Power generation equipment health state early warning method and system

Abstract

The invention provides a power generation equipment health state early warning method and system, and relates to the technical field of state early warning, and the method comprises the steps: carrying out the collection, preprocessing and dimension classification of the operation data of electrical equipment, carrying out the extraction of feature data, constructing an equipment fault feature library, carrying out the matching of dimension feature data, and carrying out the aggregation of abnormal feature matching data; dimension aggregation feature data is obtained, matching of dimension set feature data is carried out, a dimension anomaly coefficient is calculated, and anomaly monitoring dimension data is obtained; constructing a fault knowledge graph database, carrying out the matching of an equipment abnormal graph, obtaining the matching data of the equipment graph, calculating the comprehensive abnormal coefficient of the electrical equipment, and obtaining the data of abnormal monitoring equipment; according to the method, the flexibility of data processing and feature extraction is improved, and the accuracy of identification monitoring and early warning of map data such as faults and the like is improved.

Description

Technical Field

[0001] The present invention provides a method and system for early warning of the health status of power generation equipment, which relates to the technical field of early warning, specifically to the technical field of early warning of the health status of power generation equipment. Background Art

[0002] In the traditional field of early warning of the health status of power generation equipment, the technology faces multiple challenges. On the one hand, the means of data processing and feature extraction are limited, resulting in an insufficiently comprehensive and in-depth assessment of the equipment operation status. On the other hand, it is difficult to identify faults in a structured manner, and it is difficult to cover all fault types and provide intelligent early warnings. In addition, traditional technologies ignore the application of knowledge graphs in equipment health status monitoring, resulting in the lack of comprehensiveness and relevance of early warning information. More importantly, early warning systems often can only provide simple threshold alarms, lacking the ability of hierarchical early warning and linkage early warning, which limits the accuracy and practicality of early warning information. Summary of the Invention

[0003] The present invention provides a method and system for early warning of the health status of power generation equipment to solve the above problems:

[0004] A method and system for early warning of the health status of power generation equipment proposed by the present invention, the method comprising:

[0005] S1. Collect and preprocess the operation data of electrical equipment to obtain processed data of node equipment, classify the dimensions of the processed data of node equipment, and then extract feature data to obtain dimension feature data;

[0006] S2. Construct an equipment fault feature library, perform matching of dimension feature data to obtain abnormal feature matching data, perform aggregation of abnormal feature matching data to obtain dimension aggregated feature data, perform matching of dimension set feature data to obtain abnormal aggregated matching data, calculate the dimension abnormality coefficient, and obtain abnormal monitoring dimension data;

[0007] S3. Construct a fault knowledge graph library, perform matching of equipment abnormal graphs to obtain equipment graph matching data, calculate the comprehensive abnormality coefficient of electrical equipment, and obtain abnormal monitoring equipment data;

[0008] S4. Obtain equipment warning values and dimension warning values to perform hierarchical early warning on electrical equipment.

[0009] Further, the S1 includes:

[0010] Collect the operation data of electrical equipment at preset acquisition time points to obtain the operation data of node equipment;

[0011] Preprocess the operation data of the node equipment to obtain processed data of node equipment;

[0012] Obtain preset monitoring dimension information, classify the data processed by the node device according to the preset monitoring dimension information, and obtain multiple dimension node data;

[0013] Extract feature data from each dimension node data to obtain dimension feature data.

[0014] Further, the S2 includes:

[0015] Obtain historical device failure data, and construct a device failure feature library through the historical device failure data;

[0016] Match the dimension feature data of each electrical device with the device failure feature library to obtain abnormal feature matching data;

[0017] Aggregate multiple abnormal feature data of the dimension feature data of each electrical device to obtain dimension aggregated feature data;

[0018] Match the dimension aggregated feature data of the dimension feature data of each electrical device with the device failure feature library to obtain abnormal aggregated matching data;

[0019] Calculate the dimension anomaly coefficient of each preset monitoring dimension of the electrical device according to the abnormal aggregated matching data;

[0020] Compare the dimension anomaly coefficient with the preset dimension anomaly threshold to obtain a dimension anomaly comparison result;

[0021] Perform anomaly determination on the corresponding preset monitoring dimension according to the dimension anomaly comparison result to obtain abnormal monitoring dimension data.

[0022] Further, the S3 includes:

[0023] Obtain historical fault graph information according to the device failure feature library, perform fault correlation extraction on the historical fault graph information according to the preset feature correlation degree, and obtain fault knowledge graph data;

[0024] Construct a fault knowledge graph library according to the fault knowledge graph data;

[0025] Obtain the abnormal monitoring dimension data of all preset monitoring dimensions of the electrical device to obtain a device abnormal graph;

[0026] Match the device abnormal graph with the fault knowledge graph library to obtain device graph matching data;

[0027] Calculate the comprehensive anomaly coefficient of the electrical device according to the device graph matching data;

[0028] Compare the comprehensive anomaly coefficient with a preset comprehensive anomaly threshold to obtain a comprehensive anomaly comparison result;

[0029] Perform anomaly determination on the corresponding electrical equipment according to the comprehensive anomaly comparison result to obtain anomaly monitoring device data.

[0030] Further, the S4 includes:

[0031] Calculate the ratio of the comprehensive anomaly coefficient to the preset comprehensive anomaly threshold to obtain a device warning value;

[0032] Perform a health status warning on the electrical equipment according to the device warning value;

[0033] Calculate the ratio of the dimension anomaly coefficient to the preset dimension anomaly threshold to obtain a dimension warning value;

[0034] Perform a health status warning on the preset monitoring dimensions of the electrical equipment according to the dimension warning value;

[0035] Perform a linkage warning on the preset monitoring dimensions of the electrical equipment according to the device map matching data.

[0036] Further, the system includes:

[0037] A device feature extraction module, configured to collect and preprocess the operation data of the electrical equipment to obtain node device processed data, perform dimension classification on the node device processed data, and then extract feature data to obtain dimension feature data;

[0038] A feature dimension matching module, configured to construct a device fault feature library, perform matching of dimension feature data to obtain anomaly feature matching data, perform aggregation of the anomaly feature matching data to obtain dimension aggregated feature data, perform matching of dimension set feature data to obtain anomaly aggregated matching data, calculate the dimension anomaly coefficient, and obtain anomaly monitoring dimension data;

[0039] A map device matching module, configured to construct a fault knowledge map library, perform matching of device anomaly maps to obtain device map matching data, calculate the comprehensive anomaly coefficient of the electrical equipment, and obtain anomaly monitoring device data;

[0040] A multi-level warning module, configured to obtain device warning values and dimension warning values to perform hierarchical warning on the electrical equipment.

[0041] Further, the device feature extraction module includes:

[0042] A collection and processing module, configured to collect the operation data of the electrical equipment at a preset collection time point to obtain node device operation data;

[0043] Preprocess the operation data of the node device to obtain the processed data of the node device;

[0044] A dimension extraction module, configured to obtain preset monitoring dimension information, classify the processed data of the node device according to the preset monitoring dimension information, and obtain multiple dimension node data;

[0045] Extract feature data from each dimension node data to obtain dimension feature data.

