Data interaction method based on intelligent control platform

Abstract

The invention relates to the technical field of communication interaction, in particular to a data interaction method based on an intelligent control platform. The method comprises the following steps: acquiring multi-dimensional data of production equipment, performing anomaly marking and format unification processing, generating a production line data stream, performing long and short term dependency analysis on the data stream, and constructing a time-space association network, thereby realizing prediction of an anomaly situation, generating an anomaly intelligent analysis model by using a multi-modal fusion modeling technology, and performing data analysis on the anomaly intelligent analysis model. And through the model, abnormity traceability is carried out, and abnormity reasons are clarified. According to the method, a corresponding correction strategy is constructed based on an abnormal reason, tree-shaped interaction sharing is performed through an intelligent control platform, a sharing correction instruction is formed, abnormal correction processing is performed on a production line according to the sharing instruction, feedback is collected through the intelligent control platform, an effective data interaction mechanism is formed, and the reliability and efficiency of the production line are improved. The data interaction method based on the intelligent control platform is more accurate and more comprehensive.

Description

Technical Field

[0001] The present invention relates to the field of communication interaction technology, and in particular to a data interaction method based on an intelligent control platform. Background Art

[0002] In modern manufacturing, the efficient operation of production equipment is crucial to improving production efficiency and reducing costs. However, with the increasing complexity of the production process, various types of equipment will experience abnormalities during operation, which will not only affect the production progress, but also cause equipment damage and safety hazards. At present, most manufacturing companies still rely on manual inspections and experience judgments when facing equipment abnormalities, and lack a systematic abnormality monitoring and early warning mechanism. This traditional management method is powerless when processing multi-dimensional data and it is difficult to respond to equipment status changes in real time. In addition, existing anomaly detection methods often focus on the analysis of a single data source and fail to make full use of the multi-dimensional data of the equipment for comprehensive evaluation. This leads to incomplete and inaccurate information when predicting equipment abnormalities, which affects the reliability of the prediction results. The lack of effective multimodal fusion modeling technology makes the construction of intelligent analysis models complicated and cannot fully reflect the actual operating conditions and potential risks of the production line. It makes full use of the advantages of multi-dimensional data analysis and intelligent model construction, and significantly improves the abnormal monitoring, prediction and processing capabilities of the production line. Through real-time data acquisition and abnormality marking, combined with deep spatiotemporal analysis and multimodal fusion, an efficient and accurate abnormality management system has been formed. The optimized information sharing and feedback mechanism not only promotes the collaborative work of all links, but also lays a solid foundation for the future development of intelligent production. Summary of the invention

[0003] Based on this, it is necessary to provide a data interaction method based on an intelligent control platform to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a data interaction method based on an intelligent control platform includes the following steps:

[0005] Step S1: Acquire multi-dimensional data of production equipment; perform data anomaly marking on the multi-dimensional data of production equipment to generate multi-dimensional data of abnormally marked equipment; perform format homogenization processing on the multi-dimensional data of abnormally marked equipment to obtain a production line data stream;

[0006] Step S2: Perform long-term and short-term dependency analysis on the production line data stream to generate a spatiotemporal correlation network; perform abnormal situation prediction on the production line data stream based on the spatiotemporal correlation network to obtain predicted production line abnormal data;

[0007] Step S3: Perform multimodal fusion modeling on the predicted production line abnormality data to obtain an abnormality intelligent analysis model; perform abnormality tracing on the predicted production line abnormality data according to the abnormality intelligent analysis model to generate the cause of the production line abnormality;

[0008] Step S4: construct a correction strategy for the abnormal cause of the production line and generate an abnormal correction strategy; share the abnormal correction strategy in a tree-like interactive manner through the intelligent control platform to obtain a shared correction instruction;

[0009] Step S5: Perform abnormal correction processing on the production line according to the shared correction instruction to generate an abnormal correction result; collect instruction feedback on the abnormal correction result through the intelligent control platform to execute data interaction based on the intelligent control platform.

[0010] The present invention forms a complete data stream by acquiring multi-dimensional data of production equipment and marking abnormalities, which can timely discover potential equipment abnormalities and improve the monitoring capability of the production line. The construction of long-term and short-term dependency analysis and spatiotemporal correlation network makes the prediction of abnormal situation of production line data stream more accurate, makes full use of historical and real-time data, and enhances the effectiveness of abnormal detection. Through multimodal fusion technology, data from different sources can be fully integrated to improve the accuracy of abnormal intelligent analysis model, thereby improving the reliability of abnormal traceability and ensuring that problems are identified in time. According to the generated abnormal causes of production lines, effective correction strategies can be constructed to achieve targeted solutions, thereby shortening the time of abnormal processing and improving the availability of equipment. The tree-shaped interactive sharing of correction strategies is realized through the intelligent control platform, which enhances the information flow between production line operators and control systems and ensures that instructions are communicated in time and effectively executed. The collection of instruction feedback on abnormal correction results can continuously optimize the correction process, accumulate data experience, provide support for subsequent production process improvements, and promote the intelligent upgrade of production lines.

[0011] Preferably, step S1 comprises the following steps:

[0012] Step S11: Acquire multi-dimensional data of production equipment;

[0013] Step S12: removing random noise from the multi-dimensional data of the production equipment to obtain denoised multi-dimensional data of the equipment;

[0014] Step S13: marking data anomalies on the denoising device multidimensional data to generate abnormality marked device multidimensional data;

[0015] Step S14: Analyze the statistical characteristics of the multi-dimensional data of the abnormal marking device to generate data statistical characteristics;

[0016] Step S15: Perform format homogenization processing on the multi-dimensional data of the abnormal marking equipment according to the data statistical characteristics to obtain the production line data flow.

[0017] The present invention is achieved by acquiring multi-dimensional data of production equipment through a variety of sensors and monitoring systems, ensuring that the operating status of the equipment can be captured comprehensively and in real time, providing accurate basic data and promoting more accurate decision-making. Random noise rejection technology effectively filters out irrelevant signals generated by environmental or equipment interference, improves data quality, ensures the accuracy of subsequent analysis, and reduces the risk of misjudgment and missed judgment. The data anomaly marking mechanism identifies potential abnormal conditions through algorithms, enhances sensitivity to signs of faults, and achieves timely warnings, thereby reducing production losses and safety hazards caused by equipment failures. Statistical feature analysis extracts key performance indicators during equipment operation, such as mean, standard deviation, etc., to help identify the difference between normal and abnormal behavior, making fault prediction and diagnosis more accurate. Format homogenization processing unifies multi-source heterogeneous data into a standard format, ensures the integration and consistency of data streams, improves the efficiency of data processing and analysis, and provides a basis for subsequent intelligent analysis.

[0018] Preferably, step S2 comprises the following steps:

[0019] Step S21: performing time sequence conversion on the production line data stream to obtain production time sequence data of the production line;

[0020] Step S22: performing long-term and short-term dependency analysis on the production time series data of the assembly line to generate a spatiotemporal correlation network;

[0021] Step S23: performing abnormal propagation evolution on the spatiotemporal correlation network according to the multi-dimensional data of the abnormal marking device to generate simulated abnormal diffusion data;

[0022] Step S24: predicting the abnormal situation of the production line data flow according to the simulated abnormal diffusion data to obtain predicted production line abnormal data.

