Intelligent factory automatic monitoring method and system based on knowledge base enhancement

Abstract

The invention relates to the technical field of data analysis, provides an intelligent factory automatic monitoring method and system based on knowledge base enhancement, and realizes more accurate anomaly analysis and more effective process adjustment of an intelligent factory. The method comprises the steps of performing knowledge enhancement fusion processing on an obtained real-time monitoring data set of an intelligent factory through a pre-constructed process knowledge base and a pre-constructed monitoring rule base, and generating a process knowledge graph; performing abnormal mode recognition processing on the process knowledge graph based on a semantic matching strategy, extracting feature description of an abnormal event and a semantic association path with a historical monitoring text, and generating an abnormal mode analysis result containing abnormal root cause inference; according to the abnormal mode analysis result and the dynamic incidence relation in the process knowledge graph, an automatic monitoring report containing root cause priority ranking and optimization operation guidance is generated, and the automatic monitoring report is fed back to the intelligent factory control terminal to trigger process adjustment operation.

Description

Technical Field

[0001] This invention belongs to the field of data analysis technology, specifically relating to a method and system for intelligent factory automation monitoring based on knowledge base enhancement. Background Technology

[0002] With the development of artificial intelligence technology, AI knowledge graphs are gradually being applied in the field of factory monitoring. AI knowledge graphs can integrate various complex data, providing strong support for the monitoring and management of smart factories, helping enterprises to better understand production status and optimize production processes.

[0003] However, existing AI knowledge graphs and factory monitoring technologies have several shortcomings. Current technologies mostly only perform simple data analysis and processing, failing to effectively integrate scattered data into a comprehensive knowledge system. Anomaly identification often remains superficial, unable to delve into the root causes of anomalies or their semantic connections to historical data. Furthermore, in generating monitoring reports, there is a lack of systematic root cause prioritization and effective optimization guidance, resulting in information fed back to the control terminal that cannot accurately guide process adjustments, hindering the achievement of efficient and stable production in smart factories. Therefore, how to achieve more accurate anomaly analysis and more effective process adjustments is a technical problem that needs improvement. Summary of the Invention

[0004] This invention provides a knowledge-based enhanced intelligent factory automation monitoring method and system, enabling more accurate anomaly analysis and more effective process adjustment in intelligent factories.

[0005] In a first aspect, embodiments of the present invention provide a knowledge-based enhanced intelligent factory automation monitoring method, applied to an intelligent factory automation monitoring system. The method includes: performing knowledge enhancement fusion processing on the acquired real-time monitoring data set of the intelligent factory through a pre-built process knowledge base and a monitoring rule base to generate a process knowledge graph; performing anomaly pattern recognition processing on the process knowledge graph based on a semantic matching strategy to extract feature descriptions of abnormal events and semantic association paths with historical monitoring texts, generating anomaly pattern analysis results containing anomaly root cause inference; generating an automation monitoring report containing root cause priority ranking and optimization operation guidance based on the anomaly pattern analysis results and the dynamic association relationship in the process knowledge graph, and feeding back the automation monitoring report to the intelligent factory control terminal to trigger process adjustment operations.

[0006] Secondly, embodiments of the present invention provide an intelligent factory automation monitoring system, which includes a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the above-described method.

[0007] Thirdly, embodiments of the present invention provide a computer-readable storage medium including a computer program, which, when run on an intelligent factory automation monitoring system, causes the intelligent factory automation monitoring system to perform the steps of the above-described method.

[0008] In this invention, a pre-built process knowledge base and monitoring rule base are used to perform knowledge-enhanced fusion processing on the real-time monitoring data set. This integrates scattered and isolated data into a structured process knowledge graph, improving the value and usability of the data. Anomaly pattern recognition based on semantic matching strategies allows for in-depth mining of the characteristics and semantic relationships of abnormal events, enabling accurate inference of the root causes of anomalies and improving the accuracy and comprehensiveness of anomaly analysis. Automated monitoring reports are generated based on the anomaly pattern analysis results and dynamic relationships, and fed back to the control terminal to trigger process adjustment operations. This provides timely and effective guidance for production adjustments, improving the production stability and efficiency of the smart factory, ensuring continuous optimization of the production process, and realizing autonomous monitoring and intelligent decision-making in the smart factory. Thus, more accurate anomaly analysis and more effective process adjustments can be achieved. Attached Figure Description

[0009] Figure 1 This is a flowchart illustrating a knowledge-based enhanced intelligent factory automation monitoring method provided in an embodiment of the present invention.

[0010] Figure 2 This is a schematic diagram of the structure of an intelligent factory automation monitoring system provided in an embodiment of the present invention. Detailed Implementation

[0011] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the technical solutions of the present invention, and not all embodiments. Based on the embodiments described in this document, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the technical solutions of the present invention.

[0012] See Figure 1 This is a knowledge base-enhanced intelligent factory automation monitoring method provided in this embodiment of the invention. This method can be applied to intelligent factory automation monitoring systems, and the specific process is as follows: steps 110-130.

[0013] Step 110: Through the pre-built process knowledge base and monitoring rule base, perform knowledge enhancement fusion processing on the acquired real-time monitoring data set of the smart factory to generate a process knowledge graph.

[0014] In an exemplary operational scenario of a smart factory, a process knowledge base accumulates a wealth of fundamental knowledge about production processes, while a monitoring rule base establishes guidelines for analyzing and correlating various types of data. The real-time monitoring dataset is a multi-source information set reflecting the current operational status of the factory. This invention employs knowledge-enhanced fusion processing on this data, aiming to integrate knowledge from the knowledge base and rule base into the real-time monitoring data, uncover potential correlations between data, and thereby generate a process knowledge graph.

[0015] For example, the process knowledge base stores information such as the ideal parameter range for equipment operation, standard procedures for process execution, and solutions to historical faults; the monitoring rule base specifies how to identify abnormal patterns in the data and how to infer causal relationships between different data. By combining the real-time monitoring data set with the process knowledge base and the monitoring rule base, the complex relationships between equipment operating status, process execution, and environmental factors can be identified, forming a complete process knowledge graph that provides comprehensive knowledge support for subsequent analysis and decision-making.

[0016] In one example, the real-time monitoring data set includes equipment operating status data, process execution flow data, and environmental perception data. The steps for acquiring the real-time monitoring data set of the smart factory include:

[0017] Step 101: Collect equipment operating status data through multiple types of sensors deployed on the equipment nodes of the smart factory; call the process control interface of the process execution system to extract the process execution flow data of the current batch production task of the smart factory; obtain environmental perception data including temperature and humidity distribution data, gas concentration gradient data and dust particle density data from the environmental monitoring unit.

[0018] In the smart factory of this invention, various equipment nodes are distributed at different locations in the production process, and multiple types of sensors are deployed on these equipment nodes. For example, equipment operating status data is collected by various sensors; speed sensors collect the operating speed of the equipment, expressed as "V". sensor "P" indicates that the pressure sensor collects the pressure inside the device. sensor This indicates that the aforementioned sensors continuously collect data, forming a data set of equipment operating status. The process execution system is responsible for managing the execution flow of production tasks. By calling its process control interface, it can obtain detailed information about the current batch of production tasks. For example, the process execution flow data includes the identifier "Process" for each step. id The sequence of operations, "Sequence", and the estimated execution time for each operation. time Information such as "TH" is provided. The environmental monitoring unit comprehensively monitors the environmental conditions inside the factory, and the temperature and humidity distribution data are displayed using "TH". dataThe text indicates that the temperature and humidity conditions are reflected in different areas of the factory; gas concentration gradient data is represented by "GC". data The text indicates that the concentration changes of various gases in space are shown; dust particle density data are represented by "DP". data The data indicates the density of dust particles in the air, and the above data reflects the environmental conditions during factory operation from different dimensions.

[0019] Step 102: Perform timestamp alignment and unit standardization on the equipment operation status data, the process execution flow data, and the environmental perception data to generate a real-time monitoring data set with a unified data format.

[0020] After acquiring equipment operating status data, process execution data, and environmental sensing data, appropriate processing is required to ensure data consistency and usability because the collection time and data units of these data may differ. Timestamp alignment processing unifies the time stamps of data from different sources to the same time scale, enabling accurate correlation and analysis of the data across the time dimension. For example, for speed data "V" in equipment operating status data... sensor "and pressure data" P sensor ", and the process execution time "Estimated" in the process execution flow data time "and temperature and humidity data in environmental sensing data" data "..." Through timestamp alignment algorithms, the time signatures of these data are adjusted to a consistent time coordinate system. Unit standardization converts the units of different data into a unified standard unit to eliminate the impact of unit differences on data analysis. For example, speed data "V..." sensor "Previously recorded in different units of speed, these can be converted to international standard units using a unit conversion function; pressure data 'P'..." sensor "Similar unit conversions were also performed to ensure that the units matched those of other relevant data. After the aforementioned timestamp alignment and unit standardization processes, equipment operating status data, process execution data, and environmental perception data can be integrated into a real-time monitoring data set with a unified data format, facilitating subsequent knowledge-enhanced fusion processing."

