A monitoring method, system, device and medium for a substation building

Abstract

The present invention relates to the technical field of substation monitoring, and discloses a monitoring method, system, device and medium for substation rooms. By using the real-time partial discharge monitoring data and real-time environmental monitoring data of high-voltage switchgears in the substation room, a partial discharge diagnosis waveform atlas and a partial discharge diagnosis auxiliary text are constructed, and are input into a pre-constructed graphic and text multi-modal partial discharge classification model for processing to obtain a first classification result of the partial discharge type to which the high-voltage switchgear belongs; multiple partial discharge mechanism model eigenvalues of the partial discharge diagnosis waveform atlas are determined and matched with a pre-constructed partial discharge mechanism model, and then a second classification result is determined according to the matching result; based on the first and second classification results, the target partial discharge type of the high-voltage switchgear is determined, and then the abnormal cause of the high-voltage switchgear is determined, so as to realize the monitoring of the substation room, improve the accuracy of the discrimination of the partial discharge type of the switchgear and the monitoring efficiency of the substation room, and facilitate the precise maintenance of the high-voltage switchgear.

Description

Technical Field

[0001] The present invention relates to the technical field of substation monitoring, and particularly to a monitoring method, system, device and medium for substation buildings. Background Art

[0002] As a place for receiving, distributing, controlling and protecting electric energy, the substation building is a key node in the power system, and its safe operation is directly related to the reliable power consumption of residents and enterprises. Since substation buildings are usually located in remote areas, manual inspections are mostly used to monitor them, with low efficiency and the inability to process monitoring data in real time, thus unable to achieve precise management of substation buildings.

[0003] It can be seen that how to improve the monitoring efficiency of substation buildings has become a technical problem that needs to be urgently solved by those skilled in the art. Summary of the Invention

[0004] The present invention provides a monitoring method, system, device and medium for substation buildings, and solves the problem of how to improve the monitoring efficiency of substation buildings.

[0005] To solve the above technical problem, in the first aspect of the present invention, a monitoring method for substation buildings is provided, including:

[0006] Obtain real-time partial discharge monitoring data and real-time environmental monitoring data of high-voltage switch cabinets in the substation building, and construct a partial discharge diagnosis waveform atlas and a partial discharge diagnosis auxiliary text according to the real-time partial discharge monitoring data and the real-time environmental monitoring data;

[0007] Input the partial discharge diagnosis waveform atlas and the partial discharge diagnosis auxiliary text into a pre-constructed graphic and text multi-modal partial discharge classification model for processing to obtain a first classification result of the corresponding partial discharge type of the high-voltage switch cabinet;

[0008] Determine multiple partial discharge mechanism model characteristic values of the high-voltage switch cabinet according to the partial discharge diagnosis waveform atlas to match with a pre-constructed partial discharge mechanism model, and determine a second classification result of the corresponding partial discharge type of the high-voltage switch cabinet according to the matching result;

[0009] Based on the first classification result and the second classification result, determine the target partial discharge type of the high-voltage switch cabinet, and determine the abnormal cause of the high-voltage switch cabinet according to the target partial discharge type, so as to realize the monitoring of the substation building.

[0010] As one of the preferred solutions, the constructing a partial discharge diagnosis waveform atlas and a partial discharge diagnosis auxiliary text according to the real-time partial discharge monitoring data and the real-time environmental monitoring data includes:

[0011] Construct the partial discharge diagnosis waveform atlas based on the real-time partial discharge monitoring data, and extract the mechanism characteristics from the real-time partial discharge monitoring data to obtain a number of partial discharge mechanism characteristic components; the partial discharge mechanism characteristic components include amplitude dispersion, phase aggregation, polarity effect, flight pattern characteristics, pulse equalization degree, 50Hz frequency component and 100Hz frequency component;

[0012] Quantify the correlation between the real-time partial discharge monitoring data and the real-time environmental monitoring data, and generate an external influence factor for partial discharge according to the correlation quantification result;

[0013] Perform data discretization and text escape processing on each of the partial discharge mechanism characteristic components and the external influence factor for partial discharge, and represent the processing results in the form of key-value pairs to obtain the auxiliary text for partial discharge diagnosis.

[0014] As one of the preferred solutions, the real-time environmental monitoring data includes real-time temperature data and real-time humidity data; among them,

[0015] The quantification of the correlation between the real-time partial discharge monitoring data and the real-time environmental monitoring data, and the generation of an external influence factor for partial discharge according to the correlation quantification result, includes:

[0016] Perform Pearson correlation analysis and calculation on the real-time partial discharge monitoring data and the real-time temperature data at the same time period to obtain the partial discharge temperature correlation coefficient;

[0017] Perform Pearson correlation analysis and calculation on the real-time partial discharge monitoring data and the real-time humidity data at the same time period to obtain the partial discharge humidity correlation coefficient;

[0018] Weight the partial discharge temperature correlation coefficient and the partial discharge humidity correlation coefficient according to a preset ratio to obtain the external influence factor for partial discharge.

