production equipment is driven to perform parameter adjustment operations, while real-time data changes after intervention are recorded to determine whether the control has achieved the expected stable state.

2. The tracking method based on fully automated process monitoring of lithium battery cathode material production process according to claim 1, characterized in that: Obtain a comprehensive set of production process information, specifically including: By deploying a multi-source heterogeneous sensor network, diverse data from the production environment are acquired. Preprocessing techniques are used to clean and unify the collected environmental parameters, equipment status, and material characteristics data to obtain a standardized dataset. When certain environmental parameters in the standardized dataset exceed the preset threshold range, the anomaly detection mechanism is triggered to determine the location and type of the abnormal data points. Through the analysis of the abnormal data points, the trend of equipment status changes related to the process node is obtained to determine whether there are potential fault risks. Based on the trend of equipment status changes, the support vector machine algorithm is used to predict the operational stability of key process nodes and obtain stability assessment results. When the stability assessment results show that the operation is unstable, correlation analysis is performed on the material characteristic data to determine the main factors affecting stability. By continuously monitoring key factors and obtaining real-time feedback data, the acquisition frequency of the sensor network can be adjusted to optimize the update efficiency of comprehensive process information.

3. The tracking method based on fully automated process monitoring of lithium battery cathode material production process according to claim 1, characterized in that: Establish a unified, standardized data format, specifically including: By collecting data from multiple sources, we obtain the original dataset containing missing values ​​and noise interference. When the original dataset contains missing values, we use the K-nearest neighbor algorithm to interpolate and obtain the completed dataset. The wavelet transform method is used to filter noise interference to obtain a denoised dataset. The attribute features of each data source are obtained, the standardized data format is determined, and a formatted dataset is obtained. When the attributes of a formatted dataset do not conform to a unified format, field mapping technology is used to convert them to obtain a unified dataset. Then, principal component analysis is used to extract key features to obtain the dataset for analysis preparation.

4. The tracking method based on fully automated process monitoring of lithium battery cathode material production process according to claim 1, characterized in that: Obtain cross-device data association characteristics, specifically including: By constructing a distributed framework, preliminary data structuring results are obtained. Federated learning algorithms are then used to perform distributed training on multi-node data to obtain cross-node model parameter updates. When the model parameters are updated within the preset threshold range, further data collaboration processing is performed on the multi-node data to determine the potential consistency characteristics between the data. Feature mining is performed on cross-device data associations to determine the distribution pattern of association features. A joint modeling method is used to integrate multi-node data and obtain the feature mapping results of the global model. When the deviation between the feature mapping result and the preset target exceeds the threshold, the data privacy protection mechanism is adjusted to obtain an optimized privacy protection strategy. By using an optimized privacy protection strategy, the efficient sharing mechanism is verified to determine the final scope of shared data.

5. The tracking method based on fully automated process monitoring of lithium battery cathode material production process according to claim 1, characterized in that: Extracting the changing trends of key variables in the production process to determine whether there are any abnormal patterns deviating from preset thresholds, specifically including: This involves acquiring multi-source information from cross-device data, integrating and processing data streams from different devices, preprocessing the data using a unified formatting method, extracting correlation features, and using correlation analysis to identify key variables in the production process and determine the distribution characteristics of these key variables. Based on the distribution characteristics of key variables, the changing trends are analyzed. The historical data of the variables are modeled using time series analysis methods to obtain trend prediction results. When the trend prediction results deviate from the preset threshold, they are identified as potential abnormal patterns. The significance of the abnormal patterns is judged by combining historical data and current data in multiple dimensions. When the significance of an abnormal pattern exceeds a predetermined standard, the anomaly detection mechanism is triggered, generating an initial alarm signal. This signal is transmitted to the monitoring module through the system's internal channel, where contextual data from the production process is obtained. The real-time status of key variables is then verified to confirm the accuracy of the alarm signal. Finally, the relevant characteristics of the abnormal pattern are automatically recorded and stored in the anomaly log database.