[0021] In another example, if the process knowledge graph contains dynamic relationships between equipment operating status and process execution standards, then step 110 includes:

[0022] Step 111: Extract equipment operating baseline parameters, process execution standard documents, and historical fault case texts from the process knowledge base to construct an initial knowledge graph containing equipment entities, process entities, and environmental entities.

[0023] Optionally, the equipment operating baseline parameters extracted from the process knowledge base provide a reference for judging whether the equipment is operating normally, using "Device" as the reference. benchmark The term "Process" indicates that the process execution standard document specifies the standardized flow of the production process in detail. standard The term "fault history" indicates that historical fault case texts record past equipment failures and process problems, along with their corresponding solutions. case This indicates that, based on this information, an initial knowledge graph is constructed. In this process, device entities (such as various production equipment) are represented. entity (e.g., each production process) entity ") and environmental entities (such as different environmental factors "Environment") entity As nodes in the knowledge graph, the connections between nodes are determined by analyzing their inherent relationships, thereby constructing an initial knowledge graph containing these entities. This initial knowledge graph lays the foundation for the subsequent integration of real-time monitoring data.

[0024] Step 112: Input the real-time monitoring data set into the pre-trained knowledge enhancement model to simultaneously identify abnormal fluctuation patterns in the equipment operation status data, process deviation events in the process execution data, and environmental risk characteristics in the environmental perception data.

[0025] Among them, the pre-trained knowledge-enhanced model, trained on a large amount of data, possesses the ability to identify and analyze complex data patterns. After inputting a real-time monitoring dataset into this model, it performs in-depth analysis of the equipment's operating status data. For example, regarding the equipment's operating speed "V..." sensor "and pressure" P sensor The model compares data such as "Device" to other data. benchmark The baseline parameters in the data are used to identify any abnormal fluctuations. If the velocity data "V" sensor "When data deviates from the normal range, the model identifies it as an abnormal fluctuation pattern. For process execution data, the model analyzes the process identifier 'Process'..." id The execution order ("Sequence") and the estimated execution time ("Estimated") are specified. time Information such as "Process" is used to determine if there are any process deviation events. If the actual execution order differs from "Process", the process will be considered. standard If the standard process is inconsistent with the data, or if the execution time exceeds a reasonable range, it is identified as a process deviation event. For environmental sensing data, the model uses temperature and humidity distribution data to determine the process deviation. data "Gas concentration gradient data" GC data "and dust particle density data" DPdata The model identifies environmental risk characteristics by combining environmental safety standards. For example, if the temperature and humidity exceed the suitable range for production, or if the concentration of some harmful gases is too high, the model identifies these as environmental risk characteristics.

[0026] Step 113: Based on the association rules in the monitoring rule base, perform causal reasoning on the abnormal fluctuation pattern, the process deviation event, and the environmental risk characteristics to generate a dynamic correlation between the equipment operating status and the process execution standard.

[0027] In this embodiment of the invention, the association rules in the monitoring rule base are formulated based on a large amount of historical data and professional knowledge, and are used to analyze the causal relationships between different data. For abnormal fluctuation patterns in the identified equipment operating status data, process deviation events in the process execution data, and environmental risk characteristics in the environmental perception data, causal inference is performed based on the association rules. For example, if the monitored equipment operating speed "V" is detected... sensor The abnormal fluctuation pattern of "Estimated" was observed, and the execution time of one of the processes was also found to be "Estimated". time "The occurrence of deviations suggests a possible causal relationship between the abnormal speed fluctuations and the deviation event in that process, as indicated by association rules. Further analysis of environmental risk characteristics, such as temperature and humidity in the environment..." data "If the value exceeds the normal range, the association rules may indicate a potential causal relationship between this and the equipment operating status and process execution. Through the above causal reasoning, we can identify the dynamic relationship between equipment operating status and process execution standards, and reveal the mutual influence mechanism between different factors."

[0028] Step 114: Integrate the dynamic relationships into the initial knowledge graph in an incremental update manner to generate a process knowledge graph that integrates real-time monitoring data and historical knowledge.

[0029] In this step, the dynamic relationships between equipment operating status and process execution standards, obtained through causal reasoning, are incrementally updated and integrated into the initial knowledge graph. Incremental updates mean adding newly discovered relationships without disrupting the original structure of the initial knowledge graph. For example, for the equipment entity "Device" in the initial knowledge graph... entity "and process entity" entity If causal reasoning discovers new dynamic relationships between them, connecting edges are added between the corresponding nodes in the knowledge graph, and information such as the type and strength of the relationship is labeled. For the environment entity "Environment"... entity"The new relationships between the knowledge graph and the equipment and process entities are also updated incrementally. As a result, the initial knowledge graph gradually evolves into a process knowledge graph that integrates real-time monitoring data and historical knowledge, providing a richer and more accurate knowledge base for subsequent anomaly pattern recognition and analysis."

[0030] Step 120: Perform abnormal pattern recognition processing on the process knowledge graph based on the semantic matching strategy, extract the feature description of abnormal events and the semantic association path with historical monitoring text, and generate abnormal pattern analysis results containing abnormal root cause inference.

[0031] After the process knowledge graph is constructed, it is analyzed in depth based on a semantic matching strategy to identify abnormal patterns. The semantic matching strategy compares and analyzes the semantic information contained in each node and edge of the knowledge graph to identify parts that do not conform to normal patterns. For example, the knowledge graph records semantic information such as equipment operating status, process execution flow, and the relationships between them. By comparing the current state reflected by real-time monitoring data with normal semantic patterns in the knowledge graph, abnormal situations are identified. Once an anomaly is detected, a feature description of the abnormal event is extracted. This feature description includes the location where the anomaly occurred (e.g., the involved equipment entity "Device"). entity "or process entity" entity Information such as the abnormality's manifestation (e.g., abnormal equipment operating speed, deviation in process execution time, etc.) and the time when the abnormality occurred.

[0032] Simultaneously, by tracing the semantic association paths within the knowledge graph, historical monitoring texts related to the anomalous event are identified. For example, by following the connections between nodes in the knowledge graph, historical records of similar anomalies are found, and the semantic similarity between these historical records and the current anomalous event is analyzed to generate semantic association paths. Finally, based on feature descriptions and semantic association paths, root cause inference is performed, yielding anomaly pattern analysis results that include root cause inference, providing direction for subsequent problem-solving.

[0033] In an optional embodiment, step 120 includes:

[0034] Step 121: Based on the dynamic relationships in the process knowledge graph, construct a semantic matching template that includes equipment operating status thresholds.

[0035] The dynamic relationships within the process knowledge graph illustrate the interconnections between equipment operating status, process execution flow, and environmental factors. Based on these relationships, semantic matching templates are constructed. For example, for equipment operating status, based on "Device..." benchmarkThe threshold for the equipment's operating status is determined by using the baseline parameters and the normal fluctuation range in historical data. For the equipment operating speed "V",... sensor , set a reasonable speed threshold range "V" threshold "; Regarding pressure "P" sensor "Set the pressure threshold range "P" threshold By combining these threshold information with other relevant semantic information in the process knowledge graph, a semantic matching template containing equipment operating status thresholds is constructed. This template serves as a standard for subsequent comparative analysis to determine whether real-time monitoring data conforms to the normal pattern.

[0036] Step 122: Perform similarity calculation between the current equipment operating status data in the real-time monitoring data set and the semantic matching template to identify abnormal data segments that deviate from the process execution standards.

[0037] Optionally, the current device operating status data from the real-time monitoring dataset, such as speed "V", can be used. sensor "and pressure" P sensor Data such as "..." is used to calculate similarity with semantic matching templates. Similarity calculation can be achieved through various methods, such as calculating the distance between the data and a threshold range, or comparing the changing trends of the data. The speed "V" is used as the benchmark. sensor For example, calculate its relationship with "V". threshold The degree of difference between "V" and "V" sensor "Exceeded the V" threshold "A range, or a trend that differs from the trend under normal conditions, indicates an anomaly. Through the aforementioned similarity calculations, abnormal data segments deviating from process execution standards can be identified. These abnormal data segments may signal problems with equipment operation and require further in-depth analysis."

[0038] Step 123: Call the historical monitoring text database to retrieve historical fault records that have semantic similarity to the abnormal data fragment, and generate a feature description of the abnormal event.