[0019] As one of the preferred solutions, the graphic and text multi-modal partial discharge classification model includes an image encoder, a text encoder, a graphic and text multi-modal encoder and a graphic and text multi-modal decoder; among them,

[0020] The input of the partial discharge diagnosis waveform atlas and the auxiliary text for partial discharge diagnosis into a pre-constructed graphic and text multi-modal partial discharge classification model for processing to obtain the first classification result of the corresponding partial discharge type of the high-voltage switchgear, includes:

[0021] Encode the partial discharge diagnosis waveform atlas through the image encoder to obtain a partial discharge atlas feature vector, and encode the auxiliary text for partial discharge diagnosis based on the text encoder to obtain a partial discharge text feature vector;

[0022] Based on the cross-modal feature alignment loss function and the image-text matching loss function, input the partial discharge spectrum feature vector and the partial discharge text feature vector into the image-text multi-modal encoder for processing to obtain the image-text encoding result;

[0023] Input the image-text encoding result into the image-text multi-modal decoder for processing, and output the first classification result of the corresponding partial discharge type of the high-voltage switchgear.

[0024] As one of the preferred solutions, the method for determining multiple partial discharge mechanism model feature values of the high-voltage switchgear according to the partial discharge diagnosis waveform spectrum, so as to match with a pre-constructed partial discharge mechanism model, and determine the second classification result of the corresponding partial discharge type of the high-voltage switchgear includes:

[0025] Quantify multiple partial discharge mechanism model feature values of the partial discharge diagnosis waveform spectrum; the partial discharge mechanism model feature values include amplitude level, discharge times, discharge time interval, comparison of 50Hz and 100Hz frequency components, positive and negative half-axis symmetry, single-peak or double-peak characteristics, segmentation diagram, ultrasonic detection probability, and PRPS characteristics;

[0026] Match multiple partial discharge mechanism model feature values with pre-constructed partial discharge mechanism models corresponding to multiple partial discharge types to obtain several matching degrees, and use the partial discharge type corresponding to the partial discharge mechanism model with the highest matching degree as the second classification result.

[0027] As one of the preferred solutions, the first classification result includes the partial discharge type and its corresponding probability value;

[0028] The method for determining the target partial discharge type of the high-voltage switchgear based on the first classification result and the second classification result includes:

[0029] Determine the first weight value of the first classification result according to the probability value, and determine the second weight value of the second classification result based on the matching degree;

[0030] Normalize the first weight value and the second weight value to obtain the normalized weight value, and determine the target partial discharge type of the high-voltage switchgear according to the normalized weight value, the first classification result, and the second classification result.

[0031] As one of the preferred solutions, after determining the abnormal cause of the high-voltage switchgear according to the target partial discharge type, it further includes:

[0032] Construct a hidden danger feature set of the high-voltage switchgear; the hidden danger feature set includes too high humidity in the substation building, equipment aging, and family defects;

[0033] Obtain the device information of the high-voltage switchgear, and when it is determined that the high-voltage switchgear conforms to the hidden danger feature set according to the device information and the real-time environment monitoring data, determine the occurrence probability of each hidden danger feature to generate a risk identification list;

[0034] Generate a warning message for the distribution substation according to the risk identification list to realize the warning of the distribution substation.

[0035] The second aspect of the present invention provides a monitoring system for a distribution substation, including:

[0036] A partial discharge information acquisition module, configured to acquire real-time partial discharge monitoring data and real-time environment monitoring data of a high-voltage switchgear in a distribution substation, and construct a partial discharge diagnostic waveform map and a partial discharge diagnostic auxiliary text according to the real-time partial discharge monitoring data and the real-time environment monitoring data;

[0037] A first result generation module, configured to input the partial discharge diagnostic waveform map and the partial discharge diagnostic auxiliary text into a pre-constructed graphic and text multi-modal partial discharge classification model for processing, and obtain a first classification result of the corresponding partial discharge type of the high-voltage switchgear;

[0038] A second result generation module, configured to determine a plurality of partial discharge mechanism model feature values of the high-voltage switchgear according to the partial discharge diagnostic waveform map, match them with a pre-constructed partial discharge mechanism model, and determine a second classification result of the corresponding partial discharge type of the high-voltage switchgear according to the matching result;

[0039] An abnormal cause determination module, configured to determine the target partial discharge type of the high-voltage switchgear based on the first classification result and the second classification result, and determine the abnormal cause of the high-voltage switchgear according to the target partial discharge type to realize the monitoring of the distribution substation.

[0040] The third aspect of the present invention provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the monitoring method for the distribution substation as described above.

[0041] The fourth aspect of the present invention provides a computer-readable storage medium, which includes a stored computer program. When the device where the computer-readable storage medium is located executes the computer program, it implements the monitoring method for the distribution substation as described above.

