Industrial equipment fault reasoning system based on knowledge graph

Abstract

The invention discloses an industrial equipment fault inference system based on a knowledge graph, and the system comprises a knowledge graph construction module, a fault data collection module, an inference analysis module and a response processing module. The entity extraction unit extracts equipment components, fault types and maintenance record entities from an industrial equipment operation document, equipment manual unstructured data supplementation attributes are integrated, the attributes and association weights are marked, and the relationship construction unit establishes a fault causal relationship between the entities and a component association relationship to form a multi-level knowledge network; a knowledge verification unit verifies entity attribute consistency and relation rationality, a dynamic updating unit receives data updating nodes and relation strength of each unit and receives feedback data optimization weights, and in a fault data acquisition module, a real-time monitoring unit acquires operation parameters and state signals and associates equipment identifiers.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of equipment failure reasoning, in particular to an industrial equipment failure reasoning system based on a knowledge graph. BACKGROUND

[0002] Industrial equipment failure reasoning is an important part of industrial production and operation, involving industrial equipment operation data collection, fault mode identification, diagnosis logic construction and other technical links. Its core goal is to quickly identify fault types and locate fault sources through analysis of equipment operation status, to provide accurate guidance for equipment maintenance, to reduce downtime and maintenance costs, and to ensure stable operation of large and complex equipment in various industrial scenarios.

[0003] In the prior art, the system or method for industrial equipment failure reasoning has obvious deficiencies in knowledge system construction. It often cannot fully extract key entities such as equipment components, fault types and maintenance records, and the integration and attribute supplementation of unstructured data such as equipment manuals are not sufficient. The weight annotation of the association between entities is missing. The constructed knowledge network is simple in level and lacks effective consistency verification mechanism. At the same time, it is difficult to dynamically update nodes and relationship strength according to real-time data changes, resulting in one-sided and inaccurate knowledge base, which cannot provide reliable support for failure reasoning. In the reasoning analysis link, the existing method cannot effectively match real-time operation data with knowledge graph entities. The fault source positioning relies on single path search, lacks cross verification with historical cases, and cannot dynamically correct the reasoning path according to real-time data fluctuations. It is also difficult to combine historical probability and abnormal degree to quantify the possibility of fault occurrence, making the fault reasoning accuracy not high and the reliability insufficient, which makes it difficult to accurately lock the fault source and clearly present the fault probability. Therefore, an industrial equipment failure reasoning system based on a knowledge graph is proposed. SUMMARY

[0004] In view of the deficiencies of the prior art, the present application provides an industrial equipment failure reasoning system based on a knowledge graph to solve the problem of low accuracy and insufficient reliability of failure reasoning, and to accurately lock the fault source and clearly present the fault probability.

[0005] To achieve the above purpose, the present application provides the following technical scheme: an industrial equipment failure reasoning system based on a knowledge graph, comprising: a knowledge graph construction module: The entity extraction unit extracts equipment components, fault types and maintenance record entities from the industrial equipment operation document, integrates device manual unstructured data to supplement attributes, labels attributes and associated weights, and transmits to the relationship construction unit; the relationship construction unit establishes the fault causal relationship and component association relationship between entities to form a multi-level knowledge network; the knowledge verification unit checks the consistency of entity attributes and the rationality of relationships, and connects the entity extraction unit and the relationship construction unit; the dynamic updating unit receives data updates of nodes and relationship strength from each unit, and receives feedback data to optimize the weight; The fault data collection module includes: The real-time monitoring unit collects operation parameters and state signals and associates with equipment identification; the multi-dimensional collection unit supplements environmental parameters and operation records; the abnormal preprocessing sub-unit filters noise and marks abnormal values; the historical data unit stores historical fault records and aligns with real-time data timing; The reasoning analysis module includes: The knowledge matching unit matches real-time data with knowledge graph entities; the causal reasoning unit locates the fault root through multi-level path search; the reasoning path verification sub-unit cross- validates the consistency of reasoning path and historical cases; the dynamic reasoning adjustment unit corrects the reasoning path according to real-time data fluctuations; the probability calculation unit generates fault probability values by combining historical probabilities and abnormal degrees; The response processing module includes: The fault warning unit issues graded warnings based on probability values; the scheme recommendation unit generates processing suggestions by matching fault types, and outputs after adjustment by the scheme adaptation sub-unit based on the current state of the equipment; the scheme effect evaluation unit evaluates the effectiveness of the processing results and feeds back to the dynamic updating unit; the record unit stores the reasoning process and results and feeds back to the dynamic updating unit.