[0039] In this embodiment of the invention, the historical monitoring text database stores a large amount of past monitoring data and fault records. When an abnormal data fragment is identified, this database is invoked, and historical fault records with semantic similarity to the abnormal data fragment are searched through semantic retrieval. For example, for the current device operating speed "V sensorThe system identifies anomalous data fragments and searches the database for historical records containing similar speed anomalies. These historical records may detail the fault phenomena, the time of occurrence, and related equipment status. By analyzing and organizing these historical records, key information related to the current anomaly is extracted, generating a feature description of the anomaly. This feature description not only includes the specific manifestations of the current anomaly but also draws on relevant characteristics of similar historical events, making the description more comprehensive and accurate.

[0040] Step 124: Based on the entity association path in the process knowledge graph, trace the propagation trajectory of the abnormal data fragment in the equipment operation link and generate a semantic association path containing multiple levels of influence.

[0041] Optionally, the entity association paths in the process knowledge graph illustrate the connections and influence transmission methods between different entities. Based on these association paths, the propagation trajectory of anomalous data fragments within the equipment operation chain can be traced. For example, starting from the anomalous equipment entity "Device..." entity "Starting from here, follow the connecting edges in the knowledge graph to find other related equipment entities, process entities, and environment entities. If the equipment's operating speed is "V..." sensor "An anomaly has occurred; by examining the associated paths, it may be discovered that it is affecting the execution time of downstream processes." (Estimated) time This, in turn, affected the progress of the entire production process. Simultaneously, it may be discovered that this anomaly is related to certain environmental factors (such as temperature and humidity). data There is a correlation. Through the above tracing, a semantic correlation path containing multiple levels of impact is generated, clearly showing the propagation and impact range of the abnormal event throughout the production system.

[0042] Step 125: Logically integrate the feature description with the semantic association path to generate an anomaly pattern analysis result that includes anomaly root cause inference.

[0043] The feature description details the specific manifestations of the abnormal event, while the semantic association path demonstrates the propagation and impact range of the abnormal event within the system. By comprehensively analyzing both, root cause inference is performed. For example, if the feature description indicates abnormal equipment operating speed, and the semantic association path suggests that this abnormality is related to a step in the process execution flow and environmental factors, it can be inferred that environmental factors may have caused the equipment malfunction, thus affecting process execution. Through this logical integration, anomaly pattern analysis results containing root cause inference are generated, providing a strong basis for subsequent solution development.

[0044] Step 130: Based on the abnormal pattern analysis results and the dynamic correlation in the process knowledge graph, generate an automated monitoring report containing root cause priority ranking and optimization operation guidance, and feed the automated monitoring report back to the smart factory control terminal to trigger process adjustment operations.

[0045] After obtaining the results of the abnormal pattern analysis, an automated monitoring report is generated by combining the dynamic relationships in the process knowledge graph. This report aims to provide managers of smart factories with clear and accurate information to guide them in making process adjustments.

[0046] In a preferred embodiment, the step of generating an automated monitoring report containing root cause priority ranking and optimization operation guidance based on the abnormal pattern analysis results and the dynamic correlation in the process knowledge graph includes:

[0047] Step 131: Based on the causal reasoning results in the dynamic relationship, and combined with the semantic relationship path length and the number of affected entities in the abnormal pattern analysis results, calculate the comprehensive influence weight of each abnormal root cause.

[0048] The causal inference results in the dynamic correlation reveal the causal links between the anomalous event and various factors. The semantic correlation path length in the anomalous pattern analysis results reflects the scope of the anomalous event's influence, and the number of affected entities reflects the number of affected objects. When calculating the comprehensive influence weight of each anomalous root cause, the causal inference results are first analyzed to determine the direct impact of each root cause on the anomalous event. For example, if one root cause is proven by causal inference to be the main factor causing the anomalous event, it is given a higher initial weight. Then, the semantic correlation path length is considered; the longer the path, the wider the influence of the root cause, and the weight is increased accordingly. Regarding the number of affected entities, the more affected entities, the greater the influence of the root cause, and the weight is also increased. By comprehensively considering these factors, the weight of each anomalous root cause is calculated to obtain the comprehensive influence weight, which intuitively reflects the importance of each anomalous root cause in the entire anomalous event.

[0049] Step 132: Based on the inter-entity interaction strength mapped by the dynamic relationship, the comprehensive influence weight is corrected to generate the root cause priority ranking.

[0050] Optionally, dynamic relationships not only demonstrate causal connections between entities but also reflect the strength of their interactions. The overall impact weights are adjusted based on this strength. For example, if some entities have close relationships and strong interactions, their related root causes should be prioritized higher. The overall impact weights are adjusted according to the strength of interactions between entities. If the interaction strength between entities involved in one root cause is high, its weight is appropriately increased; conversely, if the interaction strength is low, its weight is decreased. Through these adjustments, a root cause priority ranking is generated. This ranking helps managers quickly determine the order in which to handle anomalies, prioritizing issues with the greatest impact on production.

[0051] Step 133: Extract optimization operation rules from the monitoring rule base that match the root cause priority ranking and the dynamic correlation, and generate optimization operation guidelines that include process parameter adjustment thresholds and equipment maintenance condition constraints.

[0052] It's understandable that the monitoring rule base stores optimized operating rules for different anomalies. Based on root cause priority and dynamic correlation, corresponding rules are extracted from the monitoring rule base. For example, if one of the root causes of an anomaly is due to equipment operating parameters exceeding the range, the monitoring rule base may specify specific process parameter adjustment thresholds. For equipment operating speed "V..." sensor The rule might indicate that the speed needs to be adjusted to a specific range, which is the process parameter adjustment threshold. Simultaneously, regarding equipment maintenance, the rule might specify maintenance constraints, such as periodically inspecting certain components or replacing specific parts. By extracting these matching rules, an optimized operation guide containing process parameter adjustment thresholds and equipment maintenance constraints is generated, providing specific operational guidance for actual process adjustments.

[0053] Step 134: Map and reorganize the root cause priority sorting and the optimization operation guidance according to the entity topology structure of the dynamic relationship to generate an automated monitoring report containing multi-dimensional optimization paths.

[0054] The dynamic entity topology structure illustrates the connections and hierarchical relationships between various entities. For example, there is a complex topology among equipment entities, process entities, and environmental entities. Each root cause in the root cause priority ranking is matched and arranged with its corresponding optimization operation guide according to this topology. For instance, for an anomaly root cause caused by one equipment entity, its priority and corresponding optimization operation guide are placed in positions related to that equipment entity. Through this mapping and reorganization, an automated monitoring report containing multi-dimensional optimization paths is generated. This report demonstrates how to resolve anomalies from multiple dimensions, providing comprehensive and organized guidance information for the control terminals of the smart factory, facilitating managers to quickly understand the key issues and corresponding solutions.

[0055] In another preferred embodiment, the step of feeding back the automated monitoring report to the smart factory control terminal to trigger process adjustment operations includes:

[0056] Step 135: Parse the process parameter adjustment thresholds corresponding to the dynamic correlation in the automated monitoring report, and generate process control instructions containing the target adjustment range and execution sequence.

[0057] In this embodiment of the invention, the automated monitoring report records in detail key information such as the adjustment thresholds of process parameters corresponding to dynamic correlations. When parsing this information, the adjustment thresholds corresponding to each process parameter are first identified. For example, for the process parameter "Temp" which involves temperature control... param Extract the adjustment threshold range "Temp" from the report. threshold range Then, based on the production process and technological requirements, the target adjustment range is determined. This target adjustment range must not only meet the threshold requirements of the process parameters themselves, but also take into account the synergistic effects with other relevant parameters and the stability of the entire production system. For example, when adjusting "Temp..." param "When adjusting process parameters, it's crucial to ensure that the operation of adjacent equipment is not negatively impacted. Simultaneously, the execution sequence must be determined based on the production plan and equipment operating rhythm. The execution sequence clearly defines when each process parameter adjustment should be performed, ensuring the orderliness and continuity of the entire process adjustment. Through these steps, process control instructions containing the target adjustment range and execution sequence are generated, providing accurate operational guidance for the smart factory control terminal."

[0058] Step 136: Generate a maintenance task instruction set according to the equipment maintenance condition constraints, and synchronously transmit the process control instructions and the maintenance task instruction set to the intelligent factory control terminal.

[0059] The equipment maintenance condition constraints are clearly defined in the automated monitoring report, specifying various requirements for equipment maintenance. A set of maintenance task instructions is generated based on these requirements. For example, for a certain critical piece of equipment... device Maintenance constraints may include periodic inspection of specific components of the equipment. X The wear and tear of the equipment, and one of the areas of the cleaning equipment. Y "and replacement of specific consumables" Z "etc." To address these requirements, detailed maintenance task instructions are generated. Each instruction clearly specifies the task's details, the personnel responsible for its execution, and the execution time. For example, "On the 15th of this month, equipment maintenance team A will perform maintenance on 'Key'..." device 'of' Component X "Conduct inspections and record wear data." After generating the maintenance task instruction set, it is transmitted synchronously to the smart factory control terminal along with the previously generated process control instructions. As a result, the control terminal can simultaneously receive task instructions for process adjustment and equipment maintenance, rationally allocate resources, and ensure that the production process can meet process requirements while guaranteeing the normal operation of equipment.