[0042] Compared with the prior art, the beneficial effects of the embodiments of the present invention are at least one of the following:

[0043] (1) By combining the graphic and text multi-modal partial discharge classification model and the partial discharge mechanism model, it is possible to more accurately determine the type of partial discharge in the high-voltage switchgear and its abnormal causes; collect and process monitoring data in real time, detect and warn of potential faults in a timely manner, and improve the operation and maintenance efficiency; through automated monitoring and diagnosis, reduce the cost of manual inspections, and at the same time improve the accuracy and timeliness of operation and maintenance;

[0044] (2) By comprehensively applying technical means such as real-time monitoring, data analysis, and model matching, the precise monitoring and diagnosis of high-voltage switchgear in the substation are realized, which is of great significance for improving the safety and reliability of the power system. Brief Description of the Drawings

[0045] In order to more clearly illustrate the technical solutions of the present invention, the drawings required for the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0046] Figure 1 is a flowchart of a monitoring method for a substation provided by an embodiment of the present invention;

[0047] Figure 2 is a training process diagram of a graphic and text multi-modal partial discharge classification model provided by an embodiment of the present invention;

[0048] Figure 3 is a structural diagram of a monitoring system for a substation provided by an embodiment of the present invention;

[0049] Figure 4 is a structural diagram of an electronic device provided by an embodiment of the present invention. Detailed Embodiments

[0050] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings and embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of them. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0051] In the description of this application, the terms "first", "second", "third", etc. are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first", "second", "third", etc. may explicitly or implicitly include one or more of such features. In the description of this application, unless otherwise specified, the meaning of "a plurality" is two or more.

[0052] In the description of this application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected to" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two components. The terms "vertical", "horizontal", "left", "right", "up", "down" and similar expressions used herein are only for the purpose of illustration and do not indicate or imply that the system or component referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus cannot be construed as a limitation of the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0053] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by those skilled in the technical field of this technology. The terms used in the specification of this invention are only for the purpose of describing specific embodiments and are not intended to limit this invention. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0054] In one embodiment, as Figure 1 shown, the first aspect of the present invention provides a monitoring method for a distribution substation building, including:

[0055] S1. Obtain the real-time partial discharge monitoring data and real-time environmental monitoring data of the high-voltage switchgear in the distribution substation building, and construct a partial discharge diagnosis waveform map and a partial discharge diagnosis auxiliary text according to the real-time partial discharge monitoring data and the real-time environmental monitoring data;

[0056] Specifically, the present invention sets up a cloud, edge, and terminal side in a cloud-edge collaboration manner to obtain monitoring data of the distribution substation building. Among them, the cloud-edge collaborative computing platform is an intelligent computing architecture developed to address the new challenges brought by the development of the Internet of Things (IoT). By managing digital cloud computing, this platform can sink computing power from the centralized cloud to edge nodes closer to the data source, optimizing data processing capabilities through a distributed computing mode. The cloud-edge collaborative computing mode not only extends the capabilities of cloud-native but also enables data processing, business applications, and artificial intelligence (AI) models to be executed at the edge closer to the data source, solving problems such as real-time response, data privacy, and maintenance convenience encountered when the IoT is implemented. Its practical applications in multiple industries have significantly improved data processing efficiency and system response speed, achieved diverse intelligent applications, and contributed to the digital transformation of various industries. The cloud, edge, and terminal sides together constitute the digital and intelligent control system of the distribution substation building. This system deploys a variety of intelligent sensors in remote distribution substation buildings to achieve insulation perception of high-voltage switch cabinets, environmental status monitoring, and active warning of abnormal fire and security.

[0057] The digital and intelligent control system of the distribution substation building adopts a three-layer architecture of cloud side, edge side, and terminal side. The cloud side includes a distribution network cloud master station for managing the distribution substation building, which can be deployed on the distribution automation platform and is responsible for the overall status overview, data display, data storage, and alarm push of the distribution substation building. The distribution network cloud master station can be deployed on one or more computing devices to achieve monitoring and early warning of the distribution substation building. The edge side includes distribution digital access node devices, which can uniformly collect data from main equipment sensing devices, dynamic environment sensors, security sensors, and fire sensors through the transmission and transformation IoT protocol and can access the distribution network cloud master station through a 4G or 5G private network. The terminal side includes various intelligent sensors, which are responsible for collecting comprehensive status information of the distribution substation building and realizing abnormal monitoring of the main equipment of the distribution substation building. By deploying intelligent sensors and distribution digital access node devices in the distribution substation building, collecting and aggregating in-station monitoring data, and deploying a digital distribution network cloud master station on the cloud side, comprehensive, real-time, and continuous monitoring of multiple distribution substation buildings can be achieved.

[0058] All kinds of intelligent sensors included on the edge side can be deployed in the distribution substation building, including main equipment sensing devices, environmental sensors, security sensors, fire sensors, etc.; among them, the main equipment sensing devices include partial discharge sensors, and the partial discharge sensors include ultrasonic partial discharge sensors and dual ultrasonic transient earth voltage partial discharge sensors, etc. Usually, one or more high-voltage switchgears are deployed in the distribution substation building, and the partial discharge sensors can be deployed on the high-voltage switchgear to detect the ultrasonic parameters of partial discharge of the high-voltage switchgear, strengthen the monitoring ability of insulation abnormality in the local area of the high-voltage switchgear, and deploying ultrasonic partial discharge sensors can improve the sensing ability of the main equipment (such as high-voltage switchgear) in the distribution substation building; specifically, 6 ultrasonic partial discharge sensors can be deployed in the distribution substation building, of which 5 are deployed in the incoming and outgoing line intervals or high-voltage switchgears, and 1 is used as a background noise sensor, which is deployed on the side of the high-voltage switchgear cabinet to collect the background noise in the station.