[0023] The present invention converts the production line data stream into assembly line production time series data through time series conversion technology, ensures the continuity and consistency of the data in the time dimension, provides a basis for subsequent analysis, and improves the effectiveness of data processing. Long-term and short-term dependency analysis reveals the relationship between the medium-term and short-term trends of the data. By constructing a spatiotemporal correlation network, it captures the dynamic changes in the production process and promotes a comprehensive understanding of potential anomalies. The abnormal propagation evolution simulation analyzes the multi-dimensional data of the abnormal marking equipment, identifies the diffusion path of the anomaly on the production line, provides a basis for prediction and intervention, and enhances the grasp of the system's vulnerabilities. The abnormal situation prediction based on the simulated abnormal diffusion data can timely identify the potential failure risks of the production line, reduce the possibility of production stagnation and loss, and ensure production efficiency and safety. The comprehensive analysis results provide real-time decision support for the intelligent control platform, making production management more flexible and efficient, and promoting the optimization and improvement of equipment and production processes.

[0024] Preferably, step S22 comprises the following steps:

[0025] Step S221: segmenting the production time series data of the assembly line into long and short time segments based on a preset segment duration threshold, generating long-term data segments and short-term data segments;

[0026] Step S222: Perform long short-term memory network analysis on the long-term data segments to obtain long-term dependency features;

[0027] Step S223: performing gated cycle unit analysis on the short-term data segment to generate short-term fluctuation features;

[0028] Step S224: performing correlation influence mining on the long-term dependency features and the short-term fluctuation features to obtain correlation mutual influence data;

[0029] Step S225: construct a correlation network for the long-term dependency characteristics and the short-term fluctuation characteristics according to the correlation and mutual influence data to generate a spatiotemporal correlation network.

[0030] The present invention divides the production time series data of the assembly line into long-term data segments and short-term data segments by setting the segment duration threshold, which facilitates targeted analysis of data of different time scales and improves the flexibility and accuracy of data processing. The analysis of long-term data segments by the long short-term memory network effectively extracts long-term dependency features, identifies the deep-level laws and trends in the production process, and enhances the understanding of the long-term performance of the system. The analysis of short-term data segments by the gated recurrent unit captures short-term fluctuation characteristics, reflects the dynamic changes of the production line in real time, and provides accurate short-term information for anomaly detection and early warning. Correlation influence mining establishes a deep connection between long-term dependency features and short-term fluctuation features, reveals the mutual influence mechanism in the production process, and promotes a comprehensive understanding of the system operation status. By constructing a spatiotemporal correlation network, a correlation structure between long-term and short-term features is formed, which provides a scientific basis for intelligent decision-making and promotes the intelligence and optimization of production line management.

[0031] Preferably, step S3 comprises the following steps:

[0032] Step S31: Perform multi-layer perceptron network mapping on the predicted production line abnormality data to generate an abnormality classification network;

[0033] Step S32: Perform multimodal fusion modeling on the predicted production line abnormal data according to the abnormal classification network to obtain an abnormal intelligent analysis model;

[0034] Step S33: performing association rule mining on the abnormal classification network and the abnormal data of the predicted production line according to the abnormal intelligent analysis model to obtain an abnormal association rule set;

[0035] Step S34: Based on the abnormal association rule set, the abnormality data of the predicted production line is traced to the source, and the cause of the production line abnormality is generated.

[0036] The present invention uses multi-layer perceptron network mapping to deeply process the data of predicted production line anomalies, improve the accuracy of anomaly classification, and ensure efficient identification of different types of anomalies in complex environments. Multimodal fusion modeling integrates data from different sources to form a comprehensive abnormal intelligent analysis model, enhance the analysis ability of abnormal situations, and provide richer perspectives and information. Association rule mining technology combines the analysis of abnormal classification networks and predicted production line abnormal data to reveal potential abnormal patterns and relationships, and improve the understanding and interpretation of abnormal behaviors. The abnormality tracing function accurately locates the root cause of the abnormality through the abnormal association rule set generated by system analysis, providing a reliable basis for subsequent improvements and preventive measures. The overall process realizes data interaction and feedback on the intelligent control platform, promotes real-time monitoring and response capabilities, and improves the intelligence level and operation efficiency of the production line.

[0037] Preferably, step S34 includes the following steps:

[0038] Step S341: performing hierarchical clustering analysis on the abnormal association rule set to obtain an abnormal rule tree;

[0039] Step S342: embedding the abnormal rule tree into a graph structure to obtain a rule vector space;

[0040] Step S343: extracting abnormal patterns from the predicted production line abnormal data to obtain abnormal feature sequences;

[0041] Step S344: performing multi-scale convolution feature matching on the abnormal feature sequence and the regular vector space to obtain a feature association matrix;

[0042] Step S345: Perform Bayesian causal inference analysis on the feature association matrix to obtain the cause of production line abnormality.

[0043] The present invention organizes the abnormal association rule set through hierarchical clustering analysis, so that the rules with strong correlation are classified together to form a clear abnormal rule tree, thereby improving the structural degree of abnormal identification and processing. Graph structure embedding technology converts the abnormal rule tree into a rule vector space, effectively captures the relationship between the rules, enhances the computability and operability of the rules, and facilitates subsequent analysis. Abnormal pattern extraction identifies key features from the abnormal data of the predicted production line, forms an abnormal feature sequence, refines the understanding of abnormal phenomena, and ensures that more subtle changes are paid attention to. Multi-scale convolutional feature matching realizes a deep comparison of the abnormal feature sequence and the rule vector space, enhances the correlation analysis between features, and provides a more comprehensive abnormal identification capability. Bayesian causal inference analysis performs in-depth reasoning based on the feature association matrix, reveals potential causal relationships, ensures a more accurate understanding of the causes of production line abnormalities, and provides a scientific basis for decision-making.

[0044] Preferably, step S4 comprises the following steps:

[0045] Step S41: constructing a knowledge network for the abnormal causes of the production line to obtain an abnormal knowledge network;

[0046] Step S42: constructing a correction strategy based on the abnormal knowledge network to generate an abnormal correction strategy;

[0047] Step S43: constructing a tree-shaped interactive chain for the production line through the intelligent control platform to obtain a strategy node cluster;

[0048] Step S44: Based on the abnormal correction result, interactive sharing processing is performed on the policy node cluster to obtain a shared correction instruction.

[0049] The present invention forms an abnormal knowledge network by systematically organizing the causes of production line abnormalities through knowledge network construction, promotes the structuring and visualization of information, and enhances the ability to understand and handle abnormalities. The correction strategy is constructed based on the abnormal knowledge network to ensure that the formulated strategy is targeted and effective, improve the scientific nature of abnormal correction, and reduce the potential risk of wrong decisions. The tree-shaped interactive chain construction realizes the efficient organization and management of policy nodes through the intelligent control platform, enhances the efficiency of information transmission, and ensures the rapid application of strategies on the production line. The interactive sharing process optimizes the policy node cluster, improves the flexibility and adaptability of shared correction instructions, enables different production lines to quickly respond to abnormal correction needs, and ensures the continuity of production efficiency.

[0050] Preferably, step S42 includes the following steps:

[0051] Perform graph reasoning on the abnormal knowledge network to obtain a set of potential correction paths;

[0052] Monte Carlo tree search is performed on the potential correction path set to obtain the optimal correction sequence;

[0053] Carry out fuzzy logic control modeling on the optimal correction sequence to obtain an adaptive correction rule base;

[0054] Multi-agent collaborative decision-making is carried out on the adaptive correction rule base to generate anomaly correction strategies.