[0006] The entity extraction unit extracts equipment components, fault types and maintenance record entities from the industrial equipment operation document, integrates device manual unstructured data to supplement attributes, labels attributes and associated weights, and transmits to the relationship construction unit; the relationship construction unit establishes the fault causal relationship and component association relationship between entities to form a multi-level knowledge network; the knowledge verification unit checks the consistency of entity attributes and the rationality of relationships, and connects the entity extraction unit and the relationship construction unit, realizes the verification and feedback of the work achievements of the first two units; the dynamic updating unit receives data from the entity extraction unit, the relationship construction unit and the knowledge verification unit to update the nodes and relationship strength in the knowledge graph, and receives feedback data to optimize the associated weight; The real-time monitoring unit collects the operating parameters and state signals of the industrial equipment and associates these data with the equipment identification; the multi-dimensional acquisition unit supplements the collection of environmental parameters and operation records to enrich the dimensions of the fault data; the abnormal pre-processing sub-unit processes the collected data, filters the noise therein and marks the abnormal values; the historical data unit stores the historical fault records and aligns them in time sequence with the real-time data; The knowledge matching unit matches the real-time collected and processed data with the entities in the knowledge graph; the causal reasoning unit locates the fault root cause by path searching on the multi-level knowledge network; the reasoning path verification sub-unit cross- verifies the reasoning path with the historical cases to ensure the consistency of the reasoning path; the dynamic reasoning adjustment unit corrects the reasoning path according to the fluctuation of the real-time data; the probability calculation unit generates the fault probability value in combination with the historical fault probability and the abnormal degree of the data; The fault early warning unit issues a hierarchical early warning according to the fault probability value generated by the probability calculation unit; the scheme recommendation unit matches the fault type and generates a processing suggestion, which is output after adjustment by the scheme adaptation sub-unit in combination with the current state of the equipment; the scheme effect evaluation unit evaluates the effectiveness of the processing result and feeds back the evaluation result to the dynamic updating unit of the knowledge graph construction module; the record unit stores the reasoning process and result and feeds them back to the dynamic updating unit of the knowledge graph construction module.

[0007] Preferably, the knowledge verification unit is provided with an attribute verification component which compares the attribute descriptions of the same entity in different data sources, marks the conflict information and triggers the manual verification process.

[0008] The attribute verification component compares the attribute descriptions of the same entity in different data sources, and when conflict information is found, triggers the manual verification process.

[0009] Preferably, the multi-dimensional acquisition unit is provided with an environment perception component which collects the temperature, humidity and vibration interference data in the equipment operating environment and associates them to the corresponding equipment identification.

[0010] The environment perception component collects the temperature, humidity and vibration interference data in the equipment operating environment and associates these collected data with the corresponding equipment identification.

[0011] Preferably, the environment perception component compares the environmental parameters with the normal operating threshold of the equipment in real time, and when the threshold is exceeded, the collection frequency of the corresponding parameters is enhanced.

[0012] The environment perception component compares the collected environmental parameters with the normal operating threshold of the equipment in real time, and when the threshold is exceeded, the collection frequency of the corresponding parameters is enhanced.

[0013] Preferably, the dynamic inference adjustment unit is provided with a fluctuation analysis component to identify the burst fluctuation characteristics of real-time data, and shorten the hierarchical range of the inference path search according to the identified characteristics.

[0014] The fluctuation analysis component identifies the burst fluctuation characteristics of real-time data, and shortens the hierarchical range of the inference path search according to the identified characteristics.

[0015] Preferably, the fluctuation analysis component takes the fluctuation mode before the historical failure of the equipment as a reference, and starts the fast inference mode when the real-time fluctuation characteristics match.

[0016] The fluctuation analysis component takes the fluctuation mode before the historical failure of the equipment as a reference, and starts the fast inference mode when the real-time fluctuation characteristics match.

[0017] Preferably, the scheme effect evaluation unit is provided with an effect quantification component to quantitatively process the effectiveness of the treatment scheme from the angles of failure solution time and recurrence rate.