[0046] Furthermore, the feature dimension matching module includes:

[0047] A feature matching analysis module, configured to obtain historical device failure data, and construct a device failure feature library through the historical device failure data;

[0048] Match the feature data of each dimension of the electrical device with the device failure feature library to obtain abnormal feature matching data;

[0049] A feature aggregation analysis module, configured to aggregate multiple abnormal feature data of the feature data of each dimension of the electrical device to obtain dimension aggregation feature data;

[0050] Match the dimension aggregation feature data of the feature data of each dimension of the electrical device with the device failure feature library to obtain abnormal aggregation matching data;

[0051] A dimension abnormality analysis module, configured to calculate the dimension abnormality coefficient of each preset monitoring dimension of the electrical device according to the abnormal aggregation matching data;

[0052] Compare the dimension abnormality coefficient with a preset dimension abnormality threshold to obtain a dimension abnormality comparison result;

[0053] Perform an abnormality determination on the corresponding preset monitoring dimension according to the dimension abnormality comparison result to obtain abnormal monitoring dimension data.

[0054] Furthermore, the graph device matching module includes:

[0055] A graph library construction module, configured to obtain historical fault graph information according to the device failure feature library, perform fault association extraction on the historical fault graph information according to a preset feature correlation degree, and obtain fault knowledge graph data;

[0056] Construct a fault knowledge graph library according to the fault knowledge graph data;

[0057] Obtain the abnormal monitoring dimension data of all preset monitoring dimensions of the electrical device to obtain a device abnormal graph;

[0058] A spectrum matching module, configured to match the device anomaly spectrum with the fault knowledge graph database to obtain device spectrum matching data;

[0059] A device anomaly analysis module, configured to calculate a comprehensive anomaly coefficient of an electrical device according to the device spectrum matching data;

[0060] Compare the comprehensive anomaly coefficient with a preset comprehensive anomaly threshold to obtain a comprehensive anomaly comparison result;

[0061] Perform anomaly determination on the corresponding electrical device according to the comprehensive anomaly comparison result to obtain anomaly monitoring device data.

[0062] Further, the multi-level early warning module includes:

[0063] A device early warning module, configured to calculate a ratio of the comprehensive anomaly coefficient to a preset comprehensive anomaly threshold to obtain a device early warning value;

[0064] Perform a health status early warning on the electrical device according to the device early warning value;

[0065] A dimension early warning module, configured to calculate a ratio of the dimension anomaly coefficient to a preset dimension anomaly threshold to obtain a dimension early warning value;

[0066] Perform a health status early warning on a preset monitoring dimension of the electrical device according to the dimension early warning value;

[0067] A linkage early warning module, configured to perform a linkage early warning on a preset monitoring dimension of the electrical device according to the device spectrum matching data.

[0068] Advantages of the present invention: Through the overall technical solution of the present invention, abnormal devices of power generation devices, abnormal knowledge graphs of abnormal devices, abnormal dimensions of abnormal knowledge graphs, and abnormal characteristics of abnormal dimensions can be obtained respectively. Furthermore, different early warnings can be carried out at different levels and degrees; the knowledge graph of the power generation device includes information such as the structure, working principle, common faults and their handling methods of the device. Through the knowledge graph, rapid diagnosis and early warning of device faults can be realized; through feature extraction, fault feature matching and the application of the fault knowledge graph, device faults or abnormal states can be more accurately identified, reducing false alarms and missed alarms. Real-time or regular collection of device operation data can quickly respond to changes in device status and send out early warning signals in a timely manner. The hierarchical early warning mechanism enables maintenance personnel to give priority to handling high-priority faults, reasonably allocate resources, and improve maintenance efficiency. By early warning and timely handling, economic losses such as downtime and damage caused by device faults can be effectively avoided, and the maintenance cost can be reduced. Continuous monitoring and early warning can timely discover and handle potential problems of the device, extend the service life of the device, and improve the overall reliability of the device. Description of the Drawings

[0069] Figure 1 It is a schematic diagram of a method for warning the health status of a power generation device. Specific implementation manners

[0070] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0071] In one embodiment of the present invention, a method and system for warning the health status of a power generation device are proposed. The method includes:

[0072] S1. Collect and preprocess the operation data of electrical equipment to obtain processed data of node equipment, classify the dimensions of the processed data of node equipment, and then extract feature data to obtain dimension feature data;

[0073] S2. Construct a device fault feature library, perform matching of dimension feature data to obtain abnormal feature matching data, aggregate the abnormal feature matching data to obtain dimension aggregated feature data, perform matching of dimension set feature data to obtain abnormal aggregated matching data, calculate the dimension abnormality coefficient, and obtain abnormal monitoring dimension data;

[0074] S3. Construct a fault knowledge graph library, perform matching of device abnormal graphs to obtain device graph matching data, calculate the comprehensive abnormality coefficient of electrical equipment, and obtain abnormal monitoring device data;

[0075] S4. Obtain device warning values and dimension warning values to perform hierarchical warning on electrical equipment.

[0076] The working principle of the above technical solution is as follows: Through sensors or other data acquisition means, the operation data of electrical equipment is collected in real time or regularly, which can include various parameters such as current, voltage, temperature, vibration, etc. These raw data are subjected to preprocessing steps such as cleaning, denoising, and normalization to eliminate noise and fill in missing values to ensure data quality. The preprocessed data is classified according to different preset dimensions, such as initial screening and operation process, etc., and then for each node device, the data extraction dimensions and key feature data are processed to form dimension feature data. Through feature extraction, various feature data of each dimension are obtained. A device fault feature library is constructed, which contains the feature patterns of known fault types. The dimension feature data is matched with the fault feature library to identify abnormal feature matching data. Then, these abnormal features are aggregated to form dimension aggregated feature data, which is further matched with the set features of the fault feature library to calculate the dimension abnormality coefficient and identify the abnormal monitoring dimension. Through matching and aggregation, the possible faults or abnormal states in the device are accurately identified. A fault knowledge graph library is constructed, which represents the association relationships between device faults in the form of a graph. The device anomaly graph (constructed based on the dimension aggregated feature data) is matched with the fault knowledge graph to calculate the comprehensive anomaly coefficient of the electrical equipment and identify the abnormal monitoring device. By using the association analysis ability of the graph, the health status of the device can be evaluated more comprehensively, and the accuracy and reliability of fault warning can be improved. According to the device warning value and dimension warning value, hierarchical warning is carried out for the electrical equipment. Through hierarchical warning, clear and intuitive fault information is provided for the operation and maintenance personnel, which is convenient for taking measures in time to prevent the expansion of faults or the occurrence of more serious consequences.