6. The tracking method based on fully automated process monitoring of lithium battery cathode material production process according to claim 1, characterized in that: Determine the specific type and scope of the anomaly, including: The system acquires preliminary anomaly alarm signals, extracts real-time data streams from multiple sensor nodes through a data acquisition system to obtain a preliminary anomaly signal set, and uses data fusion technology to perform feature association between the preliminary anomaly signal set and historical data. The system then uses principal component analysis to extract key features to obtain a feature association dataset. When the feature values ​​in the feature association dataset exceed the preset threshold, a pre-established random forest model is used to classify anomalies, determine the anomaly type, and partition the real-time data stream using cluster analysis to determine the scope of anomaly impact and obtain the impact range dataset. By employing time series analysis, the abnormality types in the affected data set are analyzed to predict the abnormality development trend. The abnormality trend prediction results are then obtained, and the parameters of the abnormality classification model are updated to obtain an optimized classification model. By re-analyzing new abnormal signals in the real-time data stream, a more accurate abnormality type and affected range are determined.

7. The tracking method based on fully automated process monitoring of lithium battery cathode material production process according to claim 1, characterized in that: Generate targeted control instructions, specifically including: By extracting key features from the data on anomaly types and impact ranges, the specific classification and action boundaries of the anomalies are determined, relevant entries in the preset process parameter adjustment rule library are obtained, and a matching set of adjustment rules is obtained. A logical comparison method is used to determine the degree of fit between the rule and the current abnormal characteristics. When the degree of fit is higher than the preset threshold, the corresponding intervention strategy configuration is determined. Through the intervention strategy configuration, a preliminary draft of the control instruction is generated, and the parameter adjustment direction and magnitude information contained therein are obtained. Based on the parameter adjustment direction and magnitude information in the draft control instructions, and combined with historical data on anomaly handling, the instructions are optimized using a support vector machine model to obtain the adjusted control instructions. The system acquires real-time process parameter status data. When there is a discrepancy between the status data and the instruction content, it uses an iterative comparison method to determine the final targeted control instruction. Combined with the boundary data of the impact range, it generates an executable instruction sequence to complete the closed-loop control for anomaly handling.

8. The tracking method based on fully automated process monitoring of lithium battery cathode material production process according to claim 1, characterized in that: To determine whether the regulation has achieved the expected stable state, the following are specific criteria: The control system generates control commands to drive production equipment to adjust parameters, obtain equipment response data, acquire real-time data from the equipment response data, and perform structured processing using data acquisition technology to obtain a formatted dataset. A recording mechanism is used to store data changes. When the data changes exceed a preset threshold, a feedback loop is triggered to obtain an adjustment signal. The control system updates the control instructions and drives the production equipment to perform parameter adjustments again, thus obtaining new equipment response data. Extract data change trends from the response data of the new equipment, use result analysis technology to calculate stability indicators, and determine whether a stable state has been reached. If the stability indicators do not meet the preset standards, generate new control commands through feedback loops, repeat the adjustment process, and obtain the final stable data. Based on the final stable data, result analysis technology is used to generate control logs, which are stored in the recording mechanism to obtain business execution records.

9. The tracking method based on fully automated process monitoring of lithium battery cathode material production process according to claim 8, characterized in that: After determining whether the regulation has reached the expected stable state, it also includes: updating the model parameters and rule base in the data collaborative analysis framework based on real-time data changes after the intervention, and obtaining the latest anomaly detection and intervention strategy basis.

10. The tracking method based on fully automated process monitoring of lithium battery cathode material production process according to claim 9, characterized in that: Obtain the latest information on anomaly detection and intervention strategies, specifically including: By acquiring real-time data from the monitoring system and analyzing its fluctuation characteristics, a preliminary data change trend is obtained. Combined with a pre-established rule base update mechanism, the relevant rule content is adjusted to determine the updated rule set. When there is a deviation between the updated rule set and the current model parameters, the direction and magnitude of parameter adjustment are obtained by comparing the differences between historical data and real-time data. The adjusted parameters are then used to update the core model in the data collaborative analysis framework, and the support vector machine algorithm is used to classify the data and determine the distribution of outliers. Based on the distribution of anomalies, a corresponding draft intervention strategy is generated. Combined with historical intervention records in the analysis framework, an optimized strategy plan is obtained. Finally, the basis for anomaly detection is generated, and the priority order of strategy execution is determined by combining real-time data change characteristics. Based on the priority order of strategy execution meeting the preset threshold conditions, the strategy plan is transmitted to the execution module through the information processing stage to obtain execution feedback data.

Images