[0060] Step 137: Construct a process parameter adjustment prediction model based on the dynamic correlation, and monitor the equipment operation status response data fed back by the smart factory control terminal in real time; when the deviation between the equipment operation status response data and the process parameter adjustment prediction model exceeds a preset threshold, trigger the backtracking update of the correlation of the process knowledge graph, and regenerate an automated monitoring report with priority correction.

[0061] Among them, the process parameter adjustment prediction model utilizes the interrelationships between equipment operating status, process execution, and environmental factors inherent in dynamic correlations to predict the equipment operating status after process parameter adjustments. For example, the model considers the process parameter "Speed"... param Adjustments to "Output" may affect the device's output quality. quality "and the operational stability of adjacent devices" stability Factors such as these are studied and analyzed through historical data to establish mathematical relationships or logical rules between them.

[0062] In actual operation, real-time monitoring of equipment operating status response data fed back from the smart factory control terminal is conducted. For example, the actual operating speed of the equipment is obtained. speed Output quality indicators "Actual" qualityData such as these are collected. These actual data are compared with the results predicted by the process parameter adjustment prediction model, and the deviation value is calculated. If the deviation value exceeds a preset threshold, it indicates a significant difference between the actual operating conditions and the predicted results, suggesting that unidentified problems or factors may be affecting the effectiveness of process adjustments. At this point, a retrospective update of the process knowledge graph is triggered. By retrospectively examining the relationships in the process knowledge graph, the causal chain of abnormal events is re-examined, and potentially missed or misjudged relationships are identified. For example, it checks whether new environmental factors have affected equipment operation, or whether there is a misunderstanding of one of the process steps. Based on the results of the retrospective update, the priority of the root causes of the anomalies is recalculated, and the automated monitoring report is revised, providing more accurate and reliable guidance for subsequent process adjustments.

[0063] In one optional embodiment, the process of constructing the process knowledge base includes:

[0064] Step 210: Collect historical production data, equipment maintenance logs, and process guidance manuals from the smart factory; parse historical statistical values ​​of equipment operating parameters from the historical production data and extract equipment operating baseline parameters; extract process execution standard documents from the structured text of the process guidance manual; perform event sequence analysis on the equipment maintenance logs and extract fault handling case texts containing fault phenomena and solutions.

[0065] During the long-term operation of smart factories, a wealth of historical production data, equipment maintenance logs, and process instruction manuals have been accumulated. Historical production data records detailed information for each production stage, including equipment operating parameters, production output, and product quality. In-depth analysis of this historical production data, using statistical analysis methods, calculates historical statistical values ​​for equipment operating parameters. For example, for the equipment's operating speed parameter "Speed..." param The system calculates the average, maximum, and minimum values ​​of the equipment over different time periods, and extracts the equipment's operating baseline parameter "Speed". benchmark The process instruction manual is a standardized document outlining the production process. Its clear text structure contains detailed requirements and standard procedures for each production step. A specific text extraction algorithm accurately extracts the process execution standard document "Process" from the structured text. standard doc The document clearly defines the operating procedures and standard ranges of process parameters for each step. The equipment maintenance log records various events during equipment maintenance, including the time and symptoms of equipment failures and the repair measures taken. Event sequence analysis is performed on the equipment maintenance logs, using specific algorithms to identify the chronological order and causal relationships between events, extracting fault handling case text containing fault symptoms and solutions. casetext The above text provides valuable experience for handling similar faults in the future.

[0066] Step 220: Perform entity identification and relationship extraction on the equipment operating baseline parameters to generate static association relationships between equipment entities and process entities.

[0067] Among these, the equipment operating baseline parameters contain various information about equipment operation. Entity identification is then performed on these parameters, that is, identifying the relevant equipment entities and related process entities. For example, for "Speed..." benchmark The parameters are used to identify the corresponding device entity, "Device". X ", and process entities related to the operation of the equipment, such as "Process A (This is one of the production processes in which the equipment participates). Then, relation extraction is performed; by analyzing the inherent relationships between parameters and entities, the static association between the equipment entity and the process entity is determined. For example, if "Speed..." benchmark "It is "Device" X "In executing "Process" A "The standard operating speed during the process, then establish the 'Device'..." X "and "Process A The static association between "the standard speed for performing this process" and "the standard speed for performing this process" is established. By performing similar entity identification and relationship extraction operations on the operating baseline parameters of all equipment, a series of static associations between equipment entities and process entities are generated. These relationships provide the basic structural framework for building the process knowledge base.

[0068] Step 230: Based on the event chain decomposition results of the fault handling case text, and combined with the constraints in the process execution standard document, construct causal reasoning rules between fault occurrence conditions and handling measures.

[0069] The fault handling case texts detail the occurrence and solutions of equipment failures, decomposing them into event chains. Failure events are broken down into a series of sub-events based on chronological order and causal relationships. For example, a fault case might include sub-events such as "equipment temperature too high," "leading to equipment performance degradation," and "ultimately causing production interruption." The process execution standard documents specify various constraints in the production process, such as the range of process parameters and the execution sequence of procedures.

[0070] Furthermore, by combining the event chain decomposition results of fault handling case texts with the constraints in the process execution standard documents, causal reasoning rules are constructed. For example, if the process execution standard specifies that the equipment operating temperature should be within a certain range, and "equipment overheating" caused the fault in the fault handling case, then the above causal reasoning rule can be constructed: when the equipment operating temperature exceeds the range specified in the process execution standard, it may cause faults such as equipment performance degradation; the corresponding handling measures are to take cooling measures and check the equipment for damage. Through the analysis and summarization of multiple fault handling cases, a complete set of causal reasoning rules between fault occurrence conditions and handling measures is constructed. These rules provide a logical basis for fault diagnosis and resolution.

[0071] Step 240: Integrate the static association with the causal reasoning rules to generate the process knowledge base.

[0072] Optionally, knowledge fusion can be performed on the static associations between the previously generated equipment entities and process entities, and the causal reasoning rules between fault occurrence conditions and handling measures. For example, the connection between equipment operating parameters and process steps in the static associations can be combined with the relationship between fault occurrence and handling measures in the causal reasoning rules. In the process knowledge base, for one of the equipment entities "Device"... Y This records both its relationship with related process entities (such as "Process") B The normal operation relationship between (e.g., executing "Process") B The system not only records the standard parameters for each process step, but also the causal reasoning rules and corresponding handling measures for malfunctions (such as those caused by violations of standard parameters). Through the integration of this knowledge, different aspects of knowledge are combined to form a complete and systematic process knowledge base. This knowledge base can provide comprehensive knowledge support for the production process of a smart factory, providing accurate information for both process guidance during normal production and diagnosis and handling of malfunctions.

[0073] In an optional embodiment, the process of constructing the monitoring rule base includes:

[0074] Step 310: Collect historical fault work orders and process execution records; perform event annotation and association parsing on the historical fault work orders to generate a preprocessed data set containing fault triggering conditions and handling actions; extract the process sequence chain and parameter fluctuation range from the process execution records, and generate a set of rule elements in combination with equipment operation constraints.

[0075] Historical fault work orders record detailed information about past equipment failures, including the time, location, symptoms, and repair measures taken. Process execution records document the actual execution of processes during production, such as the sequence of operations and the actual parameter values ​​for each operation. Historical fault work orders are annotated with events to clarify key information for each fault event, such as fault type and triggering conditions. For example, for a fault work order, the triggering condition might be annotated as "equipment pressure too high," and the action would be "shut down the equipment and adjust the pressure." Through correlation analysis, the relationships between different fault events are identified, generating a preprocessed dataset containing fault triggering conditions and corresponding actions. data set Simultaneously, the process sequence chain "Process" is extracted from the process execution record. sequence chain "Parameter" refers to the execution order of each process and the range of parameter fluctuations. fluctuation range "...", such as the actual variation range of relevant process parameters in each process step. Combined with equipment operating constraints, such as the equipment's maximum load pressure and minimum operating speed, a set of rule elements, "Rule", is generated. elements set The aforementioned set of rule elements provides the basic material for the subsequent generation of monitoring rules.

[0076] Step 320: Based on the fault triggering conditions in the preprocessed data set and the parameter fluctuation range in the rule element set, analyze the spatiotemporal correlation between abnormal fluctuation patterns and equipment operation using the event sequence pattern mining algorithm to generate causal rules.