[0077] The technical effects of the above technical solution are as follows: Through the overall technical solution of the present invention, the abnormal devices of the power generation equipment, the abnormal knowledge graphs of the abnormal devices, the abnormal dimensions of the abnormal knowledge graphs, and the abnormal features of the abnormal dimensions can be obtained respectively. Furthermore, different warnings are carried out by grading and dividing degrees; the knowledge graph of the power generation equipment includes information such as the structure, working principle, common faults and their treatment methods of the equipment. Through the knowledge graph, rapid diagnosis and warning of equipment faults can be realized; through feature extraction, fault feature matching and the application of the fault knowledge graph, the equipment faults or abnormal states can be identified more accurately, and false alarms and missed alarms can be reduced. Collecting the equipment operation data in real time or regularly can quickly respond to the changes in the equipment state and send out warning signals in time. The hierarchical warning mechanism enables the operation and maintenance personnel to give priority to handling high-priority faults, allocate resources reasonably, and improve the operation and maintenance efficiency. By warning in advance and dealing with it in time, the economic losses caused by equipment faults such as shutdown and damage can be effectively avoided, and the operation and maintenance costs can be reduced. Continuous monitoring and warning can timely discover and handle potential problems of the equipment, extend the service life of the equipment, and improve the overall reliability of the equipment.

[0078] In an embodiment of the present invention, the S1 includes:

[0079] Collect the operation data of electrical equipment according to the preset collection time points to obtain the operation data of the node equipment;

[0080] Preprocess the operation data of the node equipment to obtain the processed data of the node equipment;

[0081] Obtain the preset monitoring dimension information, classify the processed data of the node equipment according to the preset monitoring dimension information to obtain multiple dimension node data;

[0082] Extract the characteristic data for each dimension node data to obtain the dimension characteristic data.

[0083] The working principle of the above technical solution is as follows: According to the operation characteristics and monitoring requirements of the distribution equipment, a series of collection time points are preset in advance. At these time points, through sensors, measuring instruments or other data collection devices, the operation data of electrical equipment is collected in real time or near real time. These data include various parameters such as current, voltage, power, temperature, vibration, etc., specifically depending on the monitoring target, to ensure the timing and consistency of data collection. Perform preprocessing operations such as cleaning, denoising, and formatting on the collected operation data of the node equipment. This includes steps such as removing invalid data, filling in missing values, data smoothing, data conversion (such as logarithmic conversion, standardization, etc.) to ensure data quality and analysis accuracy. Improve the availability and accuracy of data; According to the characteristics and monitoring requirements of electrical equipment, a series of monitoring dimensions are predefined in advance. Then, according to these preset monitoring dimension information, classify the preprocessed operation data of the node equipment, and allocate the relevant data to the corresponding dimensions to form multiple dimension node data. Decompose the complex data set into dimensions that are easier to manage and analyze. For each dimension node data, extract the key characteristic data that can reflect the characteristics of this dimension. Through feature extraction, convert the original data into characteristic data with clear physical meaning and diagnostic value.

[0084] The technical effects of the above technical solution are as follows: Through data preprocessing, effectively remove noise and invalid data, and improve the accuracy and availability of data. The preset collection time points and monitoring dimension information make data collection and analysis more orderly and efficient, reducing manual intervention and data processing time. Through dimension classification and feature extraction, decompose the complex data set into dimensions and features that are easier to analyze, and can more accurately identify equipment status changes and potential faults for precise positioning and control; Based on the results of preset monitoring dimensions and feature extraction, it is possible to more targeted allocate monitoring resources and take maintenance measures to improve resource utilization efficiency. Provide data support and decision-making basis for subsequent anomaly detection, fault diagnosis and intelligent operation and maintenance.

[0085] In one embodiment of the present invention, the S2 includes:

[0086] Obtain historical device failure data, and construct a device failure feature library through the historical device failure data;

[0087] Match the feature data of each dimension of the electrical device with the device failure feature library for abnormal feature data to obtain abnormal feature matching data;

[0088] Judge whether the dimension feature data matches the historical data of the device failure feature library to obtain abnormal feature data.

[0089] When the dimension feature data does not match the historical data of the device failure feature library, determine that the dimension feature data is normal feature data;

[0090] When the dimension feature data matches the historical data of the device failure feature library, determine that the dimension feature data is abnormal feature data;

[0091] Aggregate multiple abnormal feature data of the feature data of each dimension of the electrical device to obtain dimension aggregation feature data;

[0092] Match the dimension aggregation feature data of the feature data of each dimension of the electrical device with the device failure feature library to obtain abnormal aggregation matching data;

[0093] Calculate the dimension abnormality coefficient of each preset monitoring dimension of the electrical device according to the abnormal aggregation matching data;

[0094] The calculation formula of the dimension abnormality coefficient is:

[0095]

[0096] Among them, WD is the dimension abnormality coefficient, YH is the amount of abnormal aggregation matching data, ZH is the total amount of dimension data, g is the number of abnormal feature matching data in the amount of abnormal aggregation matching data, Q i is the preset feature weight data of the i-th abnormal feature matching data;

[0097] Relatively speaking, when is larger and / or is larger, the dimension abnormality coefficient is larger.

[0098] Compare the dimension abnormality coefficient with a preset dimension abnormality threshold to obtain a dimension abnormality comparison result;

[0099] Perform an abnormality determination on the corresponding preset monitoring dimension according to the dimension abnormality comparison result to obtain abnormal monitoring dimension data.

[0100] The working principle of the above technical solution is as follows: By collecting and analyzing historical equipment failure data, including various characteristic parameters (such as current, voltage, temperature, vibration, etc.) when the failure occurs, an equipment failure feature library is constructed. This library contains the feature data and feature aggregation data of known failure types. The feature data of each dimension (initial stage, operation process data, etc.) of the electrical equipment is matched with the data in the equipment failure feature library for abnormal features. This step is achieved by comparing the similarity between the real-time data and the data in the library. According to the matching result, it is judged whether the dimension feature data matches the historical data in the equipment failure feature library. The matching process is based on the combination of the graph similarity algorithm and a preset threshold. If the similarity reaches the threshold, it is considered a match; otherwise, it cannot be matched. If the matching degree is high, it indicates that the current data may indicate a certain known failure; if not, it is considered normal feature data. Multiple abnormal feature data identified in each dimension feature data are aggregated to form dimension aggregation feature data. This can overall grasp the abnormal state of the equipment in a certain dimension. The dimension aggregation feature data is matched with the aggregation feature data in the equipment failure feature library to further confirm whether there is an abnormality. This step is usually used to verify the preliminary abnormality determination result. According to the abnormal aggregation matching data, the abnormality coefficient of each preset monitoring dimension is calculated. This coefficient reflects the intensity and possibility of abnormal features in this dimension. The dimension abnormality coefficient is compared with the preset dimension abnormality threshold, and according to the comparison result, the corresponding preset monitoring dimension is determined for abnormality. If the abnormality coefficient exceeds the threshold, it is determined that this dimension is an abnormal monitoring dimension, and the corresponding abnormal monitoring dimension data is output.