[0059] The operation levels of the distribution substation building include regular maintenance substations and key maintenance substations. Usually, one or more ultrasonic partial discharge sensors, smoke detectors, water immersion sensors, temperature and humidity sensors, SF6 and O2 sensors, and access control sensors are deployed in regular maintenance substations; usually, one or more dual ultrasonic transient earth voltage partial discharge sensors, smoke detectors, water immersion sensors, temperature and humidity sensors, SF6 and O2 sensors, and access control sensors can be deployed in key maintenance substations; among them, one dual ultrasonic transient earth voltage partial discharge sensor can be set as a background noise sensor, which is powered by a battery and can collect ultrasonic partial discharge signals of 20kHz~100kHz and transient earth voltage of 3MHz~100MHz generated by partial discharge of electrical equipment. Through the partial discharge identification algorithm, combined with node equipment and system applications, it can identify partial discharge and interference, and can also improve the dimension of discharge signal monitoring, improve the accuracy of discharge signal monitoring, and has the dual acquisition ability of ultrasonic partial discharge and transient earth voltage, strengthen the monitoring ability of insulation damage, and then collect the background noise of the distribution substation building.

[0060] In addition, 6 dual ultrasonic transient earth voltage partial discharge sensors can also be deployed in the distribution substation building, which can be specifically deployed in the incoming and outgoing line intervals or high-voltage switchgears. The outward ultrasonic probe of one of the dual ultrasonic transient earth voltage partial discharge sensors is the source of background noise value, and this dual ultrasonic transient earth voltage partial discharge sensor is used as a background noise sensor.

[0061] Security sensors include access control sensors, etc., which are installed on one side of the station door of the distribution substation building, above the wall of the access door, or installed on the door frame with a magnetic control strip, or can also be installed on the lintel of the door with a permanent magnet to collect the status of the station door in real time, strengthen the warning ability of illegal intrusion into the distribution substation building, and improve the security ability of the distribution substation building.

[0062] The dynamic environment sensors include water immersion sensors, temperature and humidity sensors, SF6 and O2 sensors, etc.; among them, the water immersion sensors can be installed in the cable trench of the substation building, near the gap of the top cable cover plate, or installed 15 cm above the bottom of the cable trench for the detection electrode, which is used to monitor the abnormal water level in the station to achieve the flood warning ability.

[0063] The temperature and humidity sensors can be installed beside the high-voltage switchgear in the substation building or in the measured environment, and can be installed at a suitable position on the wall or magnetically attached to the side of the high-voltage switchgear to monitor the temperature and humidity in real time, improve the perception ability of the impact of abnormal temperature and humidity on the main equipment. The monitoring data adopts the transmission and transformation Internet of Things protocol and is sent to the corresponding power distribution digital access node device through the wireless LoRa communication method. The power distribution digital access node device then transmits the monitoring data to the upper-layer monitoring system through communication methods such as Ethernet or RS485 / RS232 to achieve functions such as remote monitoring and warning of temperature and humidity.

[0064] The SF6 and O2 sensors can be installed on the surface of the high-voltage switchgear, or installed on the station wall near the power socket and the maintenance power supply box, or installed directly below the power distribution digital access node device, 30 cm above the ground, to improve the airtightness monitoring ability of the switch and reduce the risk of suffocation of the operation and maintenance personnel. Since the circuit breaker components in the high-voltage switchgear use SF6 insulation, monitoring its content to judge whether there is leakage can prevent personnel from inhaling harmful gases. Setting the dynamic environment sensors can enhance the environmental perception ability of the substation building.

[0065] The fire sensors include smoke detectors, etc. The smoke detectors can be installed at a high place in the substation building and can be deployed at the center position of the station roof, which is used to monitor the smoke concentration in the station, improve the fire warning and emergency response ability, and thus enhance the fire protection ability of the substation building.

[0066] In the distribution substation building, multiple sensors are set up in the above-mentioned manner to collect the real-time environmental monitoring data and real-time partial discharge monitoring data of the high-voltage switchgear through their collaborative cooperation; among them, the real-time environmental monitoring data includes temperature, humidity, etc.; the real-time partial discharge monitoring data includes in-cabinet partial discharge data, out-of-cabinet partial discharge data, and transient earth voltage data, which are further divided into structured data and unstructured data. Partial discharge (abbreviated as PD) is one of the main hidden dangers in the operation and maintenance of power equipment. PD will cause the insulation of the equipment to deteriorate, and in severe cases, it will trigger equipment operation failures. Timely detection and handling of PD hidden dangers have a significant impact on the safe and stable operation of power equipment. The in-cabinet partial discharge data and out-of-cabinet partial discharge data are structured data, including the highest peak value of partial discharge, the average value of partial discharge, the number of partial discharge pulses, etc. The structured data is used for threshold judgment, and it can be determined whether the switchgear partial discharge data reaches the attention interval, the severe interval, and the maintenance required interval. In the present invention, it can be set that when the switchgear partial discharge data is above 8 dBμV and below 20 dBμV, it reaches the attention interval; when the switchgear partial discharge data is above 20 dBμV and below 30 dBμV, it reaches the severe interval; when the switchgear partial discharge data is above 30 dBμV, it reaches the maintenance required interval. When it is monitored that the switchgear data reaches the above intervals, alarm information of general alarm, severe alarm, and crisis alarm can be generated respectively to prompt the operation and maintenance personnel.