[0018] The effect quantification component quantitatively processes the effectiveness of the treatment scheme from the angles of failure solution time and recurrence rate.

[0019] Preferably, the effect quantification component divides the evaluation results into three levels of effective, partially effective and ineffective, which correspond to different weight adjustment amplitudes respectively.

[0020] The effect quantification component divides the evaluation results into three levels of effective, partially effective and ineffective, which correspond to different weight adjustment amplitudes respectively.

[0021] Preferably, the historical data unit is provided with a data cleaning component that can periodically clean up repeated records and invalid data, and retain key information related to fault inference.

[0022] The data cleaning component periodically cleans up repeated records and invalid data, and retains key information related to fault inference.

[0023] Preferably, the data cleaning component deletes redundant data irrelevant to the core entities of the knowledge graph according to the entity correlation degree.

[0024] The data cleaning component deletes redundant data irrelevant to the core entities of the knowledge graph according to the entity correlation degree.

[0025] In summary, compared with the prior art, the industrial equipment fault inference system based on the knowledge graph has the following beneficial effects: The application extracts equipment components, fault types and maintenance record entities from industrial equipment operation documents through an entity extraction unit in a knowledge graph construction module, integrates device manual unstructured data to supplement attributes, and labels attributes and correlation weights, cooperates with a relationship construction unit to establish fault cause and effect relationships and component correlation relationships between entities to form a multi-level knowledge network, and then verifies entity attribute consistency and relationship rationality through a knowledge verification unit, simultaneously receives data update nodes and relationship strengths from each unit through a dynamic update unit, and receives feedback data to optimize weights, so that a comprehensive, accurate and dynamically optimized industrial equipment knowledge system can be constructed to provide a solid knowledge foundation for fault reasoning; The knowledge matching unit in the reasoning analysis module matches real-time data with knowledge graph entities, the cause and effect reasoning unit locates fault roots through multi-level path searching, the reasoning path verification subunit cross- verifies reasoning path consistency with historical cases, the dynamic reasoning adjustment unit corrects reasoning paths according to real-time data fluctuations, and the probability calculation unit generates fault probability values in combination with historical probabilities and abnormality degrees, so that precise, reliable and dynamically adaptive fault reasoning analysis can be realized to accurately locate fault roots and quantify fault likelihoods. BRIEF DESCRIPTION OF DRAWINGS

[0026] Figure 1 is a system schematic diagram of the application. DETAILED DESCRIPTION

[0027] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the application.

[0028] Please refer to Figure 1 The application provides a technical solution, an industrial equipment fault reasoning system based on a knowledge graph, comprising: A knowledge graph construction module: An entity extraction unit extracts equipment components, fault types and maintenance record entities from industrial equipment operation documents, integrates device manual unstructured data to supplement attributes, labels attributes and correlation weights, and then transmits the same to a relationship construction unit; the relationship construction unit establishes fault cause and effect relationships and component correlation relationships between entities to form a multi-level knowledge network; a knowledge verification unit verifies entity attribute consistency and relationship rationality, and connects the entity extraction unit and the relationship construction unit; and a dynamic update unit receives data update nodes and relationship strengths from each unit, and receives feedback data to optimize weights; A fault data acquisition module: The real-time monitoring unit collects operation parameters and state signals, and associates the device identification; the multi-dimensional acquisition unit supplements environmental parameters and operation records; the abnormality preprocessing sub-unit filters noise and marks abnormal values; the historical data unit stores historical fault records and aligns them with real-time data timing; The reasoning analysis module: The knowledge matching unit matches real-time data with knowledge graph entities; the causal reasoning unit locates the root cause of the fault through multi-level path search; the reasoning path verification sub-unit cross- validates the consistency of the reasoning path with historical cases; the dynamic reasoning adjustment unit corrects the reasoning path according to real-time data fluctuations; the probability calculation unit generates a fault probability value by combining historical probabilities and abnormality degrees; The response processing module: The fault warning unit issues a hierarchical warning based on the probability value; the scheme recommendation unit generates a processing suggestion by matching the fault type, which is adjusted by the scheme adaptation sub-unit based on the current state of the device and then output; the scheme effect evaluation unit evaluates the effectiveness of the processing result and feeds back to the dynamic updating unit; the record unit stores the reasoning process and result and feeds back to the dynamic updating unit.