[0101] The technical effects of the above technical solution are as follows: By constructing an equipment failure feature library and using feature matching technology, the failure features of electrical equipment can be identified more accurately, reducing false alarms and missed alarms. Through real-time monitoring and analysis of equipment feature data, abnormal features can be discovered in a timely manner, improving the timeliness and accuracy of failure warning. According to the abnormal monitoring dimension data, a more targeted maintenance plan can be formulated, avoiding unnecessary downtime and maintenance costs, and improving equipment utilization. The automated and intelligent abnormality determination process reduces manual intervention, improving the efficiency and accuracy of operation and maintenance work. The provided abnormal monitoring dimension data and dimension abnormality coefficient can provide strong support for equipment management and decision-making, helping enterprises better manage equipment and optimize production processes. By constructing an equipment failure feature library, performing feature data matching and abnormality determination, real-time monitoring and warning of the state of electrical equipment are achieved, improving the intelligent level and operation and maintenance efficiency of equipment management.

[0102] In an embodiment of the present invention, the S3 includes:

[0103] Obtain historical fault graph information according to the device fault feature library, perform fault association extraction on the historical fault graph information according to the preset feature correlation degree, and obtain fault knowledge graph data;

[0104] Construct a fault knowledge graph library according to the fault knowledge graph data;

[0105] Obtain abnormal monitoring dimension data of all preset monitoring dimensions of the electrical equipment to obtain an equipment abnormal graph;

[0106] Match the equipment abnormal graph with the fault knowledge graph library to obtain equipment graph matching data;

[0107] Calculate the comprehensive abnormality coefficient of the electrical equipment according to the equipment graph matching data;

[0108] The calculation formula of the comprehensive abnormality coefficient is:

[0109]

[0110] Where ZX is the comprehensive abnormality coefficient, m is the number of abnormal monitoring dimensions of the equipment abnormal graph, WD is the dimension abnormality coefficient, WY is the preset dimension abnormality threshold, and (WD - WY) a is the difference between the dimension abnormality coefficient and the preset dimension abnormality threshold of the a-th abnormal monitoring dimension, WD a is the dimension abnormality coefficient of the a-th abnormal monitoring dimension, YT is the equipment abnormal graph data volume, ZM is the total data volume of the fault knowledge graph library, and Q a is the preset weight data of the a-th abnormal monitoring dimension;

[0111] Relatively speaking, when YT and / or is larger, the comprehensive abnormality coefficient is larger.

[0112] Compare the comprehensive abnormality coefficient with the preset comprehensive abnormality threshold to obtain a comprehensive abnormality comparison result;

[0113] Perform abnormality determination on the corresponding electrical equipment according to the comprehensive abnormality comparison result to obtain abnormal monitoring equipment data.

[0114] The working principle of the above technical solution is as follows: Extract historical fault graph information from the device fault feature library. This graph information includes the feature data of each monitoring dimension of the device when the fault occurs and their mutual relationships. Perform fault association extraction on the historical fault graph information. This step aims to identify the correlations between fault features and construct fault knowledge graph data. Fault knowledge graph data usually appears as a set of associations between fault feature data in different dimensions. Integrate the extracted fault knowledge graph data to construct a fault knowledge graph library. This library contains graph information of various fault types and is used for matching with the real-time device graph. By real-time monitoring and analyzing all preset monitoring dimensions of the electrical device, obtain abnormal monitoring dimension data. These data indicate the abnormal states of the device in each dimension. Based on the abnormal monitoring dimension data, construct a device abnormal graph. This graph reflects the abnormal features of the current state of the device and their mutual relationships. Match the device abnormal graph with the fault knowledge graph library to find out if there is a fault graph that meets the similarity standard. The matching process is based on the combination of a graph similarity algorithm and a preset threshold. If the similarity reaches the threshold, it is a match; otherwise, it is not. According to the device graph matching data, calculate the comprehensive abnormality coefficient of the electrical device. This coefficient comprehensively considers the abnormal features of the device in each dimension and their correlations, reflecting the overall abnormal degree of the device. Compare the comprehensive abnormality coefficient with the preset comprehensive abnormality threshold. If the comprehensive abnormality coefficient exceeds the threshold, determine that the electrical device is an abnormal device and output the abnormal monitoring device data.

[0115] The technical effects of the above technical solution are as follows: By constructing a fault knowledge graph library and using the graph matching method, the fault types of electrical devices can be identified more accurately, reducing false alarms and missed alarms. Using the feature correlations in the fault knowledge graph, the possible future faults of the device can be predicted, and measures can be taken in advance for intervention. Based on the device abnormal graph and the comprehensive abnormality coefficient, a more scientific and reasonable maintenance plan can be formulated, improving the maintenance efficiency and accuracy. The automated and intelligent graph matching and abnormality determination processes reduce manual intervention and improve the intelligent level and efficiency of the operation and maintenance work. The provided abnormal monitoring device data and comprehensive abnormality coefficient can provide strong support for device health management, helping enterprises better manage devices, extend the service life of devices, and optimize the production process. By constructing a fault knowledge graph library, generating a device abnormal graph, performing graph matching and abnormality determination, the accurate monitoring and early warning of the state of electrical devices are realized, and the intelligent level of device management and the operation and maintenance efficiency are improved.

[0116] In an embodiment of the present invention, S4 includes:

[0117] Calculate the ratio of the comprehensive abnormality coefficient to the preset comprehensive abnormality threshold to obtain the device warning value;

[0118] Perform a health status warning for electrical equipment according to the warning value of the said equipment;

[0119] Calculate the ratio of the dimension anomaly coefficient to the preset dimension anomaly threshold to obtain the dimension warning value;

[0120] Perform a health status warning for the preset monitoring dimensions of electrical equipment according to the said dimension warning value;

[0121] Perform a linkage warning for the preset monitoring dimensions of electrical equipment according to the equipment graph matching data. (The knowledge graph data includes the linkage associations of each dimension. The linkage warning includes warnings for data such as the operating principle included in the knowledge graph).

[0122] The working principle of the above technical solution is as follows: Calculate the ratio of the comprehensive anomaly coefficient of the electrical equipment to the preset comprehensive anomaly threshold, and this ratio is named the equipment warning value. The equipment warning value reflects the degree of the overall abnormal state of the equipment relative to the preset threshold. According to the size of the equipment warning value, perform a health status warning for the electrical equipment. If the equipment warning value exceeds a preset warning threshold (which can be a fixed value or dynamically adjusted according to experience), then trigger a health status warning, indicating that the equipment may have potential risks or is about to malfunction. For each preset monitoring dimension of the electrical equipment, calculate the ratio of its dimension anomaly coefficient to the preset dimension anomaly threshold, and this ratio is named the dimension warning value. The dimension warning value reflects the degree of the abnormal state of each monitoring dimension relative to the preset threshold. According to the size of the dimension warning value, perform a health status warning for each preset monitoring dimension. If the warning value of a certain dimension exceeds the preset warning threshold, then trigger the health status warning of this dimension, indicating that this dimension may be abnormal or requires attention. Use the linkage association information of each dimension in the knowledge graph data to analyze the mutual influence and relationship between different monitoring dimensions. According to the equipment graph matching data and the linkage association information, perform a linkage warning for the preset monitoring dimensions of the electrical equipment. If the abnormal state of a certain dimension is in an abnormal graph with the health status of other dimensions or the entire equipment, then trigger a linkage warning, indicating the potential risks that may exist in the relevant dimensions or equipment.