[0067] The alarm level can be determined through the structured data. Different alarm levels can indicate the severity of the partial discharge abnormality of the high-voltage switchgear to the operation and maintenance personnel. However, based on the structured data, the reason for the partial discharge abnormality of the switchgear cannot be further determined, nor can the partial discharge data be analyzed when the switchgear partial discharge data does not reach the abnormal level, and then the possible types of partial discharge abnormalities can be judged to eliminate the relevant reasons. Therefore, the present invention collects the unstructured data of the high-voltage switchgear through a partial discharge sensor, which includes partial discharge pulse signals, and a partial discharge pattern of the switchgear partial discharge data can be drawn according to the partial discharge pulse signals. After obtaining these real-time data, preprocessing operations such as data cleaning are performed on them, such as deleting duplicate data, filling missing values, etc., to improve the accuracy of the data.

[0068] In one embodiment, constructing a partial discharge diagnosis waveform map and a partial discharge diagnosis auxiliary text according to the real-time partial discharge monitoring data and the real-time environmental monitoring data includes:

[0069] Constructing the partial discharge diagnosis waveform map according to the real-time partial discharge monitoring data, and extracting mechanism characteristics from the real-time partial discharge monitoring data to obtain several partial discharge mechanism characteristic components; the partial discharge mechanism characteristic components include amplitude dispersion, phase aggregation, polarity effect, flight pattern characteristics, pulse equalization degree, 50 Hz frequency component, and 100 Hz frequency component;

[0070] Quantify the correlation between the real-time partial discharge monitoring data and the real-time environmental monitoring data, and generate an external influence factor of partial discharge according to the quantification result of the correlation;

[0071] Perform data discretization and text escape processing on each of the partial discharge mechanism characteristic components and the external influence factor of partial discharge, and represent the processing result in the form of key-value pairs to obtain the auxiliary text for partial discharge diagnosis.

[0072] Specifically, the present invention constructs a partial discharge diagnosis waveform atlas based on the real-time partial discharge monitoring data, that is, a partial discharge phase distribution atlas (PRPD); subsequently, mechanism characteristics are extracted from the real-time partial discharge monitoring data to generate partial discharge mechanism characteristic components, that is, for the partial discharge sampling data, based on the mechanism of partial discharge signal generation, calculate 7 types of partial discharge mechanism characteristic components such as the amplitude dispersion of partial discharge signal (Tamp), phase aggregation (Tphase), polarity effect (Tpole), flight pattern feature (Tfly), pulse equalization degree (Tpulse), 50Hz frequency component (Tf50), and 100Hz (Tf100) frequency component.

[0073] In one embodiment, the real-time environmental monitoring data includes real-time temperature data and real-time humidity data; wherein,

[0074] The quantification of the correlation between the real-time partial discharge monitoring data and the real-time environmental monitoring data, and the generation of an external influence factor of partial discharge according to the quantification result of the correlation includes:

[0075] Perform Pearson correlation analysis and calculation on the real-time partial discharge monitoring data and the real-time temperature data in the same time period to obtain the partial discharge temperature correlation coefficient;

[0076] Perform Pearson correlation analysis and calculation on the real-time partial discharge monitoring data and the real-time humidity data in the same time period to obtain the partial discharge humidity correlation coefficient;

[0077] Weight the partial discharge temperature correlation coefficient and the partial discharge humidity correlation coefficient according to a preset ratio to obtain the external influence factor of partial discharge.

[0078] The present invention performs Pearson correlation calculations on the real-time partial discharge monitoring data and the real-time temperature data, and the real-time humidity data in the same time period respectively to obtain the partial discharge temperature and humidity correlation coefficients, and divides them into three levels according to [0, 0.3), [0.3, 0.7), [0.7, 1], and thus the external influence factor of partial discharge can be obtained. By quantifying the correlation between the real-time monitoring data and the environmental monitoring data and generating the external influence factor of partial discharge, the present invention helps to identify the influence of environmental factors on partial discharge behavior, thereby enhancing environmental adaptability.

[0079] After each partial discharge mechanism characteristic component and partial discharge external influencing factor are correspondingly translated into the text "L, M, H", data discretization and text translation are performed on them, and a partial discharge diagnosis auxiliary text S is generated in a key-value pair mode of name and value "key_value", such as: "Tamp_H, Tphase_L, Tpole_M, Tfly_L, Tpulse_H, Tf50_M, Tf100_L,Xtemp_H, Xhumid_L". By extracting multiple partial discharge mechanism characteristic components, the present invention can more comprehensively understand the characteristics of partial discharge, thereby improving the accuracy and reliability of fault diagnosis; the data discretization and text translation processing make the diagnosis result more intuitive and easy to understand, and easy to understand and explain; the generated partial discharge diagnosis auxiliary text can be used as the input of subsequent automated and intelligent diagnostic systems to support higher-level fault warning and diagnostic decisions.