[0029] The entity extraction unit extracts device components, fault types, and maintenance record entities from industrial device operation documents, and integrates unstructured data in device manuals to supplement entity attributes. After labeling entity attributes and associated weights, the relevant information is transmitted to the relationship construction unit. The relationship construction unit receives information from the entity extraction unit, establishes fault causal relationships and component association relationships between entities, and forms a multi-level knowledge network. The knowledge verification unit checks the consistency of entity attributes and the rationality of relationships, and is connected to the entity extraction unit and the relationship construction unit to verify and feedback the work results of the previous two units. The dynamic updating unit receives data from the entity extraction unit, the relationship construction unit, and the knowledge verification unit to update nodes and relationship strengths in the knowledge graph, and receives feedback data to optimize the associated weights; The real-time monitoring unit collects operation parameters and state signals, and associates the device identification; the multi-dimensional acquisition unit supplements environmental parameters and operation records, to enrich the dimensions of fault data; the abnormality preprocessing sub-unit processes the collected data, filters noise, and marks abnormal values; the historical data unit stores historical fault records, and aligns them with real-time data timing; The knowledge matching unit matches the data collected and processed in real time with entities in the knowledge graph; the causal reasoning unit locates the root cause of the fault by searching the path of the multi-level knowledge network; the reasoning path verification subunit cross- verifies the reasoning path with historical cases to ensure the consistency of the reasoning path; the dynamic reasoning adjustment unit adjusts the reasoning path according to the fluctuation of real-time data; and the probability calculation unit generates a fault probability value in combination with historical fault probability and the abnormality degree of data; The fault warning unit issues a hierarchical warning according to the fault probability value generated by the probability calculation unit; the scheme recommendation unit matches the fault type and generates a processing suggestion, which is output after being adjusted by the scheme adaptation subunit in combination with the current state of the equipment; the scheme effect evaluation unit evaluates the effectiveness of the processing result and feeds back the evaluation result to the dynamic updating unit of the knowledge graph construction module; and the recording unit stores the reasoning process and result and feeds them back to the dynamic updating unit of the knowledge graph construction module; The knowledge graph construction module constructs an accurate and dynamically optimized multi-level knowledge network through entity extraction, relationship construction, knowledge verification and dynamic updating, provides a reliable knowledge base for industrial equipment fault reasoning, and ensures the scientificity and timeliness of the reasoning basis; The fault data acquisition module comprehensively and effectively acquires equipment operation related data through real-time monitoring, multi-dimensional acquisition, abnormality preprocessing and historical data storage, ensures the integrity, accuracy and timeliness of the data used for reasoning analysis, and provides high-quality data support for subsequent fault reasoning; The reasoning analysis module can accurately locate the root cause of the fault, generate a reasonable fault probability value, improve the accuracy and reliability of fault reasoning, and help to quickly discover equipment faults, by means of knowledge matching, causal reasoning, reasoning path verification, dynamic reasoning adjustment and probability calculation; The response processing module can issue a fault warning in time, provide appropriate processing suggestions, and continuously optimize the knowledge graph through a feedback mechanism, thereby improving the response speed and processing effect of the system on equipment faults, realizing closed-loop management of fault processing, and being conducive to improving the operation reliability and maintenance efficiency of industrial equipment.

[0030] The knowledge verification unit is provided with an attribute verification component which compares the attribute descriptions of the same entity in different data sources, marks the conflict information and triggers the manual verification process.

[0031] The attribute verification component compares the attribute descriptions of the same entity in different data sources, and when conflict information is found, triggers the manual verification process; The attribute difference of the same entity in different data sources can be found in time, and through triggering the manual verification process, the accuracy and consistency of the entity attribute information can be ensured, and a reliable data basis is provided for subsequent related processing.

[0032] The multi-dimensional acquisition unit is provided with an environment perception component, which acquires temperature, humidity and vibration interference data in the equipment running environment and is associated with the corresponding equipment identifier.

[0033] The environment perception component acquires temperature, humidity and vibration interference data in the equipment running environment and associates these acquired data with the corresponding equipment identifier. The multi-dimensional acquisition unit is provided with an environment perception component, which acquires temperature, humidity and vibration interference data in the equipment running environment and is associated with the corresponding equipment identifier.