[0077] Among them, the event sequence pattern mining algorithm is used to discover hidden patterns and regularities in data. Based on the fault triggering conditions in the preprocessed dataset and the parameter fluctuation range in the rule element set, this algorithm is used to analyze the temporal and spatial correlation between abnormal fluctuation patterns and equipment operation. For example, it analyzes the relationship between the fluctuation of equipment parameters and fault triggering conditions within a certain time period. If it finds that the equipment's pressure parameter "Pressure" is affected... param "Exceeding the normal fluctuation range at several consecutive time points" Parameter fluctuation range Then, a "device malfunction" occurred, which leads to the above causal rule: when "Pressure" occurs... param"When fluctuations continuously exceed the normal range, equipment failure may be triggered. By analyzing a large amount of data and using event sequence pattern mining algorithms, the causal relationship between various abnormal fluctuation patterns and equipment operation is identified, generating a series of causal rules. These rules provide a basis for monitoring equipment operating status and detecting potential failures."

[0078] Step 330: Perform logical conflict detection on the causal rules and inject the device operation constraints to generate a set of rule constraints that includes parameter deviation tolerance and operation restrictions.

[0079] In this embodiment of the invention, the generated causal rules may contain logical conflicts; for example, some rules may contradict each other under specific circumstances. Logical conflict detection is performed on the causal rules using a specific logical analysis algorithm to check whether there are conflicts between different rules. For example, if a rule specifies that when the device temperature "Temperature" is... param "If one rule specifies a certain action when the temperature exceeds one threshold, but another rule specifies a different action under the same temperature conditions, this creates a logical conflict. Upon detecting a conflict, the rules are adjusted and corrected. Simultaneously, operational constraints are injected into the equipment, such as the equipment's safe operating range and operational feasibility limitations. For example, specifying the equipment's pressure..." param "Cannot exceed its maximum bearing pressure" Max pressure This is essentially a device operation constraint. These constraints are integrated into causal rules to generate a set of rule constraints called "Rule," which includes parameter deviation tolerances and operational limitations. constraint set The parameter deviation tolerance clarifies the acceptable range of fluctuations in equipment parameters, while the operational restrictions specify the operations that the equipment can perform under different conditions. This set of rules and constraints makes the monitoring rules more rigorous and practical.

[0080] Step 340: Input the causal rules and the rule constraint set into the rule reasoning module, verify the rule triggering conditions by combining real-time device status data, and eliminate invalid rules to form an initial monitoring rule library.

[0081] Optionally, the causal rules and rule constraint set are input into the rule reasoning module, which can verify the triggering conditions of the rules based on real-time device status data. For example, real-time acquisition of device operating parameters, such as temperature. param "Pressure" param "Data such as 'Temperature' is used to determine whether the triggering condition of a rule is met, based on causal rules and the set of rule constraints. If one of the rules stipulates that 'Temperature'..." param"Some action should be performed when the temperature exceeds one of the thresholds, but the actual 'Temperature'..." param "If the threshold is not reached, the rule will not be triggered under the current circumstances. By continuously verifying the rules based on real-time device status data, invalid rules that are unlikely to be triggered in actual operation or are unreasonable are eliminated. For example, some rules may be based on incorrect data or unreasonable strategies, and these will be removed after verification. After the above processing, an initial monitoring rule base is formed. The rules in this initial base are more accurate and practical, and can be effectively used to monitor the operating status of devices and detect anomalies."

[0082] Step 350: Based on the environmental data stream, match the rule trigger frequency and anomaly coverage of the initial monitoring rule base, dynamically adjust the rule weights of the initial monitoring rule base and perform incremental updates to generate the monitoring rule base.

[0083] The environmental data stream reflects changes in various environmental factors during the operation of the smart factory, such as temperature, humidity, and gas concentration. The initial monitoring rule base is matched with the environmental data stream to determine the trigger frequency of rules, i.e., the frequency with which each rule is triggered under different environmental conditions. Simultaneously, the anomaly coverage rate of the rules is evaluated, i.e., the proportion of anomalies that a rule can detect. For example, in a high-temperature environment, some rules related to equipment temperature monitoring may trigger more frequently. If a rule is found to be rarely triggered in a specific environment, but should be monitored for certain anomalies, its weight may need to be adjusted. Weight adjustment can be performed using specific algorithms; for example, if a rule has a low anomaly coverage rate but may play an important role in certain critical situations, its weight can be appropriately increased. By dynamically adjusting rule weights and incrementally updating the initial monitoring rule base based on new situations and data, new rules are continuously added or existing rules are optimized. For example, when a new anomaly pattern is discovered, the corresponding monitoring rule is added to the base. After the above dynamic adjustments and incremental updates, the final monitoring rule base is generated. This monitoring rule base can better adapt to the complex and ever-changing operating environment of the smart factory, improving the accuracy and effectiveness of monitoring.

[0084] In an alternative embodiment, after feeding the automated monitoring report to the smart factory control terminal to trigger a process adjustment operation, the method further includes:

[0085] Step 141: Based on the execution results of the process adjustment operation fed back by the intelligent factory control terminal, extract the adjusted target operating status response data.

[0086] It is understandable that after receiving a process adjustment command, the smart factory control terminal will execute the corresponding adjustment operation and provide feedback on the execution result. Based on these feedback results, the adjusted target operating status response data is extracted. For example, regarding the equipment's operating speed "Speed"... param The control terminal feedback indicates the actual operating speed has been adjusted. speed "; Regarding the process parameter "Quality" param (e.g., product quality indicators), the adjusted actual quality value is "Adjusted". quality The above data reflects the actual operating status of the equipment and processes after the process adjustment operation, and is an important basis for evaluating the adjustment effect.

[0087] Step 142: Match and verify the target operating status response data with the dynamic correlation in the process knowledge graph to identify the impact characteristics of the process adjustment operation on the root cause of the anomaly.

[0088] Optionally, the extracted target operating status response data can be matched and validated against the dynamic relationships in the process knowledge graph. For example, in the process knowledge graph, the equipment operating speed "Speed" param "Quality" param There is a certain dynamic relationship between them. By comparing the adjusted "Adjusted" speed "and "Adjusted" quality "Based on the expected results from the relationships in the knowledge graph, determine whether the process adjustment operation has achieved the expected effect. If 'Adjusted' is found..." speed "Within a reasonable range, but" Adjusted quality "The fact that the improvement did not occur as expected indicates that the impact of the process adjustment on the root cause of the anomaly may differ from expectations. Through the above matching verification, we can identify the characteristics of the impact of the process adjustment on the root cause of the anomaly, such as determining which step's adjustment did not achieve the expected results, or discovering new factors that interfered with the adjustment effect."

[0089] Step 143: Based on the influence features, perform confidence-weighted correction on the corresponding dynamic relationships in the process knowledge graph to generate updated dynamic relationships.

[0090] In this step, based on the impact characteristics of the identified process adjustment operations on the root causes of anomalies, the corresponding dynamic relationships in the process knowledge graph are weighted and adjusted according to confidence levels. If the process adjustment operation achieves the expected results, it indicates that the corresponding dynamic relationships in the knowledge graph are relatively reliable, and their confidence weights are appropriately increased; conversely, if the adjustment effect is not ideal, it indicates that the relationships may be inaccurate or incomplete, and their confidence weights are decreased. For example, regarding the equipment operating speed "Speed..." param "Quality" param The relationship between "Speed" and "Speed" needs to be adjusted. param "After" Quality param "Significant improvement was achieved, and the confidence weight of the association was increased from 0.8 to 0.9; if no improvement was achieved, the confidence weight was reduced to 0.6. Through the above confidence weighting correction, an updated dynamic association was generated, enabling the process knowledge graph to more accurately reflect the various relationships in the actual production process."

[0091] Step 144: Perform path backtracking verification between the updated dynamic correlation and the abnormal pattern analysis results to generate an updated automated monitoring report containing verification indicators of the adjustment effect.

[0092] It's understandable that the updated dynamic correlations are used to verify the path backtracking results against the anomaly pattern analysis results. The path and root cause of the abnormal event are determined from the anomaly pattern analysis results. Then, based on the updated dynamic correlations, it's checked whether this path is still reasonable and whether the root cause is still accurate. For example, if the anomaly pattern analysis results indicate that abnormal equipment operating speed is the root cause of product quality problems, and the updated dynamic correlations show a close correlation between the two that matches the actual adjusted situation, then the anomaly pattern analysis results are reliable to a certain extent. If a contradiction is found between the path or root cause and the updated dynamic correlations, the anomaly pattern analysis results need to be corrected.