[0080] S2. Inputting the partial discharge diagnosis waveform spectrum and the partial discharge diagnosis auxiliary text into a pre-constructed graphic-text multimodal partial discharge classification model for processing, and obtaining a first classification result of the partial discharge type corresponding to the high-voltage switchgear; wherein the graphic-text multimodal partial discharge classification model includes an image encoder, a text encoder, a graphic-text multimodal encoder, and a graphic-text multimodal decoder;

[0081] In one embodiment, step S2 includes:

[0082] Encoding the partial discharge diagnosis waveform spectrum through the image encoder to obtain a partial discharge spectrum feature vector, and encoding the partial discharge diagnosis auxiliary text based on the text encoder to obtain a partial discharge text feature vector;

[0083] Based on the cross-modal feature alignment loss function and the image-text matching loss function, the partial discharge spectrum feature vector and the partial discharge text feature vector are input into the image-text multimodal encoder for processing to obtain an image-text encoding result;

[0084] The image-text encoding result is input into the image-text multimodal decoder for processing, and a first classification result of the partial discharge type corresponding to the high-voltage switch cabinet is output.

[0085] Specifically, due to the complex partial discharge phenomenon in the switchgear of the substation building, there are limitations in using only traditional methods to classify and identify partial discharge spectrograms, such as PRPD spectrogram pictures, by means of CNN convolutional neural network, which seriously affects the accuracy of partial discharge diagnosis, that is: the on-site signals in the substation building are complex and are easily affected by internal and external interferences, and the PRPD waveform spectrograms vary greatly, which affects the classification accuracy based on image recognition; the PRPD spectrogram pictures are generated by phase superposition of the pulse sampling signal sequences based on the first sampling, reflecting the phase and amplitude distribution relationship of the partial discharge pulses within one power frequency cycle; information loss will occur during the generation of the PRPD spectrogram pictures, for example, the time interval between adjacent pulses cannot be expressed, and the amplitude of the frequency domain components cannot be expressed, which limits the accurate diagnosis of the partial discharge type. Based on this, the present invention proposes a text-image multi-modal partial discharge classification model, and its training process is as Figure 2 shown. It not only uses partial discharge spectrograms, such as PRPD spectrogram images, for partial discharge type classification, but also extracts the partial discharge mechanism feature components from the partial discharge monitoring data to assist in classification, and incorporates the external influence parameters related to the generation of partial discharges into the diagnostic model, which is conducive to the accurate diagnosis of partial discharges. In addition to partial discharges caused by the deterioration of the insulation of the equipment itself, when the temperature or humidity in the substation building is too high, the probability of partial discharges in the switchgear will also increase. Therefore, considering the external influence parameters can improve the accuracy of partial discharge diagnosis.

[0086] Moreover, due to the different temperatures and humidities in the substation building, the types of partial discharges that occur in the switchgear or the probabilities of different partial discharge types occurring will also be different. For example, at different temperatures and humidities, different components of the switchgear will generate different types of partial discharges. Therefore, the temperature and humidity are trained together with the partial discharge data to improve the accuracy of the text-image multi-modal partial discharge classification model in determining the partial discharge type.

[0087] When training the text-image multi-modal partial discharge classification model, historical partial discharge sampling data of the switchgear is obtained based on the partial discharge sensors deployed in the substation building. The historical partial discharge sampling data includes the sampling data of multiple partial discharge signals. For example, within a total sampling time of 1 s, there is a signal sampling every 0.27 ms, and a total of 3600 ultrasonic partial discharge sampling pulse amplitudes; the present invention also obtains the environmental data of the substation building at each signal sampling, and the environmental data of the substation building includes temperature data and humidity data; then, corresponding historical partial discharge spectrograms and historical auxiliary texts are generated according to each sampling data in the above manner, that is, the picture-text data pairs generated based on each sampling of the historical partial discharges, and manual annotation work is carried out. The annotation types include 6 categories: no partial discharge, surface discharge, corona discharge, particle discharge, floating discharge, and air gap discharge, and based on the Transformer model, the trained data is used to train the text-image multi-modal partial discharge classification model. The specific training process is similar to its classification process for real-time data, only the data used is different.

[0088] The classification process of the image-text multimodal partial discharge classification model for real-time data includes: encoding the partial discharge diagnostic waveform atlas through an image encoder (such as CNN, ViT, etc.), and based on the self-attention mechanism and the feed-forward propagation function to form a partial discharge atlas feature vector, which can capture key information in the waveform atlas, such as discharge intensity, discharge mode, etc.; that is, applying the Transformer architecture to the field of image processing by using ViT, by dividing the image into a series of small patches, converting these small patches into embedding vectors, and then encoding through the Encoder part of the Transformer to extract the partial discharge atlas feature vector;

[0089] Encoding the partial discharge diagnostic auxiliary text through a text encoder (such as the Transformer architecture, RNN, LSTM, GRU, etc.), and based on the self-attention mechanism and the feed-forward propagation function to form a partial discharge text feature vector, which can capture semantic information in the text, such as descriptions of discharge type, discharge location, etc.; that is, similar to the VisionTransformer in the image encoder, the text encoder also uses the Transformer architecture to convert the input text into a hidden representation through the self-attention mechanism and the feed-forward neural network, leveraging the advantages of the Transformer in processing long sequences and capturing global dependencies to convert the text into a high-dimensional hidden representation;