[0034] The environment perception component compares the environmental parameters with the normal running threshold of the equipment in real time, and increases the acquisition frequency of the corresponding parameters when the threshold is exceeded.

[0035] The environment perception component compares the acquired environmental parameters with the normal running threshold of the equipment in real time, and increases the acquisition frequency of the corresponding parameters when the threshold is exceeded. By comparing the environmental parameters with the normal running threshold in real time, the abnormal situation of the environmental parameters can be detected in time, the acquisition frequency of the abnormal parameters is increased, the parameter changes can be captured in more detail, and the abnormal development trend of the environmental parameters can be grasped in time, thereby providing more timely information for the maintenance and fault prevention of the equipment.

[0036] The dynamic reasoning adjustment unit is provided with a fluctuation analysis component, which identifies the sudden fluctuation characteristics of real-time data and shortens the hierarchical range of reasoning path search accordingly.

[0037] The fluctuation analysis component identifies the sudden fluctuation characteristics of real-time data and shortens the hierarchical range of reasoning path search according to the identified characteristics. The sudden fluctuation of real-time data can be quickly identified, the efficiency of reasoning path search can be improved by shortening the hierarchical range of reasoning path search, the speed of dynamic reasoning adjustment can be accelerated, and the system can respond to data fluctuation more quickly.

[0038] The fluctuation analysis component takes the fluctuation mode before the historical failure of the equipment as a reference, and starts the fast reasoning mode when the real-time fluctuation characteristics match.

[0039] The fluctuation analysis component takes the fluctuation mode before the historical failure of the equipment as a reference, and starts the fast reasoning mode when the real-time fluctuation characteristics match. With the fluctuation mode before the historical failure of the device as a reference, the real-time fluctuation characteristics are matched to start the fast reasoning mode, which helps to react to the possible failure in advance and improves the timeliness of failure reasoning, and gains more time for failure handling.

[0040] The scheme effect evaluation unit is provided with an effect quantification component to quantify the effectiveness of the processing scheme from the angles of failure resolution time and recurrence rate.

[0041] The effect quantification component quantifies the effectiveness of the processing scheme from the angles of failure resolution time and recurrence rate. Quantifying the effectiveness of the processing scheme from the angles of failure resolution time and recurrence rate can objectively and comprehensively evaluate the actual effect of the processing scheme and provide a clear basis for judging the pros and cons of the scheme.

[0042] The effect quantification component divides the evaluation results into three levels of effective, partially effective and ineffective, which correspond to different weight adjustment amplitudes.

[0043] The effect quantification component divides the evaluation results into three levels of effective, partially effective and ineffective, which correspond to different weight adjustment amplitudes. By grading the evaluation results and corresponding different weight adjustment amplitudes, the weight adjustment is more targeted and reasonable, which helps to optimize the application of the processing scheme by the system and improve the overall performance of the system.

[0044] The historical data unit is provided with a data cleaning component that can periodically clean up duplicate records and invalid data and retain key information related to failure reasoning.

[0045] The data cleaning component periodically cleans up duplicate records and invalid data and retains key information related to failure reasoning. Periodically cleaning up duplicate and invalid data reduces data redundancy while retaining key information, which helps to improve the quality of historical data and make historical data better support failure reasoning.

[0046] The data cleaning component deletes redundant data irrelevant to the core entities of the knowledge graph based on entity correlation degree.

[0047] The data cleaning component deletes redundant data irrelevant to the core entities of the knowledge graph based on entity correlation degree. Deleting redundant data based on entity correlation degree can accurately remove data irrelevant to the core entities, further simplify the data volume, improve the data processing efficiency, while ensuring the correlation of the retained data with the core entities of the knowledge graph and improving the practical value of the data.