[0093] By backtracking and verifying through the aforementioned path, the effectiveness of the process adjustment operations is comprehensively evaluated. Various indicators obtained during the verification process, such as the correction status of root causes of anomalies, changes in the confidence level of correlations, and the compliance status of adjusted equipment operating status and process indicators, are compiled into adjustment effectiveness verification indicators. These indicators are integrated into the automated monitoring report, generating an updated automated monitoring report that includes the adjustment effectiveness verification indicators. This updated report not only records the execution of the process adjustment operations but also provides in-depth analysis and verification of the adjustment effects, offering more valuable reference information for subsequent production management and optimization in the smart factory. It helps managers better understand problems and improvement directions in the production process, further improving production efficiency and product quality.

[0094] In a non-limiting embodiment, after feeding back the automated monitoring report to the smart factory control terminal to trigger a process adjustment operation, the method further includes: parsing the historical fault case text corresponding to the abnormal root cause inference from the automated monitoring report; comparing the feature similarity between the equipment operating status data after the process adjustment operation is executed and the solution execution results in the historical fault case text to obtain a comparison result; annotating the causal reasoning rules of the historical fault case text in the process knowledge graph according to the comparison result, generating annotated nodes containing case matching tags; and updating the weights of the semantic association paths in the process knowledge graph based on the annotated nodes to generate an optimized process knowledge graph.

[0095] In this embodiment of the invention, the automated monitoring report records in detail the inference of the root causes of anomalies and the corresponding historical failure case texts. First, these historical failure case texts are accurately parsed from the report. These texts contain detailed descriptions of similar past anomalies, the causes of the failures, the solutions implemented, and the final results. For example, a historical failure case text records a production interruption caused by excessive equipment pressure. The solution involved adjusting the equipment's pressure control system parameters and inspecting and maintaining the relevant pipelines, ultimately restoring the equipment to normal operation.

[0096] Optionally, after the process adjustment operation is executed, equipment operating status data, such as the actual pressure value and operational stability indicators, is acquired. This data is then compared with the solution execution results in historical failure case texts for feature similarity. The comparison considers multiple dimensions, such as changes in equipment operating parameters, the degree of recovery in production efficiency, and the stability of product quality. For the feature of equipment pressure, if the current adjusted pressure value is similar to the pressure value after the solution execution in historical failure cases, and if operational stability and product quality also show similar trends, then a high similarity is considered to exist for this feature. By comprehensively comparing the results of various features, a comprehensive comparison result is obtained, reflecting the degree of matching between the current process adjustment operation and the historical failure solution.

[0097] Furthermore, based on the comparison results, the causal reasoning rules of historical failure case texts are labeled in the process knowledge graph. If the comparison results show a high similarity, it indicates that the current anomaly is strongly correlated with the historical failure, and the causal reasoning rule of the historical failure case has high credibility. In this case, a case matching tag indicating a high degree of matching is assigned to the corresponding causal reasoning rule. Conversely, if the similarity is low, a low degree of matching tag is assigned. For example, for the aforementioned historical failure case of excessive equipment pressure, if the current equipment status after process adjustment has a high similarity to the execution result of the historical case solution, a "high degree of matching" tag is assigned to the causal reasoning rule of this case in the knowledge graph, generating a labeled node containing the case matching tag.

[0098] Furthermore, based on these labeled nodes, the weights of semantic association paths in the process knowledge graph are updated. For semantic association paths closely related to labeled nodes, if they are labeled with high matching degree, their weights are appropriately increased to highlight their importance in the knowledge graph; if they are labeled with low matching degree, the weights of the relevant paths are decreased. For example, if a semantic association path connects the current abnormal equipment entity, related process entities, and historical fault case nodes labeled with high matching degree, then the weight of that path is increased from 0.7 to 0.8. Through the above weight updates, the process knowledge graph can more accurately reflect the relationship between the current production situation and historical experience. The optimized process knowledge graph can provide more accurate knowledge support in subsequent anomaly analysis and processing, helping to improve the accuracy and efficiency of fault diagnosis.

[0099] In a non-limiting embodiment, after the automated monitoring report is fed back to the smart factory control terminal to trigger a process adjustment operation, the method further includes: acquiring multiple batches of equipment operating status data after the smart factory control terminal continuously executes the process adjustment operation; extracting parameter adjustment sequences from the multiple batches of equipment operating status data that correspond to the optimization operation guidelines in the automated monitoring report; matching entity nodes with the same dynamic relationship in the process knowledge graph according to the parameter adjustment sequence to generate an adjustment pattern set; and performing frequency statistics on the association edges between equipment entities and process entities in the process knowledge graph based on the adjustment pattern set to generate an optimization knowledge graph containing adjustment success probability labels.

[0100] First, after receiving an automation monitoring report and triggering a process adjustment operation, the smart factory control terminal will continuously execute multiple process adjustment operations to optimize the production process. During this process, it acquires multiple batches of equipment operating status data, covering the equipment's operation status across different time periods and production batches. For example, it records parameter values ​​such as operating speed, pressure, and temperature after each adjustment operation, forming a rich dataset of equipment operating status.

[0101] Secondly, from these multiple batches of equipment operating status data, the parameter adjustment sequences corresponding to the optimized operation guidelines in the automated monitoring reports were carefully extracted. The optimized operation guidelines in the automated monitoring reports clearly specify the process parameters that need to be adjusted for different abnormal situations, as well as the direction and magnitude of the adjustments. For example, the optimized operation guidelines might indicate that the equipment operating speed needs to be gradually adjusted within a certain range to improve product quality. The corresponding speed parameter adjustment records were found from the equipment operating status data and organized into a parameter adjustment sequence according to time.

[0102] Then, based on the parameter adjustment sequence, entity nodes with the same dynamic relationships are matched in the process knowledge graph. The process knowledge graph records equipment entities, process entities, and the dynamic relationships between them. For example, there is a certain relationship between equipment operating speed and product quality, which is represented by corresponding nodes and connecting edges in the knowledge graph. By analyzing the parameter values ​​and trends in the parameter adjustment sequence, matching equipment and process entity nodes are found in the knowledge graph. These nodes constitute the adjustment pattern set. For example, the equipment node "Device" related to speed parameter adjustment is found in the knowledge graph. M "and process nodes related to product quality" N The relationships between them are consistent with the relationships reflected in the current parameter adjustment sequence, and these nodes are included in the adjustment mode set.

[0103] Furthermore, based on this set of adjustment patterns, frequency statistics are performed on the association edges between equipment entities and process entities in the process knowledge graph. Association edges represent a certain relationship between equipment entities and process entities, such as how changes in equipment operating parameters affect the process execution result. The frequency of each association edge being involved in multiple process adjustment operations is counted. For example, in multiple adjustment operations, the equipment node "Device"... M "Process node" N The edge connecting the two was involved 10 times, while the other edge was involved only 3 times.

[0104] Finally, based on the frequency statistics, the success probability of adjustments is labeled for the associated edges in the process knowledge graph. If an associated edge is frequently involved and the process adjustment operation achieves good results, it indicates that the association played an important role in the adjustment process, and its success probability is high; conversely, if the association is involved infrequently and the adjustment effect is poor, the success probability is low. For example, for an associated edge that is involved 10 times and the product quality is significantly improved after adjustment, the success probability is labeled as 0.8; for an associated edge that is involved 3 times and the adjustment effect is not obvious, the success probability is labeled as 0.3. Through the above method, an optimized knowledge graph containing adjustment success probability labels is generated. This optimized knowledge graph can intuitively show the importance and effectiveness of the association between different equipment entities and process entities in process adjustment operations, providing a more intuitive and accurate basis for subsequent optimization decisions in smart factories, helping managers to better grasp the key factors and relationships in the production process, and further improving the stability and efficiency of production.

[0105] It should be noted that when implementing the above technical solutions, those skilled in the field can refactor the construction process of the process knowledge base and the monitoring rule base based on existing workflow engines (such as Apache Airflow). By defining the construction of the process knowledge base (steps 210-240) and the construction of the monitoring rule base (steps 310-350) as independent schedulable subtasks and setting their execution priority to be higher than that of the knowledge enhancement fusion processing in step 110, the potential risk of circular dependencies between steps can be eliminated.

[0106] To address the issue of inconsistent units, an industrial data standardization framework (such as the IEEE 1451 standard) can be introduced. In the multi-type sensor deployment stage of step 101, a data acquisition protocol is preset, which mandates that all sensor output data adopt the International System of Units (SI) (e.g., velocity is uniformly in m / s, pressure in Pa, and gas concentration in ppm). At the same time, in the unit standardization process of step 102, an automated unit conversion module based on ISO 80000 is embedded. By parsing the original unit identifier in the sensor metadata in real time and calling the standard unit conversion function library, the physical units of all parameters in the real-time monitoring data set are globally consistent.