[0055] The present invention uses graph reasoning to conduct in-depth analysis of the abnormal knowledge network, identify potential correction path sets, and enhance the comprehensive understanding of abnormal handling and strategy selection capabilities. Monte Carlo tree search provides a probabilistic evaluation of potential correction paths to ensure that the optimal correction sequence obtained is efficient and reasonable, reducing the uncertainty in the correction process. Fuzzy logic control modeling combines the adaptive correction rule base with fuzzy logic to improve the flexibility and adaptability of the rules, effectively respond to complex changes in the production environment, and ensure dynamic adjustment of the correction strategy. Multi-agent collaborative decision-making promotes efficient communication and collaboration among various agents. The generated abnormal correction strategy is comprehensive and forward-looking, ensuring rapid response to abnormalities in the production line and optimizing overall production efficiency.

[0056] Preferably, step S43 includes the following steps:

[0057] Identify the machine interaction relationships on the production line and obtain potential interaction pairs;

[0058] Construct a tree-shaped interaction chain for potential interaction pairs to generate a node connection tree structure;

[0059] Map the node connection tree structure to a distributed hash table to obtain a device addressing scheme;

[0060] Perform network coverage processing on the device addressing scheme to obtain the device communication protocol;

[0061] The consensus mechanism is integrated into the device communication protocol to obtain a strategy node cluster.

[0062] The present invention can effectively reveal potential interaction pairs between devices in the production line through machine interaction relationship identification, enhance the understanding of the collaborative relationship between devices, and provide basic data for subsequent optimization. The tree-shaped interaction chain construction presents the interaction relationship between devices in a structured manner, promotes the visualization of information flow, and is conducive to identifying key nodes and optimizing the interaction process. Distributed hash table mapping realizes efficient management of device addressing schemes, significantly improves the speed and accuracy of communication between devices, and reduces potential communication delays. Network coverage processing ensures the comprehensiveness and reliability of communication between devices, enhances the adaptability of the system in complex environments, and improves the stability of the overall system. The integration of the consensus mechanism improves the decision-making efficiency and consistency of the policy node cluster, which helps to achieve efficient coordination and control when multiple devices work together, and optimizes the intelligence level of the production line.

[0063] Preferably, step S5 comprises the following steps:

[0064] Step S51: performing dependency analysis on the shared correction instruction to obtain a task dependency network;

[0065] Step S52: Prioritize the task-dependent networks to obtain a device execution plan;

[0066] Step S53: performing abnormal correction processing on the production line according to the equipment execution plan and generating an abnormal correction result;

[0067] Step S54: collect command feedback on the abnormal correction results through the intelligent control platform to perform data interaction based on the intelligent control platform.

[0068] The present invention conducts in-depth analysis of shared correction instructions through dependency resolution to form a task dependency network, which helps to clarify the relationship between tasks and ensure the systematicness and efficiency of the correction process. Priority sorting optimizes the order of task execution, ensures that key correction tasks are executed first, reduces the time for abnormal correction processing, and improves the response speed. The equipment execution plan is generated based on the task dependency network and priority sorting, ensuring that the abnormal correction processing of the production line is carried out in an orderly manner and reducing conflicts and errors in the correction process. The abnormal correction processing ensures the predictability and reliability of the correction effect by accurately executing the equipment plan, and improves the overall operating efficiency of the production line. Instruction feedback collection provides real-time data for the intelligent control platform, which can dynamically adjust the correction strategy, enhance the adaptability and flexibility of the system, and promote the continuous optimization of intelligent decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 A schematic diagram of the steps of the data interaction method based on the intelligent control platform;

[0070] Figure 2 for Figure 1 Detailed implementation steps of step S2 in the flowchart;

[0071] Figure 3 for Figure 1 Detailed implementation steps of step S3 in FIG.

[0072] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0073] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0074] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0075] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0076] To achieve this, please refer to Figures 1 to 3 , a data interaction method based on an intelligent control platform includes the following steps:

[0077] Step S1: Acquire multi-dimensional data of production equipment; perform data anomaly marking on the multi-dimensional data of production equipment to generate multi-dimensional data of abnormally marked equipment; perform format homogenization processing on the multi-dimensional data of abnormally marked equipment to obtain a production line data stream;

[0078] Step S2: Perform long-term and short-term dependency analysis on the production line data stream to generate a spatiotemporal correlation network; perform abnormal situation prediction on the production line data stream based on the spatiotemporal correlation network to obtain predicted production line abnormal data;

[0079] Step S3: Perform multimodal fusion modeling on the predicted production line abnormality data to obtain an abnormality intelligent analysis model; perform abnormality tracing on the predicted production line abnormality data according to the abnormality intelligent analysis model to generate the cause of the production line abnormality;

[0080] Step S4: construct a correction strategy for the abnormal cause of the production line and generate an abnormal correction strategy; share the abnormal correction strategy in a tree-like interactive manner through the intelligent control platform to obtain a shared correction instruction;

[0081] Step S5: Perform abnormal correction processing on the production line according to the shared correction instruction to generate abnormal correction results; collect instruction feedback on the abnormal correction results through the intelligent control platform to execute data interaction based on the intelligent control platform.

[0082] The present invention forms a complete data stream by acquiring multi-dimensional data of production equipment and marking abnormalities, which can timely discover potential equipment abnormalities and improve the monitoring capability of the production line. The construction of long-term and short-term dependency analysis and spatiotemporal correlation network makes the prediction of abnormal situation of production line data stream more accurate, makes full use of historical and real-time data, and enhances the effectiveness of abnormal detection. Through multimodal fusion technology, data from different sources can be fully integrated to improve the accuracy of abnormal intelligent analysis model, thereby improving the reliability of abnormal traceability and ensuring that problems are identified in time. According to the generated abnormal causes of production lines, effective correction strategies can be constructed to achieve targeted solutions, thereby shortening the time of abnormal processing and improving the availability of equipment. The tree-shaped interactive sharing of correction strategies is realized through the intelligent control platform, which enhances the information flow between production line operators and control systems and ensures that instructions are communicated in time and effectively executed. The collection of instruction feedback on abnormal correction results can continuously optimize the correction process, accumulate data experience, provide support for subsequent production process improvements, and promote the intelligent upgrade of production lines.

[0083] In the embodiment of the present invention, reference Figure 1 The above is a schematic diagram of the steps of the data interaction method based on the intelligent control platform of the present invention. In this example, the data interaction method based on the intelligent control platform includes the following steps:

[0084] Step S1: Acquire multi-dimensional data of production equipment; perform data anomaly marking on the multi-dimensional data of production equipment to generate multi-dimensional data of abnormally marked equipment; perform format homogenization processing on the multi-dimensional data of abnormally marked equipment to obtain a production line data stream;

[0085] In this embodiment, when acquiring multi-dimensional data of production equipment, IoT sensors are used to monitor various parameters of the equipment in real time, including temperature, pressure, rotation speed, vibration, etc., and the data is transmitted to the data storage server via a wireless network. Data anomaly marking is achieved through a machine learning model, which is trained with historical data and can identify parameter values ​​that are beyond the normal range and mark them as abnormal. The generated multi-dimensional data of the abnormal marking equipment must follow a unified JSON format in terms of data format to facilitate subsequent processing. The format homogenization process uses data cleaning tools to process missing values ​​and abnormal values, and finally obtains a formatted production line data stream, which contains the status information and timestamps of each device to form a complete database.