[0048] The scheme: the entity extraction unit of the knowledge graph construction module extracts equipment components, fault types and maintenance record entities from the industrial equipment operation document, integrates the unstructured data of the equipment manual to supplement the attributes, labels the attributes and associated weights, and transmits them to the relationship construction unit; the relationship construction unit receives information from the entity extraction unit, establishes the fault causal relationship and component association relationship between entities to form a multi-level knowledge network; the knowledge verification unit verifies the consistency of entity attributes and the rationality of relationships, and is connected with the entity extraction unit and the relationship construction unit respectively to realize the verification and feedback of the work results of the previous two units; the attribute verification component of the knowledge verification unit compares the attribute descriptions of the same entity in different data sources, marks the conflict information and triggers the manual verification process; the dynamic updating unit receives data from the entity extraction unit, the relationship construction unit and the knowledge verification unit to update the nodes and relationship strength in the knowledge graph, and receives feedback data to optimize the associated weight; The real-time monitoring unit of the fault data acquisition module collects the operation parameters and state signals of the industrial equipment, and associates these data with the equipment identification; the multi-dimensional acquisition unit supplements the collection of environmental parameters and operation records to enrich the dimensions of fault data; the environmental perception component of the multi-dimensional acquisition unit collects temperature, humidity and vibration interference data in the equipment operating environment and associates them to the corresponding equipment identification; the environmental perception component compares the collected environmental parameters with the normal operation threshold of the equipment in real time, and increases the collection frequency of the corresponding parameters when the threshold is exceeded; the abnormal preprocessing sub-unit processes the collected data, filters the noise and marks the abnormal values; the historical data unit stores historical fault records and aligns them with real-time data in time sequence; the data cleaning component of the historical data unit can periodically clean up duplicate records and invalid data, and retain key information related to fault reasoning; the data cleaning component deletes redundant data that has no association with the core entities of the knowledge graph based on the entity association degree; The knowledge matching unit of the reasoning analysis module matches the data collected and processed in real time with the entities in the knowledge graph; the causal reasoning unit locates the fault root cause by searching the path of the multi-level knowledge network; the reasoning path verification sub-unit cross- verifies the reasoning path with historical cases to ensure the consistency of the reasoning path; the dynamic reasoning adjustment unit adjusts the reasoning path according to the fluctuation of real-time data; the fluctuation analysis component of the dynamic reasoning adjustment unit identifies the sudden fluctuation characteristics of real-time data, and accordingly shortens the hierarchical range of the reasoning path search; when the real-time fluctuation characteristics match the reference mode before the historical fault of the equipment, the fluctuation analysis component starts the fast reasoning mode; the probability calculation unit combines the historical fault probability and the abnormal degree of data to generate a fault probability value; The fault early warning unit of the response processing module sends a hierarchical early warning according to the fault probability value generated by the probability calculation unit; the scheme recommendation unit matches the fault type and generates a processing suggestion, which is output after being adjusted by the scheme adaptation subunit in combination with the current state of the equipment; the scheme effect evaluation unit evaluates the effectiveness of the processing result and feeds back the evaluation result to the dynamic updating unit of the knowledge graph construction module; the effect quantification component provided by the scheme effect evaluation unit quantifies the effectiveness of the processing scheme from the perspectives of fault solving time and recurrence rate; the effect quantification component divides the evaluation result into three levels of effective, partially effective and ineffective, which correspond to different weight adjustment amplitudes; the recording unit stores the reasoning process and result and feeds them back to the dynamic updating unit of the knowledge graph construction module; The knowledge graph construction module constructs an accurate and dynamically optimized multi-level knowledge network through entity extraction, relationship construction, knowledge verification and dynamic updating, provides a reliable knowledge base for industrial equipment fault reasoning, and ensures the scientificity and timeliness of the reasoning basis; The fault data acquisition module comprehensively and effectively acquires equipment operation related data through real-time monitoring, multi-dimensional acquisition, abnormal preprocessing and historical data storage, ensures the integrity, accuracy and timeliness of the data used for reasoning analysis, and provides high-quality data support for subsequent fault reasoning; The reasoning analysis module can accurately locate the fault root cause, generate a reasonable fault probability value, improve the accuracy and reliability of fault reasoning, and help quickly find equipment faults, with the help of knowledge matching, causal reasoning, reasoning path verification, dynamic reasoning adjustment and probability calculation; The response processing module can timely issue a fault early warning, provide appropriate processing suggestions, and continuously optimize the knowledge graph through a feedback mechanism, thereby improving the response speed and processing effect of the system on equipment faults, realizing closed-loop management of fault processing, and being conducive to improving the operation reliability and maintenance efficiency of industrial equipment.