[0123] The technical effects of the above technical solution are as follows: By calculating the device warning value and the dimension warning value and combining with the preset warning threshold, the health status of the device and each monitoring dimension can be judged more accurately, reducing false alarms and missed alarms. The linkage warning mechanism can timely detect and warn of potential device failures or abnormal states, providing earlier intervention opportunities for maintenance personnel and avoiding the expansion of failures or affecting production. According to the warning information, a more scientific and reasonable maintenance plan can be formulated, giving priority to dealing with devices and dimensions with high warning values, improving the maintenance efficiency and accuracy. The warning system can help enterprises better manage devices, timely discover and solve problems, extend the device life, and improve the device utilization rate and production efficiency. The warning system can be combined with an intelligent operation and maintenance platform to achieve automated and intelligent device monitoring, warning, and maintenance management, improving the intelligent level and efficiency of operation and maintenance work. By calculating the device warning value and the dimension warning value and combining with the linkage warning mechanism, accurate warning and intelligent management of the health status of electrical devices are achieved, improving the efficiency and accuracy of device management.

[0124] In one embodiment of the present invention, the system includes:

[0125] A device feature extraction module, configured to collect and preprocess the operation data of the electrical device to obtain node device processed data, perform dimension classification on the node device processed data, and then extract feature data to obtain dimension feature data;

[0126] A feature dimension matching module, configured to construct a device fault feature library, perform matching of dimension feature data to obtain abnormal feature matching data, perform aggregation of abnormal feature matching data to obtain dimension aggregation feature data, perform matching of dimension set feature data to obtain abnormal aggregation matching data, calculate the dimension abnormality coefficient, and obtain abnormal monitoring dimension data;

[0127] A graph device matching module, configured to construct a fault knowledge graph library, perform matching of device abnormal graphs to obtain device graph matching data, calculate the comprehensive abnormality coefficient of the electrical device, and obtain abnormal monitoring device data;

[0128] A multi-level warning module, configured to obtain the device warning value and the dimension warning value to perform hierarchical warning on the electrical device.

[0129] The working principle of the above technical solution is as follows: Through sensors or other data acquisition means, the operation data of electrical equipment is collected in real time or regularly, which can include various parameters such as current, voltage, temperature, vibration, etc. These raw data are subjected to preprocessing steps such as cleaning, denoising, and normalization to eliminate noise, fill in missing values, and ensure data quality. The preprocessed data is classified according to different preset dimensions, such as initial screening and operation process, etc., and then for each node device, key feature data is extracted from the processed data, such as current volatility, temperature change rate, etc., to form dimensional feature data. Through feature extraction, various feature data for each dimension are obtained. A device fault feature library is constructed, which contains the feature patterns of known fault types. The dimensional feature data is matched with the fault feature library to identify abnormal feature matching data. Then, these abnormal features are aggregated to form dimensional aggregated feature data, which is further matched with the set features of the fault feature library to calculate the dimensional abnormality coefficient and identify the abnormal monitoring dimension. Through matching and aggregation, faults or abnormal states that may exist in the device are accurately identified. A fault knowledge graph library is constructed, which represents the association relationships between device faults in the form of a graph. The device anomaly graph (constructed based on dimensional aggregated feature data) is matched with the fault knowledge graph to calculate the comprehensive anomaly coefficient of the electrical equipment and identify the abnormal monitoring device. Using the association analysis ability of the graph, the health status of the device can be evaluated more comprehensively, and the accuracy and reliability of fault early warning can be improved. According to the device warning value and dimensional warning value, hierarchical early warning is carried out for the electrical equipment. Through hierarchical early warning, clear and intuitive fault information is provided for the operation and maintenance personnel, facilitating timely measures to prevent the expansion of faults or the occurrence of more serious consequences.

[0130] The technical effects of the above technical solution are as follows: Through the overall technical solution of the present invention, the abnormal devices of the power generation equipment, the abnormal knowledge graphs of the abnormal devices, the abnormal dimensions of the abnormal knowledge graphs, and the abnormal features of the abnormal dimensions can be obtained respectively. Furthermore, different early warnings of different levels and degrees are carried out; the knowledge graph of the power generation equipment includes information such as the structure, working principle, common faults and their treatment methods of the equipment. Through the knowledge graph, rapid diagnosis and early warning of equipment faults can be realized; through feature extraction, fault feature matching, and the application of the fault knowledge graph, equipment faults or abnormal states can be identified more accurately, reducing false alarms and missed alarms. Collecting equipment operation data in real time or regularly can quickly respond to changes in equipment status and send early warning signals in a timely manner. The hierarchical early warning mechanism enables the operation and maintenance personnel to give priority to handling high-priority faults, reasonably allocate resources, and improve operation and maintenance efficiency. By early warning and timely handling, economic losses such as downtime and damage caused by equipment faults can be effectively avoided, and operation and maintenance costs can be reduced. Continuous monitoring and early warning can timely discover and handle potential problems of the equipment, extend the service life of the equipment, and improve the overall reliability of the equipment.

[0131] In one embodiment of the present invention, the device feature extraction module includes:

[0132] An acquisition and processing module, configured to acquire the operation data of the electrical equipment at preset acquisition time points to obtain the operation data of the node equipment;

[0133] Preprocess the operation data of the node equipment to obtain the processed data of the node equipment;

[0134] A dimension extraction module, configured to obtain preset monitoring dimension information, classify the processed data of the node equipment according to the preset monitoring dimension information to obtain multiple dimension node data;

[0135] Extract feature data from each dimension node data to obtain dimension feature data.

[0136] The working principle of the above technical solution is as follows: According to the operation characteristics and monitoring requirements of the power distribution equipment, a series of acquisition time points are preset in advance. At these time points, the operation data of the electrical equipment is collected in real time or near real time through sensors, measuring instruments or other data acquisition devices. These data include various parameters such as current, voltage, power, temperature, vibration, etc., depending on the monitoring target, to ensure the timing and consistency of data acquisition. Perform preprocessing operations such as cleaning, denoising, and formatting on the collected operation data of the node equipment. This includes steps such as removing invalid data, filling missing values, data smoothing, data conversion (such as logarithmic conversion, standardization, etc.) to ensure data quality and analysis accuracy. Improve the availability and accuracy of the data; According to the characteristics and monitoring requirements of the electrical equipment, a series of monitoring dimensions are predefined in advance. Then, according to the preset monitoring dimension information, classify the preprocessed operation data of the node equipment, and allocate the relevant data to the corresponding dimensions to form multiple dimension node data. Decompose the complex data set into dimensions that are easier to manage and analyze. For each dimension node data, extract key feature data that can reflect the characteristics of this dimension. Through feature extraction, convert the original data into feature data with clear physical meaning and diagnostic value.