[0086] Step S2: Perform long-term and short-term dependency analysis on the production line data stream to generate a spatiotemporal correlation network; perform abnormal situation prediction on the production line data stream based on the spatiotemporal correlation network to obtain predicted production line abnormal data;

[0087] In this embodiment, when performing long-term and short-term dependency analysis on the production line data stream, a long short-term memory network (LSTM) is used to train the data. This process includes dividing the production line data stream into multiple time series segments, each segment containing data of several time steps. During the training phase, the LSTM model extracts the long-term and short-term dependency features in the time series and generates a spatiotemporal correlation network that can reveal the relationship between devices. Based on the constructed spatiotemporal correlation network, a regression model is applied to predict abnormal situations in the data stream, and the prediction results obtained are abnormal data that will occur. This data will serve as the basis for subsequent analysis to ensure timely identification of potential risks.

[0088] Step S3: Perform multimodal fusion modeling on the predicted production line abnormality data to obtain an abnormality intelligent analysis model; perform abnormality tracing on the predicted production line abnormality data according to the abnormality intelligent analysis model to generate the cause of the production line abnormality;

[0089] In this embodiment, when multimodal fusion modeling is performed on the data of predicted production line abnormalities, data from different sensors, such as temperature sensor and pressure sensor data, are integrated, and respective features are extracted through feature extraction algorithms. The fusion model is trained using a convolutional neural network (CNN) to comprehensively consider the impact of multiple data sources. After the training is completed, anomaly tracing analysis is performed based on the fusion model, and the predicted data is analyzed through a back propagation algorithm to determine the specific cause of the anomaly. For example, if the data of the vibration sensor is abnormal, combined with the temperature and pressure data, it can be traced back to a failure in the cooling system of the equipment or wear of components.

[0090] Step S4: construct a correction strategy for the abnormal cause of the production line and generate an abnormal correction strategy; share the abnormal correction strategy in a tree-like interactive manner through the intelligent control platform to obtain a shared correction instruction;

[0091] In this embodiment, when constructing a correction strategy for the cause of production line abnormality, the nature of the cause of the abnormality is first evaluated, and corresponding correction strategies are formulated based on historical data and expert experience. These strategies include replacing parts, adjusting operating parameters, or upgrading equipment. The correction strategies are shared interactively in a tree-like manner through the intelligent control platform, and a decision tree model is constructed to associate each correction strategy with specific equipment and abnormal conditions to ensure that relevant personnel can clearly understand the necessity and specific implementation methods of each correction. After the shared correction instructions are generated, they will be issued to relevant positions in the form of electronic documents for implementation.

[0092] Step S5: Perform abnormal correction processing on the production line according to the shared correction instruction to generate abnormal correction results; collect instruction feedback on the abnormal correction results through the intelligent control platform to execute data interaction based on the intelligent control platform.

[0093] In this embodiment, when the production line is corrected for anomalies according to the shared correction instructions, the on-site operators need to implement specific correction measures in accordance with the instructions. For example, if the instructions include replacing a part of a certain device, the operator will disassemble and replace it according to the steps. During the execution process, all operations must be monitored in real time through the intelligent control platform to ensure the integrity and accuracy of the information. The generated abnormal correction results, including the operating status of the corrected equipment, will be recorded through the instruction feedback function, and the system will automatically collect feedback data for subsequent analysis and optimization decisions to ensure the efficient operation of the entire production line.

[0094] Preferably, step S1 comprises the following steps:

[0095] Step S11: Acquire multi-dimensional data of production equipment;

[0096] Step S12: removing random noise from the multi-dimensional data of the production equipment to obtain denoised multi-dimensional data of the equipment;

[0097] Step S13: marking data anomalies on the denoising device multidimensional data to generate abnormality marked device multidimensional data;

[0098] Step S14: Analyze the statistical characteristics of the multi-dimensional data of the abnormal marking device to generate data statistical characteristics;

[0099] Step S15: Perform format homogenization processing on the multi-dimensional data of the abnormal marking equipment according to the data statistical characteristics to obtain the production line data flow.

[0100] In this embodiment, in the process of obtaining multi-dimensional data of production equipment, various sensors deployed on the equipment are used to collect data in real time, and the sensors include temperature, pressure, rotation speed, vibration and other types. The data acquisition device reads information from the sensor at a fixed frequency and transmits the data to the central data processing unit via industrial Ethernet. The data is stored in a database in a structured format, for example, using a timestamp as a keyword, and recording the corresponding sensor type and reading to ensure the integrity and traceability of the data. At this time, all data will be saved in the local database and uploaded to the cloud regularly for further analysis to ensure the real-time and security of the data. When the multi-dimensional data of the production equipment is subjected to random noise removal, the original data is first processed using a wavelet transform algorithm. The wavelet transform decomposes the signal into components of different frequencies, which can effectively identify and separate noise and signal. For the decomposed signal, a threshold is set to remove the noise components below the threshold, so that only the meaningful signal part is retained. After denoising, the generated denoising equipment multi-dimensional data is verified to ensure the accuracy and stability of the data, and the parameter settings of the processing process are recorded for subsequent backtracking and reproduction. When marking data anomalies for the multidimensional data of the denoising device, a statistical method is used to set the standard for data anomaly detection. For example, the Z-score method is used to calculate the deviation of each data point from its mean, and the points whose deviation exceeds the preset threshold are marked as abnormal. The process includes first calculating the mean and standard deviation of the denoised data, and then calculating the Z-score for each data point. Values ​​exceeding ±3 will be marked as abnormal. Finally, the generated multidimensional data of the abnormal marking device contains the identification of all normal and abnormal data, which is convenient for subsequent analysis and processing. In the process of statistical feature analysis of the multidimensional data of the abnormal marking device, various features are first extracted by descriptive statistical methods, including mean, standard deviation, maximum value, minimum value and quartiles. The abnormal marked data is analyzed separately from the normal data, and data visualization tools such as box plots and histograms are used to display the data distribution, so as to observe the characteristics of the abnormal data. The statistical analysis results are organized into reports, and the generated data statistical features will be used for subsequent model training and predictive analysis to ensure the accuracy and effectiveness of subsequent steps. When formatting the multi-dimensional data of abnormal marking equipment according to the statistical characteristics of the data, first set a unified data structure format, such as JSON or CSV, to ensure the consistency of subsequent processing. During the processing, extract and reorganize the relevant feature information of each data point, including timestamp, sensor type, value and abnormal mark. For data from different sources, use standardized methods to normalize to eliminate the impact of different dimensions on the analysis. Finally, the production line data stream that has been formatted is stored in a structured database to ensure that the data can be retrieved and used quickly and efficiently in the subsequent processing and analysis process.

[0101] Preferably, step S2 comprises the following steps:

[0102] Step S21: performing time sequence conversion on the production line data stream to obtain production time sequence data of the production line;

[0103] Step S22: performing long-term and short-term dependency analysis on the production time series data of the assembly line to generate a spatiotemporal correlation network;

[0104] Step S23: performing abnormal propagation evolution on the spatiotemporal correlation network according to the multi-dimensional data of the abnormal marking device to generate simulated abnormal diffusion data;

[0105] Step S24: predicting the abnormal situation of the production line data flow according to the simulated abnormal diffusion data to obtain predicted production line abnormal data.