[0107] To optimize terminology consistency, ontology building tools (such as Protégé) can be used to explicitly define the entity classification tree and attribute constraint rules of core concepts such as "equipment entity" and "process entity" during the process knowledge base construction process in steps 210-240. In the initial knowledge graph generation in step 111, the logical consistency of node relationships can be verified through the OWL ontology reasoning engine, thereby avoiding association confusion caused by ambiguous entity definitions.

[0108] To address the risk of causal rule conflicts, a rule engine (such as Drools) can be used to implement multi-rule logical conflict detection based on the Rete algorithm in the rule constraint set generation stage of step 330. By defining equipment operation safety constraints as the highest priority rule and process parameter adjustments as the second priority rule, the logical completeness of the dynamic correlation reasoning results can be ensured.

[0109] Regarding the optimization of the dynamic association verification mechanism, the A / B testing framework can be used to introduce control group data analysis in the path backtracking verification in step 144. The equipment operation status response data before and after process adjustment and the knowledge graph prediction values ​​are subjected to a two-sample T test. The effectiveness of the association update is verified by statistical significance difference analysis (p<0.05), thereby forming a logical closed loop.

[0110] Finally, to address the issue of mixed units for environmental parameters, a lightweight ETL tool can be deployed at the data acquisition interface layer of the environmental monitoring unit based on an industrial IoT platform (such as PTC ThingWorx). This tool can perform unit standardization preprocessing for environmental parameters such as temperature and humidity (converted to ℃ and %RH) and dust density (converted to μg / m3) before the data is entered into the database, thereby eliminating dimensional ambiguity in subsequent analysis at its source.

[0111] This invention utilizes a pre-built process knowledge base and monitoring rule base to perform knowledge-enhanced fusion processing on real-time monitoring data sets. This integrates scattered and isolated data into a structured process knowledge graph, enhancing the value and usability of the data. Anomaly pattern recognition based on semantic matching strategies allows for in-depth mining of the characteristics and semantic relationships of abnormal events, enabling accurate inference of root causes and improving the accuracy and comprehensiveness of anomaly analysis. Automated monitoring reports are generated based on the anomaly pattern analysis results and dynamic correlations, and fed back to the control terminal to trigger process adjustment operations. This provides timely and effective guidance for production adjustments, improving the production stability and efficiency of smart factories, ensuring continuous optimization of the production process, and enabling autonomous monitoring and intelligent decision-making in smart factories. Thus, more accurate anomaly analysis and more effective process adjustments can be achieved.

[0112] Based on the same inventive concept, embodiments of the present invention also provide an intelligent factory automation monitoring system. See also... Figure 2 As shown, it is a schematic diagram of a possible intelligent factory automation monitoring system provided in an embodiment of the present invention. Figure 2 In the intelligent factory automation monitoring system 200, a processor 210 and a memory 220 are included. The memory 220 stores computer programs that can be executed by the processor 210. By executing the instructions stored in the memory 220, the processor 210 can perform the steps of the aforementioned knowledge-based enhanced intelligent factory automation monitoring method.

[0113] Based on the same inventive concept, embodiments of the present invention provide a computer-readable storage medium including a computer program. When the computer program is run on a smart factory automation monitoring system, it causes the smart factory automation monitoring system to perform the steps of the aforementioned knowledge-based enhanced smart factory automation monitoring method. In some possible embodiments, various aspects of the knowledge-based enhanced smart factory automation monitoring method provided by the present invention can also be implemented in the form of a program product, including a computer program. When the program product is run on a smart factory automation monitoring system, the computer program causes the smart factory automation monitoring system to perform the steps of the aforementioned knowledge-based enhanced smart factory automation monitoring method. For example, the smart factory automation monitoring system can perform actions such as... Figure 1 The steps are shown in the figure.

[0114] In the technical solutions involved in the above embodiments of the present invention, whether performing comparison calculations of multi-dimensional features or constructing composite parameters, if there are problems caused by significant differences in the number of dimensions, units of measurement, and semantic meanings of different features, those skilled in the art, based on their professional knowledge and past practical experience, can fully understand that these differences need to be properly handled so that the calculation results are accurate and comparable, and to avoid situations such as logical confusion and unclear mathematical meaning.

[0115] In detail, when faced with features of different numbers of dimensions, those skilled in the art can employ various strategies to accurately calculate the similarity, matching degree, or feature distance between different features.

[0116] Feature selection is a common method. For high-dimensional feature sets, a subset of features that matches the number of low-dimensional features and is most representative can be selected based on indicators such as feature importance and relevance. By using methods such as chi-square test and information gain, the most valuable features for the technical solution can be screened out, thereby reducing the dimensionality of high-dimensional features to a level comparable to low-dimensional features, and then similarity or distance calculations can be performed.

[0117] Feature extraction is also an effective method. By constructing a suitable feature extraction model, features of different dimensions can be mapped to a common low-dimensional feature space. Principal Component Analysis (PCA) can not only handle differences in dimensions but also project high-dimensional features onto a low-dimensional space composed of principal components, making features of different dimensions comparable in this low-dimensional space. Furthermore, deep learning models such as autoencoders can also be used for feature extraction. They can automatically learn the latent representation of input features, transforming features of different dimensions into feature vectors of the same dimension for subsequent similarity, matching degree, or feature distance calculations.

[0118] Alternatively, kernel methods can be used. Kernel functions can calculate the similarity between features in a high-dimensional space without explicitly mapping features to that space. For features with different numbers of dimensions, appropriate kernel functions, such as Gaussian kernels or polynomial kernels, can be selected to directly calculate their similarity. This method avoids the computational difficulties caused by different feature dimensions and can effectively measure the relationships between features in the original feature space or the implicit high-dimensional space.

[0119] In order to achieve comparability alignment of feature spaces, those skilled in the art can use a variety of existing and common technical means when processing the comparison of multidimensional features.

[0120] Standardization preprocessing is a widely used and effective method that transforms raw feature data into a standard normal distribution with a mean of 0 and a standard deviation of 1 by applying a specific linear transformation. This process essentially eliminates the influence of different dimensions on the data, allowing all features to be compared on the same scale. For example, in a dataset containing features with different dimensions, after standardization preprocessing, these features can be used to calculate similarity or distance on the same scale, avoiding calculation bias caused by differences in dimensions.

[0121] Mapping transformation is also an effective way to solve the problem of dimensional differences. It can map the original features to a completely new space based on the specific properties of the features and actual business needs. In this new space, features with different dimensions can have better comparability. For features with nonlinear relationships, those skilled in the art can use logarithmic transformations, power transformations, etc., to convert them into linear relationships, thus facilitating similarity or distance calculations. For example, when dealing with features with exponential growth trends, logarithmic transformations can convert them into linear relationships, making subsequent calculations more accurate and convenient.

[0122] Spatial projection is also an important technique that projects a high-dimensional feature space onto a low-dimensional space while preserving as much important information as possible between features. By carefully selecting appropriate projection directions and dimensions, those skilled in the art can effectively reduce the impact of dimensional differences on computational results while reducing data dimensionality. Common spatial projection methods include Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA). Taking PCA as an example, it projects high-dimensional data into a low-dimensional space composed of principal components by finding the principal component directions, simplifying the data structure and reducing the interference of dimensional differences on feature comparison.

[0123] In the process of constructing composite parameters (such as loss function values), different parameter terms often have different dimensions. Those skilled in the art can use normalization processing or an adaptive weight allocation mechanism based on distribution characteristics.

[0124] Normalization unifies the value range of different parameter terms into a fixed interval, such as [0, 1]. This method eliminates the influence of differences in dimensions, ensuring that each parameter term has the same importance during weighted fusion. Common normalization methods include min-max normalization and Z-score normalization. Taking min-max normalization as an example, it performs a linear transformation on the parameter terms, scaling their value range to the [0, 1] interval, allowing parameter terms with different dimensions to be weighted and fused under the same standard.

[0125] An adaptive weight allocation mechanism based on distribution characteristics dynamically adjusts the weights of different parameter terms according to their distribution characteristics. For parameter terms with large variance, those skilled in the art can appropriately reduce their weights; for parameter terms with small variance, they can appropriately increase their weights. This allows the composite loss function to focus more on parameter terms with smaller variances, thereby improving the model's stability and generalization ability. For example, in a composite loss function containing multiple parameter terms, if the variance of a certain parameter term is large, it indicates that its fluctuations are relatively drastic, which may adversely affect the model's stability. In this case, reducing its weight can mitigate this adverse effect; conversely, for parameter terms with small variance, increasing their weights allows the model to focus more on the information reflected by that parameter term, thereby improving the overall performance of the model.

[0126] The aforementioned general techniques for solving feature matching and loss balance problems are all common knowledge in this field. These techniques have been fully verified and widely used in numerous practical applications, and those skilled in the art can skillfully and flexibly apply these methods to handle similar problems involving differences in dimensions.