[0137] The technical effects of the above technical solution are as follows: Through data preprocessing, effectively remove noise and invalid data, and improve the accuracy and availability of the data. The preset acquisition time points and monitoring dimension information make data acquisition and analysis more orderly and efficient, reducing manual intervention and data processing time. Through dimension classification and feature extraction, decompose the complex data set into dimensions and features that are easier to analyze, and can more accurately identify equipment status changes and potential faults for precise positioning and control; Based on the results of preset monitoring dimensions and feature extraction, more targeted allocation of monitoring resources and taking maintenance measures can improve resource utilization efficiency. Provide data support and decision-making basis for subsequent anomaly detection, fault diagnosis and intelligent operation and maintenance.

[0138] In one embodiment of the present invention, the feature dimension matching module includes:

[0139] A feature matching analysis module, configured to obtain historical device failure data and construct a device failure feature library through the historical device failure data;

[0140] Match the feature data of each dimension of the electrical device with the device failure feature library for abnormal feature data to obtain abnormal feature matching data;

[0141] Judge whether the dimension feature data matches the historical data of the device failure feature library to obtain abnormal feature data.

[0142] When the dimension feature data does not match the historical data of the device failure feature library, determine that the dimension feature data is normal feature data;

[0143] When the dimension feature data matches the historical data of the device failure feature library, determine that the dimension feature data is abnormal feature data;

[0144] A feature aggregation analysis module, configured to aggregate multiple abnormal feature data of the feature data of each dimension of the electrical device to obtain dimension aggregation feature data;

[0145] Match the dimension aggregation feature data of the feature data of each dimension of the electrical device with the device failure feature library for aggregation feature data to obtain abnormal aggregation matching data;

[0146] A dimension abnormality analysis module, configured to calculate a dimension abnormality coefficient for each preset monitoring dimension of the electrical device according to the abnormal aggregation matching data;

[0147] The calculation formula of the dimension abnormality coefficient is:

[0148]

[0149] Where WD is the dimension abnormality coefficient, YH is the amount of abnormal aggregation matching data, ZH is the total amount of dimension data, g is the number of abnormal feature matching data in the amount of abnormal aggregation matching data, and Q i Is the preset feature weight data of the i-th abnormal feature matching data;

[0150] Compare the dimension abnormality coefficient with a preset dimension abnormality threshold to obtain a dimension abnormality comparison result;

[0151] Perform an abnormality determination on the corresponding preset monitoring dimension according to the dimension abnormality comparison result to obtain abnormal monitoring dimension data.

[0152] The working principle of the above technical solution is as follows: By collecting and analyzing historical equipment failure data, including various characteristic parameters (such as current, voltage, temperature, vibration, etc.) at the time of failure, an equipment failure characteristic library is constructed. This library contains characteristic data and characteristic aggregation data of known failure types. The characteristic data of each dimension (initial stage, operation process data, etc.) of the electrical equipment is matched with the data in the equipment failure characteristic library for abnormal characteristics. This step is achieved by comparing the similarity between the real-time data and the data in the library and comparing the similarity with a preset matching threshold. If the matching degree is high, it indicates that the current data may indicate a certain known failure; if there is no match, it is considered normal characteristic data. Multiple abnormal characteristic data identified in the characteristic data of each dimension are aggregated to form dimension aggregation characteristic data. This can comprehensively grasp the abnormal state of the equipment in a certain dimension. The dimension aggregation characteristic data is matched with the aggregation characteristic data in the equipment failure characteristic library to further confirm whether there is an abnormality. This step is usually used to verify the preliminary abnormal determination result. According to the abnormal aggregation matching data, the abnormal coefficient of each preset monitoring dimension is calculated. This coefficient reflects the intensity and possibility of abnormal characteristics in this dimension. The dimension abnormal coefficient is compared with the preset dimension abnormal threshold, and according to the comparison result, an abnormal determination is made for the corresponding preset monitoring dimension. If the abnormal coefficient exceeds the threshold, it is determined that this dimension is an abnormal monitoring dimension, and the corresponding abnormal monitoring dimension data is output.

[0153] The technical effects of the above technical solution are as follows: By constructing an equipment failure characteristic library and using characteristic matching technology, the failure characteristics of electrical equipment can be more accurately identified, reducing false alarms and missed alarms. By real-time monitoring and analyzing equipment characteristic data, abnormal characteristics can be detected in a timely manner, improving the timeliness and accuracy of fault warning. According to the abnormal monitoring dimension data, a more targeted maintenance plan can be formulated, avoiding unnecessary downtime and maintenance costs, and improving equipment utilization. The automated and intelligent abnormal determination process reduces manual intervention, improving the efficiency and accuracy of operation and maintenance work. The provided abnormal monitoring dimension data and dimension abnormal coefficient can provide strong support for equipment management and decision-making, helping enterprises better manage equipment and optimize production processes. By constructing an equipment failure characteristic library, performing characteristic data matching and abnormal determination, real-time monitoring and warning of the status of electrical equipment are realized, improving the intelligent level of equipment management and operation and maintenance efficiency.

[0154] In an embodiment of the present invention, the graph device matching module includes:

[0155] A graph library construction module, configured to obtain historical fault graph information according to the equipment failure characteristic library, perform fault association extraction on the historical fault graph information according to a preset characteristic correlation degree, and obtain fault knowledge graph data;

[0156] Construct a fault knowledge graph library based on the fault knowledge graph data;

[0157] Obtain the abnormal monitoring dimension data of all preset monitoring dimensions of the electrical equipment to obtain the equipment abnormal graph;

[0158] A graph matching module for matching the equipment abnormal graph with the fault knowledge graph library to obtain equipment graph matching data;

[0159] An equipment abnormal analysis module for calculating the comprehensive abnormal coefficient of the electrical equipment according to the equipment graph matching data;

[0160] The calculation formula of the comprehensive abnormal coefficient is:

[0161]

[0162] Among them, ZX is the comprehensive abnormal coefficient, m is the number of abnormal monitoring dimensions of the equipment abnormal graph, WD is the dimension abnormal coefficient, WY is the preset dimension abnormal threshold, and (WD - WY) a is the difference between the dimension abnormal coefficient of the a-th abnormal monitoring dimension and the preset dimension abnormal threshold, WD a is the dimension abnormal coefficient of the a-th abnormal monitoring dimension, YT is the equipment abnormal graph data volume, ZM is the total data volume of the fault knowledge graph library, and Q a is the preset weight data of the a-th abnormal monitoring dimension;

[0163] Compare the comprehensive abnormal coefficient with the preset comprehensive abnormal threshold to obtain a comprehensive abnormal comparison result;

[0164] Perform abnormal determination on the corresponding electrical equipment according to the comprehensive abnormal comparison result to obtain abnormal monitoring equipment data.