[0106] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes:

[0107] Step S21: performing time sequence conversion on the production line data stream to obtain production time sequence data of the production line;

[0108] In this embodiment, when performing time series conversion on the production line data stream, first extract the timestamp and related sensor data from the stored structured data to ensure the order and time integrity of the data. Interpolation methods are used to fill in the data at irregular time intervals and unify the data to specified time intervals, such as every second or every minute. Then, the Pandas library in Python is used to resample the data, convert the original data into a time series format indexed by time, and generate production line time series data. All time series data are saved in a unified format to provide a reliable basis for subsequent analysis and ensure the coherence and consistency of the data.

[0109] Step S22: performing long-term and short-term dependency analysis on the production time series data of the assembly line to generate a spatiotemporal correlation network;

[0110] In this embodiment, when performing long-term and short-term dependency analysis on the production time series data of the assembly line, the long short-term memory network (LSTM) is first selected as the analysis model. The LSTM network is widely used due to its superior performance in processing time series data. The production time series data of the assembly line is divided into a training set and a test set, and the training set is input for model training, and the network parameters are optimized using a loss function such as mean square error (MSE). After the model training is completed, the trained LSTM model is used to predict the test set to generate a spatiotemporal correlation network. The network represents the dependency relationship between data in the form of nodes and edges. The nodes represent different time periods and data features, and the edges represent the degree and intensity of dependency between them.

[0111] Step S23: performing abnormal propagation evolution on the spatiotemporal correlation network according to the multi-dimensional data of the abnormal marking device to generate simulated abnormal diffusion data;

[0112] In this embodiment, when the spatiotemporal correlation network is subjected to anomaly propagation evolution according to the multidimensional data of the anomaly marking device, the marked anomaly data is first extracted, and the propagation model in graph theory, such as the SIR model (susceptible-infected-recovered model), is used to simulate the propagation process of the anomaly in the spatiotemporal correlation network. The anomaly probability of each node is associated with the state of the adjacent node, and the node state is updated iteratively to generate simulated anomaly diffusion data. This process requires setting the initial infection rate and propagation rate to ensure the accuracy of the model. Finally, the state of each node at different time steps is recorded to form a dynamic model of anomaly propagation, which is convenient for subsequent situation analysis.

[0113] Step S24: predicting the abnormal situation of the production line data flow according to the simulated abnormal diffusion data to obtain predicted production line abnormal data.

[0114] In this embodiment, when predicting the abnormal situation of the production line data flow based on the simulated abnormal diffusion data, a regression model is first constructed using the generated simulated data, such as random forest regression or gradient boosting regression, and the input features include timestamp, abnormal propagation status, etc. By training the historical data, the model parameters are adjusted so that it can effectively capture the patterns and trends of abnormal data. After the training is completed, the model is used to predict the current production line data flow to obtain the predicted production line abnormal data. The prediction results include the probability, time and potential impact range of the abnormality, and are finally output in the form of a structured report to provide a basis for decision-making on the production line.

[0115] Preferably, step S22 comprises the following steps:

[0116] Step S221: segmenting the production time series data of the assembly line into long and short time segments based on a preset segment duration threshold, generating long-term data segments and short-term data segments;

[0117] Step S222: Perform long short-term memory network analysis on the long-term data segments to obtain long-term dependency features;

[0118] Step S223: performing gated cycle unit analysis on the short-term data segment to generate short-term fluctuation features;

[0119] Step S224: performing correlation influence mining on the long-term dependency features and the short-term fluctuation features to obtain correlation mutual influence data;

[0120] Step S225: construct a correlation network for the long-term dependency characteristics and the short-term fluctuation characteristics according to the correlation and mutual influence data to generate a spatiotemporal correlation network.

[0121] In this embodiment, when the production time series data of the assembly line is segmented into long and short time segments, the segmentation time threshold is first set, for example, in hours and minutes. During the processing, the timestamps in the time series data are extracted using Python's Pandas library, and the data is divided into multiple segments based on the set threshold. In the specific operation, the pd.Grouper function is used to group the data to generate long-term data segments (for example, including hourly averages) and short-term data segments (for example, including real-time data per minute). Each segment is saved in an independent data frame (DataFrame) for subsequent analysis. When the long-term data segment is analyzed by a long short-term memory network (LSTM), the long-term data segment is first formatted into a shape suitable for LSTM input, that is, a three-dimensional array (number of samples, time steps, number of features). The LSTM model is constructed using the TensorFlow or Keras library, including an input layer, one or more LSTM layers, and an output layer. During the training process, an appropriate loss function (such as mean square error) and an optimizer (such as Adam) are selected, and the model weights are updated by the back propagation algorithm until convergence. After training, the model is used to predict long-term data segments, extract long-term dependency features, and form feature vectors for subsequent association mining. When performing gated recurrent unit (GRU) analysis on short-term data segments, the short-term data segments are first converted into a format suitable for GRU input, similar to the processing of LSTM. Use Keras to build a GRU model, set the appropriate number of layers and nodes, select a suitable activation function (such as ReLU), and configure the loss function and optimizer. Short-term data is input into the model, and the output layer obtained after training contains short-term fluctuation features. The feature extraction process should ensure the stability and prediction accuracy of the model, and finally output the short-term fluctuation feature vector for subsequent analysis. When performing association impact mining on long-term dependency features and short-term fluctuation features, the extracted feature vectors are first merged to form a comprehensive feature set. Then, the degree of association between each feature is evaluated using methods such as Pearson correlation coefficient or Spearman rank correlation coefficient. With the help of the mutual_info_regression function in the Scikit-learn library, the degree of influence of different features on the abnormal state is calculated to identify the key influencing features. Finally, the generated correlation and mutual influence data are saved in the form of a data frame, recording the intensity and direction of each feature's influence on other features. When constructing a correlation network for long-term dependent features and short-term fluctuation features based on the correlation and mutual influence data, the graph theory method is used to treat the features as nodes, and the degree of association between nodes as edge weights. The graph structure is created through the NetworkX library, and the edge weight value is set as the correlation strength. The properties of the nodes and edges are stored in the graph object to form a complete spatiotemporal correlation network.The network can be used to analyze the complex interactions between features, provide a basis for subsequent intelligent decision-making, and ensure the efficient operation of the production line. Finally, the constructed spatiotemporal correlation network is presented in a visual graph to facilitate the analysis and display of the correlation between features.

[0122] Preferably, step S3 comprises the following steps:

[0123] Step S31: Perform multi-layer perceptron network mapping on the predicted production line abnormality data to generate an abnormality classification network;

[0124] Step S32: Perform multimodal fusion modeling on the predicted production line abnormal data according to the abnormal classification network to obtain an abnormal intelligent analysis model;

[0125] Step S33: performing association rule mining on the abnormal classification network and the abnormal data of the predicted production line according to the abnormal intelligent analysis model to obtain an abnormal association rule set;

[0126] Step S34: Based on the abnormal association rule set, the abnormality data of the predicted production line is traced to the source, and the cause of the production line abnormality is generated.

[0127] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes:

[0128] Step S31: Perform multi-layer perceptron network mapping on the predicted production line abnormality data to generate an abnormality classification network;

[0129] In this embodiment, when a multi-layer perceptron network mapping is performed on the predicted production line abnormal data, a multi-layer perceptron (MLP) model is first established, including an input layer, multiple hidden layers and an output layer. The input layer receives the processed abnormal data, including multiple feature dimensions. The network is constructed using the TensorFlow or Keras framework, and the appropriate number of hidden layers and the number of neurons in each layer are set. The ReLU activation function is usually selected to improve the nonlinear modeling ability. During the training process, the cross entropy loss function is used to optimize the classification results, and the Adam optimizer is used to update the parameters. After the training is completed, the model output layer will generate an abnormal classification network based on the input data. The network is used to distinguish different types of abnormalities and form a set of classification labels for subsequent analysis.