[0127] The formulas and calculation processes involved in the embodiments of this invention, whether used for multidimensional feature comparison or composite loss function construction, strictly adhere to the principle of dimensional correspondence. Each variable in the formula has a clear and explicit physical meaning, and its operational logic fully conforms to basic mathematical and physical logic. The calculation results are necessarily the reasonable results expected by this invention. Those skilled in the art are capable of effectively solving various problems arising from the number of dimensions, dimensional differences, etc., in the multidimensional feature comparison calculation and composite loss function construction in the embodiments, based on specific data conditions and business needs, by comprehensively utilizing the above-mentioned general technical means, thus ensuring the accuracy, reliability, and implementability of the technical solution of this invention.

Claims

1. A knowledge-based enhanced intelligent factory automation monitoring method, characterized in that, include: By using a pre-built process knowledge base and monitoring rule base, knowledge enhancement and fusion processing is performed on the real-time monitoring data set of the smart factory to generate a process knowledge graph. Based on a semantic matching strategy, the process knowledge graph is subjected to abnormal pattern recognition processing to extract the feature descriptions of abnormal events and the semantic association paths with historical monitoring texts, and generate abnormal pattern analysis results containing abnormal root cause inferences. Based on the abnormal pattern analysis results and the dynamic correlation in the process knowledge graph, an automated monitoring report containing root cause priority ranking and optimization operation guidance is generated, and the automated monitoring report is fed back to the smart factory control terminal to trigger process adjustment operations.

2. The method as described in claim 1, characterized in that, The real-time monitoring data set includes equipment operating status data, process execution flow data, and environmental perception data. The steps for acquiring the real-time monitoring data set of the smart factory include: The system collects equipment operating status data by deploying various types of sensors on the equipment nodes of the smart factory. Call the process control interface of the process execution system to extract the process execution flow data of the current batch production task of the smart factory; Environmental sensing data, including temperature and humidity distribution data, gas concentration gradient data, and dust particle density data, are obtained from the environmental monitoring unit. The equipment operating status data, the process execution flow data, and the environmental perception data are processed by timestamp alignment and unit standardization to generate a real-time monitoring data set with a unified data format.

3. The method as described in claim 2, characterized in that, The process knowledge graph contains the dynamic correlation between equipment operating status and process execution standards. The process knowledge graph is generated by performing knowledge enhancement fusion processing on the acquired real-time monitoring data set of the smart factory through a pre-built process knowledge base and monitoring rule base, including: Extract equipment operating baseline parameters, process execution standard documents, and historical fault case texts from the process knowledge base to construct an initial knowledge graph containing equipment entities, process entities, and environmental entities; The real-time monitoring data set is input into a pre-trained knowledge enhancement model to simultaneously identify abnormal fluctuation patterns in the equipment operating status data, process deviation events in the process execution data, and environmental risk characteristics in the environmental perception data. Based on the association rules in the monitoring rule base, causal reasoning is performed on the abnormal fluctuation pattern, the process deviation event and the environmental risk characteristics to generate a dynamic correlation between equipment operating status and process execution standards. The dynamic relationships are incrementally updated and then integrated into the initial knowledge graph to generate a process knowledge graph that integrates real-time monitoring data and historical knowledge.

4. The method as described in claim 3, characterized in that, The process knowledge graph is subjected to anomaly pattern recognition processing based on a semantic matching strategy. Feature descriptions of abnormal events and semantic association paths with historical monitoring texts are extracted, generating anomaly pattern analysis results containing root cause inferences, including: Based on the dynamic relationships in the process knowledge graph, a semantic matching template containing equipment operating status thresholds is constructed. The current equipment operating status data in the real-time monitoring data set is compared with the semantic matching template to calculate the similarity and identify abnormal data segments that deviate from the process execution standards. The historical monitoring text database is invoked to retrieve historical fault records that have semantic similarity to the abnormal data fragment, and a feature description of the abnormal event is generated. Based on the entity association paths in the process knowledge graph, the propagation trajectory of the abnormal data fragment in the equipment operation link is traced to generate a semantic association path containing multiple levels of influence. The feature descriptions and semantic association paths are logically integrated to generate anomaly pattern analysis results that include anomaly root cause inference.

5. The method as described in claim 4, characterized in that, The process generates an automated monitoring report based on the abnormal pattern analysis results and the dynamic correlation in the process knowledge graph, including root cause priority ranking and optimization operation guidance. Based on the causal reasoning results in the dynamic relationship, and combined with the semantic relationship path length and the number of affected entities in the abnormal pattern analysis results, the comprehensive influence weight of each abnormal root cause is calculated. Based on the strength of interaction between entities mapped by the dynamic relationship, the comprehensive influence weight is adjusted to generate the root cause priority ranking. Optimization operation rules that match the root cause priority ranking and the dynamic correlation are extracted from the monitoring rule base to generate optimization operation guidelines that include process parameter adjustment thresholds and equipment maintenance condition constraints. The root cause priority sorting and the optimization operation guidance are mapped and recombined according to the entity topology structure of the dynamic relationship to generate an automated monitoring report containing multi-dimensional optimization paths.

6. The method as described in claim 5, characterized in that, The step of feeding back the automated monitoring report to the smart factory control terminal to trigger process adjustment operations includes: The process parameter adjustment thresholds corresponding to the dynamic correlation in the automated monitoring report are analyzed to generate process control instructions containing the target adjustment range and execution sequence. A maintenance task instruction set is generated based on the equipment maintenance condition constraints, and the process control instructions and the maintenance task instruction set are synchronously transmitted to the intelligent factory control terminal. Based on the dynamic correlation, a process parameter adjustment prediction model is constructed, and the equipment operation status response data fed back by the intelligent factory control terminal is monitored in real time. When the deviation between the equipment operation status response data and the process parameter adjustment prediction model exceeds a preset threshold, the process knowledge graph is triggered to backtrack and update the association relationship, and an automated monitoring report with priority correction is regenerated.

7. The method as described in claim 3, characterized in that, The process of constructing the process knowledge base includes: Collect historical production data, equipment maintenance logs, and process guidance manuals from the smart factory; parse historical statistical values ​​of equipment operating parameters from the historical production data and extract equipment operating baseline parameters; extract process execution standard documents from the structured text of the process guidance manual; perform event sequence analysis on the equipment maintenance logs and extract fault handling case texts containing fault phenomena and solutions; Entity identification and relationship extraction are performed on the equipment operating baseline parameters to generate static association relationships between equipment entities and process entities; Based on the event chain decomposition results of the fault handling case text, and combined with the constraints in the process execution standard document, causal reasoning rules between fault occurrence conditions and handling measures are constructed. The static association relationship is fused with the causal reasoning rule to generate the process knowledge base.

8. The method as described in claim 3, characterized in that, The process of constructing the monitoring rule base includes: Collect historical fault work orders and process execution records; perform event annotation and correlation parsing on the historical fault work orders to generate a preprocessed data set containing fault triggering conditions and handling actions; extract the process sequence chain and parameter fluctuation range from the process execution records, and generate a set of rule elements in combination with equipment operation constraints; Based on the fault triggering conditions in the preprocessed data set and the parameter fluctuation range in the rule element set, the spatiotemporal correlation between abnormal fluctuation patterns and equipment operation is analyzed through the event sequence pattern mining algorithm to generate causal rules. Logical conflict detection is performed on the causal rules and the device operation constraints are injected to generate a set of rule constraints that includes parameter deviation tolerance and operation restrictions; The causal rules and the set of rule constraints are input into the rule reasoning module, and the rule triggering conditions are verified by combining real-time device status data. Invalid rules are eliminated to form an initial monitoring rule library. Based on the environmental data stream, the rule trigger frequency and anomaly coverage of the initial monitoring rule base are matched, the rule weights of the initial monitoring rule base are dynamically adjusted and incrementally updated to generate the monitoring rule base.

9. The method as described in claim 1, characterized in that, After the automated monitoring report is fed back to the smart factory control terminal to trigger process adjustment operations, the method further includes: Based on the execution results of the process adjustment operation fed back by the intelligent factory control terminal, the target operating status response data after adjustment is extracted; The target operating status response data is matched and verified with the dynamic correlation in the process knowledge graph to identify the impact characteristics of the process adjustment operation on the root cause of the anomaly. Based on the aforementioned impact characteristics, the corresponding dynamic relationships in the process knowledge graph are corrected with confidence weighting to generate updated dynamic relationships. The updated dynamic correlation is verified by path backtracking with the abnormal pattern analysis results, and an updated automated monitoring report containing verification indicators of the adjustment effect is generated.

10. An intelligent factory automation monitoring system, characterized in that, It includes a processor and a memory, wherein the memory stores a computer program that, when executed by the processor, causes the processor to perform the steps of any one of the methods described in claims 1 to 9.

Images