[0165] The working principle of the above technical solution is as follows: Extract historical fault graph information from the device fault feature library. This graph information includes the feature data of each monitoring dimension of the device when the fault occurs and their mutual relationships. Perform fault association extraction on the historical fault graph information. This step aims to identify the associations between fault features and construct fault knowledge graph data. The fault knowledge graph data usually appears as a set of associations between fault feature data of different dimensions. Integrate the extracted fault knowledge graph data to construct a fault knowledge graph library. This library contains graph information of various fault types and is used for matching with the real-time device graph. By real-time monitoring and analyzing all preset monitoring dimensions of the electrical device, obtain abnormal monitoring dimension data. These data indicate the abnormal states of the device in each dimension. According to the abnormal monitoring dimension data, construct a device abnormal graph. This graph reflects the abnormal features of the current state of the device and their mutual relationships. Match the device abnormal graph with the fault knowledge graph library to find out if there is a fault graph that meets the similarity standard. The matching process is based on the combination of a graph similarity algorithm and a preset threshold. If the similarity reaches the threshold, it is a match; otherwise, it is not. According to the device graph matching data, calculate the comprehensive abnormality coefficient of the electrical device. This coefficient comprehensively considers the abnormal features of the device in each dimension and their associations, reflecting the overall abnormal degree of the device. Compare the comprehensive abnormality coefficient with the preset comprehensive abnormality threshold. If the comprehensive abnormality coefficient exceeds the threshold, determine that the electrical device is an abnormal device and output the abnormal monitoring device data.

[0166] The technical effects of the above technical solution are as follows: By constructing a fault knowledge graph library and using the graph matching method, the fault types of electrical devices can be identified more accurately, reducing false alarms and missed alarms. Using the feature associations in the fault knowledge graph, the possible future faults of the device can be predicted, and measures can be taken in advance for intervention. According to the device abnormal graph and the comprehensive abnormality coefficient, a more scientific and reasonable maintenance plan can be formulated, improving the maintenance efficiency and accuracy. The automated and intelligent graph matching and abnormality determination processes reduce manual intervention and improve the intelligent level and efficiency of the operation and maintenance work. The provided abnormal monitoring device data and comprehensive abnormality coefficient can provide strong support for device health management, helping enterprises better manage devices, extend the service life of devices, and optimize the production process. By constructing a fault knowledge graph library, generating a device abnormal graph, performing graph matching and abnormality determination, the accurate monitoring and early warning of the state of electrical devices are realized, and the intelligent level of device management and the operation and maintenance efficiency are improved.

[0167] In one embodiment of the present invention, the multi-level early warning module includes:

[0168] A device early warning module, used to calculate the ratio of the comprehensive abnormality coefficient to the preset comprehensive abnormality threshold to obtain a device early warning value;

[0169] Perform health status warning on electrical equipment according to the equipment warning value;

[0170] A dimension warning module, which is used to calculate the ratio of the dimension anomaly coefficient to the preset dimension anomaly threshold to obtain the dimension warning value;

[0171] Perform health status warning on the preset monitoring dimensions of electrical equipment according to the dimension warning value;

[0172] A linkage warning module, which is used to perform linkage warning on the preset monitoring dimensions of electrical equipment according to the equipment graph matching data. (The knowledge graph data includes the linkage associations of each dimension. The linkage warning includes warnings for data such as the operating principle included in the knowledge graph).

[0173] The working principle of the above technical solution is as follows: Calculate the ratio of the comprehensive anomaly coefficient of the electrical equipment to the preset comprehensive anomaly threshold, and this ratio is named the equipment warning value. The equipment warning value reflects the degree of the overall abnormal state of the equipment relative to the preset threshold. According to the size of the equipment warning value, perform health status warning on the electrical equipment. If the equipment warning value exceeds a preset warning threshold (which can be a fixed value or dynamically adjusted according to experience), then trigger a health status warning, indicating that the equipment may have potential risks or is about to fail. For each preset monitoring dimension of the electrical equipment, calculate the ratio of the dimension anomaly coefficient to the preset dimension anomaly threshold, and this ratio is named the dimension warning value. The dimension warning value reflects the degree of the abnormal state of each monitoring dimension relative to the preset threshold. According to the size of the dimension warning value, perform health status warning on each preset monitoring dimension. If the warning value of a certain dimension exceeds the preset warning threshold, then trigger the health status warning of this dimension, indicating that this dimension may be abnormal or needs attention. Use the linkage association information of each dimension in the knowledge graph data to analyze the mutual influence and relationship between different monitoring dimensions. According to the equipment graph matching data and the linkage association information, perform linkage warning on the preset monitoring dimensions of the electrical equipment. If the abnormal state of a certain dimension is in an abnormal graph with other dimensions or the health status of the entire equipment, then trigger a linkage warning, indicating the potential risks that may exist in the relevant dimensions or equipment.

[0174] The technical effects of the above technical solution are as follows: By calculating the device warning value and the dimension warning value and combining with the preset warning threshold, the health status of the device and each monitoring dimension can be judged more accurately, reducing false alarms and missed alarms. The linkage warning mechanism can timely detect and warn of potential device failures or abnormal states, providing earlier intervention opportunities for maintenance personnel and avoiding the expansion of failures or impacts on production. According to the warning information, a more scientific and reasonable maintenance plan can be formulated, giving priority to dealing with devices and dimensions with high warning values, improving the maintenance efficiency and accuracy. The warning system can help enterprises better manage devices, timely discover and solve problems, extend the service life of devices, and improve the utilization rate of devices and production efficiency. The warning system can be combined with an intelligent operation and maintenance platform to achieve automated and intelligent device monitoring, warning, and maintenance management, improving the intelligent level and efficiency of operation and maintenance work. By calculating the device warning value and the dimension warning value and combining with the linkage warning mechanism, accurate warning and intelligent management of the health status of electrical devices are achieved, improving the efficiency and accuracy of device management.

[0175] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these changes and modifications.

Claims

1. A method for early warning of the health status of power generation equipment, characterized in that: The method comprises: S1. Collect and preprocess the operation data of the electrical equipment to obtain the node equipment processing data, classify the node equipment processing data by dimensions, and then extract the feature data to obtain the dimension feature data; S2. Build a device fault feature library, match dimensional feature data, obtain abnormal feature matching data, aggregate abnormal feature matching data, obtain dimensional aggregate feature data, match dimensional set feature data, obtain abnormal aggregate matching data, calculate dimensional abnormality coefficient, and obtain abnormal monitoring dimensional data; S3. Build a fault knowledge graph library, match the equipment abnormal graph, obtain equipment graph matching data, calculate the comprehensive abnormal coefficient of electrical equipment, and obtain abnormal monitoring equipment data; S4. Obtain equipment warning values ​​and dimension warning values ​​to perform graded warnings on electrical equipment.