[0130] Step S32: Perform multimodal fusion modeling on the predicted production line abnormal data according to the abnormal classification network to obtain an abnormal intelligent analysis model;

[0131] In this embodiment, when multimodal fusion modeling is performed on the abnormal data of the predicted production line according to the abnormal classification network, features from different data sources are first extracted, such as sensor data, operation logs, and environmental factors. These data are uniformly formatted into a specific vector form. Using the classification results generated by the abnormal classification network, weighted fusion of different modal data is performed, and the attention mechanism or splicing method is selected for fusion to ensure that each modal data is fully expressed in the model. Then, by constructing a new network (such as a deep neural network), the fused data is input to generate an abnormal intelligent analysis model. The model will use the fused multidimensional information for more accurate anomaly detection and analysis.

[0132] Step S33: performing association rule mining on the abnormal classification network and the abnormal data of the predicted production line according to the abnormal intelligent analysis model to obtain an abnormal association rule set;

[0133] In this embodiment, when mining association rules for the abnormal classification network and the abnormal data of the predicted production line according to the abnormal intelligent analysis model, it is first necessary to define the minimum support and minimum confidence thresholds for mining. The data output by the abnormal classification network is analyzed using the Apriori algorithm or the FP-Growth algorithm to identify frequent item sets and strong association rules. The obtained rules are organized into an abnormal association rule set, which contains the premise and conclusion of the rules, and the support and confidence of each rule are calculated to evaluate its reliability. Finally, the generated abnormal association rule set provides an important basis for subsequent traceability analysis, recording the relationship and potential causes between different abnormal events.

[0134] Step S34: Based on the abnormal association rule set, the abnormality data of the predicted production line is traced to the source, and the cause of the production line abnormality is generated.

[0135] In this embodiment, when tracing the abnormal data of the predicted production line based on the abnormal association rule set, the newly acquired abnormal data is first matched with the constructed rule set to find the rules that meet the conditions. For each matching rule, its premise and results are analyzed, and the potential cause of the abnormality is determined by backtracking analysis. The identified abnormal causes are sorted by priority and recorded in the database for subsequent query and verification. This process uses a graph database or a relational database to store traceability information to ensure data traceability and ease of management. Ultimately, the generated production line abnormality cause report will provide a clear basis for subsequent decision-making and ensure effective adjustment of the production process.

[0136] Preferably, step S34 includes the following steps:

[0137] Step S341: performing hierarchical clustering analysis on the abnormal association rule set to obtain an abnormal rule tree;

[0138] Step S342: embedding the abnormal rule tree into a graph structure to obtain a rule vector space;

[0139] Step S343: extracting abnormal patterns from the predicted production line abnormal data to obtain abnormal feature sequences;

[0140] Step S344: performing multi-scale convolution feature matching on the abnormal feature sequence and the regular vector space to obtain a feature association matrix;

[0141] Step S345: Perform Bayesian causal inference analysis on the feature association matrix to obtain the cause of production line abnormality.

[0142] In this embodiment, the abnormal rule set is converted into a format suitable for clustering. K-means or hierarchical clustering algorithm is used for processing, and the specific operation is to calculate the feature vector of each rule, and extract the keywords, support and confidence in the rule as features. In the Python environment, the hierarchical clustering model is constructed using the scikit-learn library, and a suitable distance metric (such as cosine distance) is selected to measure the similarity between the rules. After clustering, the generated abnormal rule tree displays the hierarchical relationship between different rules in the form of a tree diagram. When the abnormal rule tree is embedded in the graph structure, the tree structure is converted into a vector representation using a graph neural network (GNN). The nodes of the constructed graph represent each abnormal rule, and the edges represent the association relationship between the rules. By using Node2Vec or GraphSAGE algorithm, the nodes in the graph structure are embedded in the vector space to generate a vector representation of the rule. In this process, an adjacency matrix is ​​first constructed to calculate the features of each node during the training process, and then the node vector is adjusted by an iterative optimization algorithm (such as stochastic gradient descent), and the rule vector space finally obtained can be used for further abnormal pattern recognition and analysis. When extracting abnormal patterns from abnormal data of predicted production lines, feature extraction algorithms, such as principal component analysis (PCA) or autoencoders, are used to extract potential abnormal patterns from predicted data. The specific steps are as follows: first, the predicted data is standardized to eliminate the influence of different dimensions. Then, the processed data is input into the selected feature extraction model to extract sequences representing abnormal features. These sequences cover the key information reflecting abnormal behavior in the data, forming a structured abnormal feature sequence. When performing multi-scale convolution feature matching on the abnormal feature sequence and the rule vector space, a convolutional neural network (CNN) is first constructed to adapt to feature extraction at different scales. The input data is the abnormal feature sequence and the rule vector, and the network design contains multiple convolution layers and pooling layers to extract multi-scale features. During the training process, the cross entropy loss function is used to evaluate the model performance, optimize the model parameters, and ensure the effective fusion of features at different scales. Finally, the generated feature association matrix describes the degree of association between the abnormal features and the rule vector, providing basic data for subsequent analysis. When performing Bayesian causal inference analysis on the feature association matrix, a Bayesian network model is constructed to reveal the causal relationship between the features. Use Bayesian Network Toolbox (such as the pgmpy library) for modeling, take the feature association matrix as input data, and define the nodes and their dependencies. Determine the network structure through the maximum a posteriori estimation method, and use the observed data for parameter learning. Ultimately, the generated model can provide potential causal explanations for each abnormal feature and identify the root cause of the production line abnormality.

[0143] Preferably, step S4 comprises the following steps:

[0144] Step S41: constructing a knowledge network for the abnormal causes of the production line to obtain an abnormal knowledge network;

[0145] Step S42: constructing a correction strategy based on the abnormal knowledge network to generate an abnormal correction strategy;

[0146] Step S43: constructing a tree-shaped interactive chain for the production line through the intelligent control platform to obtain a strategy node cluster;

[0147] Step S44: Based on the abnormal correction result, interactive sharing processing is performed on the policy node cluster to obtain a shared correction instruction.

[0148] In this embodiment, when constructing a knowledge network for the abnormal causes of the production line, key factors are first extracted from the identified abnormal causes. These factors include equipment failure types, environmental impacts, operational errors, etc. A knowledge graph construction tool, such as Neo4j, is used to represent each abnormal cause and its related factors in the form of nodes and edges. Each node represents an abnormal cause or related factor, and the edge represents the relationship between them. By constructing a knowledge network, the mutual influence and dependency between different abnormal causes can be clearly presented, forming a systematic abnormal knowledge network to provide basic data for subsequent correction strategies. Analyze each node and its attributes in the knowledge network to identify the correction strategy. Apply the decision tree algorithm, combine the abnormal causes and correction measures, and formulate the corresponding correction strategy by analyzing the correlation and influence of each factor. In this process, historical data and experience need to be collected to verify the effectiveness of the correction strategy, and the strategy needs to be optimized to ensure that the formulated correction strategy can solve specific abnormalities in a targeted manner, and finally generate a comprehensive abnormal correction strategy. When constructing a tree-shaped interactive chain for the production line through the intelligent control platform, the generated abnormal correction strategy first needs to be converted into an operable instruction format. Using the graphical user interface (GUI) tool, a tree structure is constructed on the intelligent control platform to present each correction strategy and its execution order in the form of nodes. Each node represents a specific correction measure, and the child node represents its subordinate steps and conditions. Using the tree-shaped interactive chain, the execution process of the correction strategy can be clearly displayed to form a strategy node cluster. Collect the execution feedback information of each node, and use the real-time monitoring function of the intelligent control platform to record the execution status of each strategy. By setting up a data sharing interface, the execution results of each strategy node are integrated to generate shared correction instructions. In this process, data visualization tools are used to present the correction effect and execution status, so that relevant personnel can intuitively understand the implementation effect of the strategy. At the same time, the instructions are sent to each execution device through the control platform to realize the real-time update and sharing of targeted correction measures, ensuring that the production line can flexibly respond to various abnormal situations.