[0049] It should be noted that, in this text, relational terms such as first and second are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "contain" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or equipment.

[0050] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary of the principles and application of the present application. Numerous modifications and adaptions can be effected without departing from the spirit and scope of the present application, which is not limited to the exact construction and arrangement described. It is intended, therefore, to cover all modifications and adaptions that fall within the scope of the claims and their equivalents.

Claims

1. A knowledge graph based industrial equipment failure reasoning system, characterized in that, Comprise: A knowledge graph construction module: An entity extraction unit extracts equipment components, fault types, and maintenance record entities from industrial equipment operation documents, integrates equipment manual unstructured data to supplement attributes, labels attributes and associated weights, and transmits them to a relationship construction unit; the relationship construction unit establishes fault cause-and-effect relationships and component association relationships between entities to form a multi-level knowledge network; a knowledge verification unit checks the consistency of entity attributes and the rationality of relationships, connecting the entity extraction and relationship construction units; a dynamic update unit receives data updates for nodes and relationship strengths, and receives feedback data to optimize weights; A fault data collection module: A real-time monitoring unit collects operating parameters and status signals and associates them with equipment identifiers; a multi-dimensional collection unit supplements environmental parameters and operation records; an abnormality preprocessing subunit filters noise and marks abnormal values; a historical data unit stores historical fault records and aligns them with real-time data time series; A reasoning analysis module: A knowledge matching unit matches real-time data with knowledge graph entities; a causal reasoning unit locates fault roots through multi-level path searching; a reasoning path verification subunit cross- validates reasoning paths with historical cases for consistency; a dynamic reasoning adjustment unit corrects reasoning paths based on real-time data fluctuations; a probability calculation unit generates fault probability values by combining historical probabilities and abnormality levels; A response processing module: A fault warning unit issues graded warnings based on probability values; a scheme recommendation unit generates processing recommendations by matching fault types, which are adjusted by a scheme adaptation subunit based on the current state of the equipment and then output; a scheme effectiveness evaluation unit evaluates the effectiveness of the processing results and feeds back to the dynamic update unit; A record unit stores the reasoning process and results and feeds them back to the dynamic update unit. 2.The knowledge graph based industrial equipment failure reasoning system of claim 1, wherein: The knowledge verification unit is provided with an attribute verification component that compares attribute descriptions of the same entity in different data sources, marks conflicting information, and triggers an artificial verification process.

3. The knowledge graph based industrial equipment failure reasoning system of claim 1, wherein: The multi-dimensional collection unit is provided with an environmental perception component that collects temperature, humidity, and vibration interference data in the equipment operating environment and associates them with corresponding equipment identifiers.

4. The knowledge graph based industrial equipment failure reasoning system of claim 3, wherein: The environmental perception component compares environmental parameters with equipment normal operating thresholds in real time, and increases the collection frequency of corresponding parameters when the thresholds are exceeded.

5. The knowledge graph based industrial equipment failure reasoning system of claim 1, wherein: The dynamic reasoning adjustment unit is provided with a fluctuation analysis component that identifies sudden fluctuation characteristics of real-time data, thereby shortening the hierarchical range of reasoning path searching.

6. The knowledge graph based industrial equipment failure reasoning system of claim 5, wherein: The fluctuation analysis component uses the fluctuation pattern before the historical fault of the equipment as a reference, and starts a fast reasoning mode when the real-time fluctuation characteristics match.

7. The knowledge graph based industrial equipment failure reasoning system of claim 1, wherein: The scheme effectiveness evaluation unit is provided with an effectiveness quantification component that quantifies the effectiveness of the processing scheme from the perspectives of fault resolution time and recurrence rate.

8. The knowledge graph based industrial equipment failure reasoning system of claim 7, wherein: The effectiveness quantification component divides the evaluation results into three levels: effective, partially effective, and ineffective, corresponding to different weight adjustment amplitudes.

9. The knowledge graph based industrial equipment failure reasoning system of claim 1, wherein: The historical data unit is provided with a data cleaning component that can periodically clean up duplicate records and invalid data, retaining key information related to fault reasoning.

10. The knowledge graph based industrial equipment failure reasoning system of claim 9, wherein: The data cleaning component deletes redundant data that is not associated with core entities of the knowledge graph based on entity association degree.

Images