2. A power generation equipment health status early warning method according to claim 1, characterized in that: The S1 includes: Collect the operation data of the electrical equipment according to the preset collection time point to obtain the node equipment operation data; Preprocessing the node device operation data to obtain node device processing data; Acquire preset monitoring dimension information, and classify the node device processing data according to the preset monitoring dimension information to obtain node data of multiple dimensions; Feature data is extracted from each dimension node data to obtain dimension feature data.

3. A power generation equipment health status early warning method according to claim 1, characterized in that: The S2 includes: Acquire historical equipment failure data, and construct an equipment failure feature library based on the historical equipment failure data; Match each dimension feature data of the electrical equipment with the equipment fault feature library for abnormal feature data to obtain abnormal feature matching data; Aggregate multiple abnormal feature data of each dimensional feature data of the electrical equipment to obtain dimensional aggregated feature data; Match the dimension aggregated feature data of each dimension feature data of the electrical equipment with the aggregated feature data of the equipment fault feature library to obtain abnormal aggregated matching data; Calculate the dimension anomaly coefficient of each preset monitoring dimension of the electrical equipment based on the anomaly aggregate matching data; Compare the dimension anomaly coefficient with a preset dimension anomaly threshold to obtain a dimension anomaly comparison result; According to the dimension abnormality comparison result, an abnormality determination is performed on the corresponding preset monitoring dimension to obtain abnormal monitoring dimension data.

4. A power generation equipment health status early warning method according to claim 1, characterized in that: The S3 includes: Obtain historical fault graph information based on the equipment fault feature library, extract fault associations from the historical fault graph information based on the preset feature association, and obtain fault knowledge graph data; Build a fault knowledge graph library based on fault knowledge graph data; Obtain abnormal monitoring dimension data of all preset monitoring dimensions of electrical equipment and obtain equipment abnormality maps; Matching the equipment abnormality graph with the fault knowledge graph library to obtain equipment graph matching data; Calculate the comprehensive abnormality coefficient of the electrical equipment according to the equipment spectrum matching data; Comparing the comprehensive abnormality coefficient with a preset comprehensive abnormality threshold to obtain a comprehensive abnormality comparison result; According to the comprehensive abnormality comparison result, the corresponding electrical equipment is judged to be abnormal, and abnormal monitoring equipment data is obtained.

5. A power generation equipment health status early warning method according to claim 1, characterized in that: The S4 includes: Calculate the ratio of the comprehensive abnormality coefficient to the preset comprehensive abnormality threshold to obtain the equipment warning value; Performing health status warning on electrical equipment according to the equipment warning value; Calculate the ratio of the dimension anomaly coefficient to the preset dimension anomaly threshold to obtain the dimension warning value; Performing health status warning on preset monitoring dimensions of electrical equipment according to the dimension warning value; According to the equipment map matching data, the preset monitoring dimensions of the electrical equipment are linked to early warning.

6. A power generation equipment health status early warning system, characterized in that: The system comprises: The equipment feature extraction module is used to collect and pre-process the operation data of the electrical equipment, obtain the node equipment processing data, classify the node equipment processing data by dimensions, and then extract the feature data to obtain the dimension feature data; The feature dimension matching module is used to build an equipment fault feature library, match the dimension feature data, obtain abnormal feature matching data, aggregate the abnormal feature matching data, obtain dimension aggregate feature data, match the dimension set feature data, obtain abnormal aggregate matching data, calculate the dimension abnormality coefficient, and obtain abnormal monitoring dimension data; The graph equipment matching module is used to build a fault knowledge graph library, match equipment abnormality graphs, obtain equipment graph matching data, calculate the comprehensive abnormality coefficient of electrical equipment, and obtain abnormal monitoring equipment data; The multi-level warning module is used to obtain equipment warning values ​​and dimension warning values ​​to provide graded warnings for electrical equipment.

7. A power generation equipment health status early warning system according to claim 6, characterized in that: The device feature extraction module comprises: The collection and processing module is used to collect the operation data of the electrical equipment according to the preset collection time point to obtain the node equipment operation data; Preprocessing the node device operation data to obtain node device processing data; A dimension extraction module is used to obtain preset monitoring dimension information, and dimensionally classify the node device processing data according to the preset monitoring dimension information to obtain multiple dimension node data; Feature data is extracted from each dimension node data to obtain dimension feature data.

8. A power generation equipment health status early warning system according to claim 6, characterized in that: The feature dimension matching module includes: A feature matching analysis module is used to obtain historical equipment failure data and build an equipment failure feature library based on the historical equipment failure data; Match each dimension feature data of the electrical equipment with the equipment fault feature library for abnormal feature data to obtain abnormal feature matching data; A feature aggregation analysis module is used to aggregate multiple abnormal feature data of each dimension feature data of the electrical equipment to obtain dimension aggregated feature data; Match the dimension aggregated feature data of each dimension feature data of the electrical equipment with the aggregated feature data of the equipment fault feature library to obtain abnormal aggregated matching data; A dimension anomaly analysis module, used to calculate the dimension anomaly coefficient of each preset monitoring dimension of the electrical equipment based on the anomaly aggregation matching data; Compare the dimension anomaly coefficient with a preset dimension anomaly threshold to obtain a dimension anomaly comparison result; According to the dimension abnormality comparison result, an abnormality determination is performed on the corresponding preset monitoring dimension to obtain abnormal monitoring dimension data.

9. A power generation equipment health status early warning system according to claim 6, characterized in that: The graph device matching module includes: A graph library construction module is used to obtain historical fault graph information based on the equipment fault feature library, extract fault associations from the historical fault graph information based on a preset feature association, and obtain fault knowledge graph data; Build a fault knowledge graph library based on fault knowledge graph data; Obtain abnormal monitoring dimension data of all preset monitoring dimensions of electrical equipment and obtain equipment abnormality maps; A graph matching module, used to match the equipment abnormality graph with the fault knowledge graph library to obtain equipment graph matching data; An equipment anomaly analysis module, used to calculate the comprehensive anomaly coefficient of the electrical equipment according to the equipment spectrum matching data; Comparing the comprehensive abnormality coefficient with a preset comprehensive abnormality threshold to obtain a comprehensive abnormality comparison result; According to the comprehensive abnormality comparison result, the corresponding electrical equipment is judged to be abnormal, and abnormal monitoring equipment data is obtained.

10. A power generation equipment health status early warning system according to claim 6, characterized in that: The multi-level early warning module includes: The equipment early warning module is used to calculate the ratio of the comprehensive abnormality coefficient to the preset comprehensive abnormality threshold to obtain the equipment early warning value; Performing health status warning on electrical equipment according to the equipment warning value; The dimension warning module is used to calculate the ratio of the dimension anomaly coefficient to the preset dimension anomaly threshold to obtain the dimension warning value; Performing health status warning on preset monitoring dimensions of electrical equipment according to the dimension warning value; The linkage warning module is used to provide linkage warning for preset monitoring dimensions of electrical equipment according to equipment map matching data.

Images