[0149] Preferably, step S42 includes the following steps:

[0150] Perform graph reasoning on the abnormal knowledge network to obtain a set of potential correction paths;

[0151] Monte Carlo tree search is performed on the potential correction path set to obtain the optimal correction sequence;

[0152] Carry out fuzzy logic control modeling on the optimal correction sequence to obtain an adaptive correction rule base;

[0153] Multi-agent collaborative decision-making is carried out on the adaptive correction rule base to generate anomaly correction strategies.

[0154] In this embodiment, a graph reasoning algorithm is used to analyze the relationship between nodes in the abnormal knowledge network. A graph database system (such as Neo4j) is used to query each node in the graph, extract relevant abnormal correction information, and construct a potential correction path set. In this process, by setting reasoning rules, the mutual influence of different abnormal causes is identified, thereby forming a correction path for reference. For example, for a specific type of equipment failure, the reasoning algorithm can automatically generate a series of correction suggestions based on its adjacent nodes (such as environmental factors, operating habits), and construct a potential correction path set. Set search parameters and evaluation functions, and perform multiple random simulations for each path. Apply the Monte Carlo tree search algorithm to gradually expand the path and evaluate the effect of each correction measure to generate the optimal correction sequence. This process uses the efficient computing power of the computer to repeatedly simulate the implementation effect of each correction measure, and select the correction sequence with the best performance through statistical analysis of the results. For example, for the evaluation of multiple correction schemes, different scoring criteria (such as time, cost, effect, etc.) can be set, and finally the correction scheme with the highest score is output as the optimal correction sequence. Convert each correction measure in the correction sequence into fuzzy rules. For example, for different equipment states, corresponding fuzzy variables (such as "normal", "minor fault", "serious fault") are set, and an adaptive correction rule base is established in combination with a fuzzy reasoning system. A fuzzy logic controller is applied to automatically match the corresponding correction rules by inputting the current equipment state, and output the corresponding correction measures. Fuzzy controllers are designed using tools such as MATLAB or Simulink to enable real-time adjustment of the rules in the rule base to adapt to different equipment state changes. When making multi-agent collaborative decisions on the adaptive correction rule base, a multi-agent system is constructed, with each agent representing different equipment or decision-making units. By defining the communication mechanism between agents, they can share information and collaboratively formulate correction strategies. Each agent participates in the decision-making process based on its own state and the instructions received, forming a globally optimized abnormal correction strategy. Using platforms such as JADE or MATLAB for the development and coordination of agents, all agents can make real-time decisions in a shared environment, and ultimately generate a comprehensive abnormal correction strategy to ensure the efficient operation of the production line.

[0155] Preferably, step S43 includes the following steps:

[0156] Identify the machine interaction relationships on the production line and obtain potential interaction pairs;

[0157] Construct a tree-shaped interaction chain for potential interaction pairs to generate a node connection tree structure;

[0158] Map the node connection tree structure to a distributed hash table to obtain a device addressing scheme;

[0159] Perform network coverage processing on the device addressing scheme to obtain the device communication protocol;

[0160] The consensus mechanism is integrated into the device communication protocol to obtain a strategy node cluster.

[0161] In this embodiment, the operation data of each production device, including status, fault records and running time, etc., are collected, and the interaction behavior between devices is analyzed by data mining technology. Machine learning algorithms (such as cluster analysis and association rule mining) are applied to identify potential interaction pairs. For example, adjacent sensors and actuators are regarded as potential interaction pairs, and the Scikit-learn library in Python is used for analysis to generate an interaction relationship matrix to clarify which devices have significant interaction relationships. The identified interaction pairs are converted into a tree structure, and the devices are connected as nodes and the interaction relationships as edges. The minimum spanning tree algorithm in graph theory is used to connect devices by optimizing paths to minimize communication delays and resource consumption. This process can be visualized using the Graphviz tool to display the interactive chain structure between devices and clarify the interactive role of each device. The address of each node is mapped to a distributed hash table to implement a device addressing scheme. The Consistent Hashing algorithm is used to ensure that the device can be efficiently and accurately addressed in the network. By mapping device features (such as ID and status information) to a hash table, rapid positioning and interaction of devices are achieved. Use tools such as Redis to build and manage hash tables to ensure dynamic expansion capabilities and high availability. When performing network coverage processing on the device addressing scheme, analyze the physical location and communication range of the device to ensure that the communication link of each device in the network is complete. Use the network topology optimization algorithm to ensure that all devices can be effectively covered and reduce communication blind spots. Test and optimize the network coverage scheme through simulation tools (such as NS2 or OMNeT++) to generate effective device communication protocols. When integrating the consensus mechanism of the device communication protocol, use a blockchain-based consensus algorithm (such as PBFT or PoS) to ensure that a consistent communication protocol is reached between devices. By establishing a decentralized protocol management system, each device participates in the formulation and verification of the protocol to form a policy node cluster. The cluster can achieve automated management and coordination through smart contracts to ensure the security and effectiveness of information transmission between devices. Use platforms such as Ethereum to develop and deploy smart contracts to achieve efficient interaction between devices.

[0162] Preferably, step S5 comprises the following steps:

[0163] Step S51: performing dependency analysis on the shared correction instruction to obtain a task dependency network;

[0164] Step S52: Prioritize the task-dependent networks to obtain a device execution plan;

[0165] Step S53: performing abnormal correction processing on the production line according to the equipment execution plan and generating an abnormal correction result;

[0166] Step S54: collect command feedback on the abnormal correction results through the intelligent control platform to perform data interaction based on the intelligent control platform.

[0167] In this embodiment, all correction instructions and their related task information are collected. Using the directed graph structure in graph theory, each task is regarded as a node, and edges are established according to the dependencies between tasks. A depth-first search algorithm is used to traverse the task network, identify the order of tasks, and generate a task dependency network. For example, sensor calibration and data restart are regarded as interdependent tasks, and a dependency graph is constructed through Python's NetworkX library to ensure the effective execution of subsequent steps. When prioritizing the task dependency network, a topological sorting algorithm is used to sort the tasks to determine the execution order. According to the urgency and dependency of the task, the weight value is set, and the dependency network is analyzed using the Kahn algorithm to generate an equipment execution plan. This plan clarifies which tasks should be executed first, thereby improving the efficiency of the entire production line. The task sorting results are displayed using a visualization tool to facilitate operators to quickly understand the task priority. When the production line is subjected to abnormal correction processing according to the equipment execution plan, the tasks are first refined into specific operation instructions and executed one by one according to the priority. An automated control system is used to control the equipment to operate according to the execution plan, such as first calibrating the sensor and then restarting the relevant equipment. Real-time monitoring and control are performed through PLC (Programmable Logic Controller) to ensure the accuracy and timeliness of each operation. After execution, the results of each correction operation are recorded and a report on the abnormal correction results is generated. When collecting instruction feedback on the abnormal correction results through the intelligent control platform, the correction results are first uploaded to the database of the control platform. The results are summarized and analyzed through data analysis tools to evaluate the effectiveness and impact of the correction measures. Data visualization tools (such as Grafana) are used to display the feedback results to facilitate decision makers to make subsequent adjustments. Ensure that the status information and correction results of all equipment are fed back to the control platform in a timely manner to achieve efficient data interaction and continuous optimization. Through the real-time monitoring system, operators can respond quickly to abnormal situations and ensure the stable operation of the production line.

[0168] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0169] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A data interaction method based on an intelligent control platform, characterized in that: The following steps are involved: Step S1: Acquire multi-dimensional data of production equipment; Mark the multi-dimensional data of production equipment for data anomalies and generate multi-dimensional data of abnormally marked equipment; Perform format homogenization on the multi-dimensional data of abnormal marking equipment to obtain the production line data flow; Step S2: Perform long-term and short-term dependency analysis on the production line data stream to generate a spatiotemporal correlation network; perform abnormal situation prediction on the production line data stream based on the spatiotemporal correlation network to obtain predicted production line abnormal data; Step S3: Perform multimodal fusion modeling on the predicted production line abnormality data to obtain an abnormality intelligent analysis model; perform abnormality tracing on the predicted production line abnormality data according to the abnormality intelligent analysis model to generate the cause of the production line abnormality; Step S4: construct a correction strategy for the abnormal cause of the production line and generate an abnormal correction strategy; share the abnormal correction strategy in a tree-like interactive manner through the intelligent control platform to obtain a shared correction instruction; Step S5: performing abnormal correction processing on the production line according to the shared correction instruction to generate an abnormal correction result; The intelligent control platform is used to collect command feedback on the abnormal correction results to execute data interaction based on the intelligent control platform.

2. The data interaction method based on the intelligent control platform according to claim 1 is characterized in that: Step S1 includes the following steps: Step S11: Acquire multi-dimensional data of production equipment; Step S12: removing random noise from the multi-dimensional data of the production equipment to obtain denoised multi-dimensional data of the equipment; Step S13: marking data anomalies on the denoising device multidimensional data to generate abnormality marked device multidimensional data; Step S14: Analyze the statistical characteristics of the multi-dimensional data of the abnormal marking device to generate data statistical characteristics; Step S15: Perform format homogenization processing on the multi-dimensional data of the abnormal marking equipment according to the data statistical characteristics to obtain the production line data flow.

3. The data interaction method based on the intelligent control platform according to claim 1 is characterized in that: Step S2 includes the following steps: Step S21: performing time sequence conversion on the production line data stream to obtain production time sequence data of the production line; Step S22: performing long-term and short-term dependency analysis on the production time series data of the assembly line to generate a spatiotemporal correlation network; Step S23: performing abnormal propagation evolution on the spatiotemporal correlation network according to the multi-dimensional data of the abnormal marking device to generate simulated abnormal diffusion data; Step S24: predicting the abnormal situation of the production line data flow according to the simulated abnormal diffusion data to obtain predicted production line abnormal data.

4. The data interaction method based on the intelligent control platform according to claim 3 is characterized in that: Step S22 includes the following steps: Step S221: segmenting the production time series data of the assembly line into long and short time segments based on a preset segment duration threshold, generating long-term data segments and short-term data segments; Step S222: Perform long short-term memory network analysis on the long-term data segments to obtain long-term dependency features; Step S223: performing gated cycle unit analysis on the short-term data segment to generate short-term fluctuation features; Step S224: performing correlation influence mining on the long-term dependency features and the short-term fluctuation features to obtain correlation mutual influence data; Step S225: construct a correlation network for the long-term dependency characteristics and the short-term fluctuation characteristics according to the correlation and mutual influence data to generate a spatiotemporal correlation network.

5. The data interaction method based on the intelligent control platform according to claim 1 is characterized in that: Step S3 includes the following steps: Step S31: Perform multi-layer perceptron network mapping on the predicted production line abnormality data to generate an abnormality classification network; Step S32: Perform multimodal fusion modeling on the predicted production line abnormal data according to the abnormal classification network to obtain an abnormal intelligent analysis model; Step S33: performing association rule mining on the abnormal classification network and the abnormal data of the predicted production line according to the abnormal intelligent analysis model to obtain an abnormal association rule set; Step S34: Based on the abnormal association rule set, the abnormality data of the predicted production line is traced to the source, and the cause of the production line abnormality is generated.

6. The data interaction method based on the intelligent control platform according to claim 5 is characterized in that: Step S34 includes the following steps: Step S341: performing hierarchical clustering analysis on the abnormal association rule set to obtain an abnormal rule tree; Step S342: embedding the abnormal rule tree into a graph structure to obtain a rule vector space; Step S343: extracting abnormal patterns from the predicted production line abnormal data to obtain abnormal feature sequences; Step S344: performing multi-scale convolution feature matching on the abnormal feature sequence and the regular vector space to obtain a feature association matrix; Step S345: Perform Bayesian causal inference analysis on the feature association matrix to obtain the cause of production line abnormality.

7. The data interaction method based on the intelligent control platform according to claim 1 is characterized in that: Step S4 includes the following steps: Step S41: constructing a knowledge network for the abnormal causes of the production line to obtain an abnormal knowledge network; Step S42: constructing a correction strategy based on the abnormal knowledge network to generate an abnormal correction strategy; Step S43: constructing a tree-shaped interactive chain for the production line through the intelligent control platform to obtain a strategy node cluster; Step S44: Based on the abnormal correction result, interactive sharing processing is performed on the policy node cluster to obtain a shared correction instruction.

8. The data interaction method based on the intelligent control platform according to claim 7 is characterized in that: Step S42 includes the following steps: Perform graph reasoning on the abnormal knowledge network to obtain a set of potential correction paths; Monte Carlo tree search is performed on the potential correction path set to obtain the optimal correction sequence; Carry out fuzzy logic control modeling on the optimal correction sequence to obtain the adaptive correction rule base; Multi-agent collaborative decision-making is carried out on the adaptive correction rule base to generate anomaly correction strategies.

9. The data interaction method based on the intelligent control platform according to claim 7 is characterized in that: Step S43 includes the following steps: Identify the machine interaction relationships on the production line and obtain potential interaction pairs; Construct a tree-shaped interaction chain for potential interaction pairs to generate a node connection tree structure; Map the node connection tree structure to a distributed hash table to obtain a device addressing scheme; Perform network coverage processing on the device addressing scheme to obtain the device communication protocol; The consensus mechanism is integrated into the device communication protocol to obtain a strategy node cluster.

10. The data interaction method based on the intelligent control platform according to claim 1, characterized in that: Step S5 includes the following steps: Step S51: performing dependency analysis on the shared correction instruction to obtain a task dependency network; Step S52: Prioritize the task-dependent networks to obtain a device execution plan; Step S53: performing abnormal correction processing on the production line according to the equipment execution plan and generating an abnormal correction result; Step S54: collect command feedback on the abnormal correction results through the intelligent control platform to perform data interaction based on the intelligent control platform.

Images