[0090] Importing the two types of vectors into the image-text multimodal encoder, which can learn the correlation information between images and texts, and optimize based on the cross-modal feature alignment loss function and the image-text matching loss function, and then obtain the image-text encoding result, which integrates the information of images and texts and provides rich feature representations for subsequent decoding and classification; that is, by introducing the attention mechanism, the image-text multimodal encoder can dynamically focus on the key information in images and texts and capture the correlation between them, which is achieved by calculating the similarity score between image and text features, using a shared Transformer layer and further fusing and encoding through the self-attention mechanism and the feed-forward neural network to enable the model to learn the deep correlation between images and texts;

[0091] Finally, the graphic and text encoding results are output through the graphic and text multi-modal decoder to obtain the classification results. That is, a decoder based on Transformer is adopted, and the Decoder part of Transformer is used to generate the corresponding output according to the encoding results, that is, the partial discharge type classification result of the high-voltage switchgear. An example of the classification result is as follows: the probability of being surface discharge is 80%, the probability of being floating discharge is 3%, the probability of being air gap discharge is 5%, the probability of being corona discharge is 1%, and the probability of being particle discharge is 11%. Finally, the partial discharge type with the highest probability value is used as the first classification result. For example, when the classifier layer outputs that the probability of being surface discharge is 80%, then surface discharge is used as the first classification result.

[0092] The overall architecture of the graphic and text multi-modal partial discharge classification model includes: Input layer: Receive image and text data respectively and perform preprocessing (such as image scaling, text tokenization, etc.); Image encoder and text encoder: Process image and text data respectively to extract their respective feature representations; Graphic and text multi-modal encoder: Fuse the feature representations of the image and text and extract cross-modal correlation information; Classification layer (i.e., graphic and text multi-modal decoder): Based on the fused feature representations, use a fully connected layer or other classifier to perform partial discharge classification.

[0093] The present invention captures the partial discharge characteristics of the high-voltage switchgear more comprehensively by combining image and text information, thereby improving the classification accuracy. Since image and text information have different characteristics and advantages, this solution can resist the noise and interference of a single information source to a certain extent and enhance the robustness of the model. This solution promotes cross-modal learning between images and texts through cross-modal feature alignment and graphic and text matching loss functions, providing a basis for subsequent graphic and text fusion and joint analysis. In addition, the graphic and text multi-modal partial discharge classification model uses partial discharge spectrograms to extract the characteristic components of partial discharge mechanisms and combines them with the data of external influencing factors of partial discharge to give full play to the joint diagnostic advantages of multi-faceted characterization data of partial discharge, avoid the problem of insufficient information represented by a single data, and achieve accurate judgment of the partial discharge type of the switchgear.

[0094] S3. Determine multiple characteristic values of the partial discharge mechanism model of the high-voltage switchgear according to the partial discharge diagnosis waveform spectrogram, match them with a pre-constructed partial discharge mechanism model, and determine the second classification result of the corresponding partial discharge type of the high-voltage switchgear according to the matching result.

[0095] In one embodiment, step S3 includes:

[0096] Quantify multiple partial discharge mechanism model eigenvalues of the partial discharge diagnosis waveform atlas; the partial discharge mechanism model eigenvalues include amplitude level, discharge times, discharge time interval, comparison of 50Hz and 100Hz frequency components, positive and negative half-axis symmetry, single-peak or double-peak characteristics, classification diagram, ultrasonic detection probability, and PRPS characteristics;

[0097] Match multiple partial discharge mechanism model eigenvalues with partial discharge mechanism models corresponding to various pre-constructed partial discharge types to obtain several matching degrees, and take the partial discharge type corresponding to the partial discharge mechanism model with the highest matching degree as the second classification result.

[0098] Specifically, the present invention pre-constructs partial discharge mechanism models corresponding to each partial discharge type, including surface discharge mechanism models, floating discharge mechanism models, corona discharge mechanism models, air gap discharge mechanism models, and particle discharge mechanism models. Each partial discharge mechanism model is determined according to the partial discharge atlas corresponding to the partial discharge type, including extracting the eigenvalues of each partial discharge atlas as the partial discharge mechanism model eigenvalues of the corresponding partial discharge type's partial discharge mechanism model features. For example, if the eigenvalue corresponding to the amplitude dispersion of the surface discharge partial discharge atlas is relatively large, then the relatively large value is taken as the eigenvalue of the amplitude dispersion feature in the surface discharge mechanism model; among them, the partial discharge mechanism model eigenvalues include one or more of amplitude level, amplitude dispersion, discharge times, discharge time interval, 50Hz frequency component, 100Hz frequency component, comparison of 50Hz and 100Hz frequency components, positive and negative half-axis symmetry, single-peak or double-peak characteristics, polarity effect, classification diagram, ultrasonic detection probability, and PRPS characteristics. The partial discharge mechanism model eigenvalues of each partial discharge mechanism model include the eigenvalues of multiple partial discharge mechanism model features; among them, examples of the eigenvalues of each mechanism model feature are shown in Table 1:

[0099] Table 1 Example table of eigenvalues of each mechanism model

[0100]

[0101] The "low", "relatively large", and "high" in terms of amplitude level can be classified as follows: The highest amplitude value among all training samples for training the text-image multi-modal partial discharge classification model is used as the amplitude standard value. The amplitude standard value is divided into four equal parts to obtain four successively increasing value ranges, namely the first to the fourth value ranges. The amplitude in the first value range is "low", the amplitude in the third value range is "relatively large", and the amplitude in the fourth value range is "high". In terms of amplitude dispersion, "relatively large" means that the amplitude covers more than 60% of all acquisition phases, "small" means that the amplitude covers less than 30% of all acquisition phases, and "very large" means that the amplitude covers more than 80% of all acquisition phases. The "few" and "relatively many" in terms of the number of discharges can also be classified as follows: The quartiles of all the number of discharges in all training samples for training the text-image multi-modal partial discharge classification model are determined to obtain the positions of the first to the fourth quartiles. The number of discharges less than the lower quartile is regarded as "few", and the number of discharges greater than the second quartile is regarded as "relatively many". Similarly, the "obvious", "not obvious", and "none or very small" in the 50 Hz frequency component can also be classified as follows: The highest amplitude value among all training samples for training the text-image multi-modal partial discharge classification model is used as the amplitude standard value. The amplitude standard value is divided into four equal parts to obtain four successively increasing value ranges, namely the first to the fourth value ranges. The amplitude in the first value range is "none or very small"; the amplitude in the second value range is "not obvious", and the amplitudes in the third and fourth value ranges are "obvious". Thus, the present invention can perform feature extraction based on the drawn partial discharge pattern, determine the eigenvalue of one or more partial discharge mechanism model features for partial discharge mechanism model matching, and the specific method for performing feature extraction based on the partial discharge pattern is not limited, such as it can be extracted by means of image recognition algorithms, etc.

[0102] When determining the matching degree between multiple partial discharge mechanism model eigenvalues and the partial discharge mechanism model corresponding to each partial discharge type, the partial discharge mechanism model eigenvalues can be compared with the partial discharge mechanism model eigenvalues corresponding to the partial discharge mechanism model features in the partial discharge mechanism model. If the partial discharge mechanism model eigenvalues are the same, the partial discharge mechanism model feature is matched; if not, the partial discharge mechanism model feature is not matched. Then, it is judged whether the next partial discharge mechanism model eigenvalue is the same until the comparison of all partial discharge mechanism model eigenvalues is completed, and the ratio of the number of partial discharge mechanism model features with the same partial discharge mechanism model eigenvalues as the partial discharge mechanism model to the total number of all partial discharge mechanism model features of the partial discharge mechanism model is calculated as the matching degree. It is also possible to set corresponding weights for each partial discharge mechanism model feature, and calculate the matching degree according to the partial discharge mechanism model features that can be matched and the corresponding weights set. The weights set for each partial discharge mechanism model feature can be obtained according to the importance of the partial discharge mechanism model feature or the empirical values obtained through multiple tests. The present invention does not limit the weight values and setting methods corresponding to each partial discharge mechanism model feature. Finally, the partial discharge type corresponding to the partial discharge mechanism model with the highest matching degree is used as the second classification result.

[0103] By quantifying and matching multiple partial discharge mechanism model eigenvalues, the present invention can more comprehensively reflect the essence of the discharge phenomenon, thereby improving the accuracy of diagnosis; realizing the automation of partial discharge diagnosis, reducing manual intervention, and improving work efficiency; considering a variety of partial discharge mechanism model eigenvalues to adapt to different types of partial discharge phenomena, with strong versatility and adaptability; through the quantification and matching of partial discharge mechanism model eigenvalues, it can provide strong support for the fault diagnosis of power equipment, helping to detect and handle potential fault hazards in a timely manner.

[0104] S4. Determine the target partial discharge type of the high-voltage switchgear based on the first classification result and the second classification result, and determine the abnormal cause of the high-voltage switchgear according to the target partial discharge type, so as to realize the monitoring of the substation building;

[0105] In an embodiment, the first classification result includes the partial discharge type and its corresponding probability value;

[0106] The determining of the target partial discharge type of the high-voltage switchgear based on the first classification result and the second classification result includes:

[0107] Determine the first weight value of the first classification result according to the probability value, and determine the second weight value of the second classification result based on the matching degree;

[0108] Normalize the first weight value and the second weight value to obtain a normalized weight value, and determine the target partial discharge type of the high-voltage switchgear according to the normalized weight value, the first classification result, and the second classification result.

[0109] Specifically, the present invention normalizes the probability value in the first classification result and the matching degree between the second classification result and the corresponding partial discharge mechanism model, and uses them as the first weight value and the second weight value respectively; wherein, the maximum values of the first weight value and the second weight value are both 50%. For example, when the probability value is 80%, the first weight value obtained after normalization is 40%; when the matching degree is 70%, the second weight value obtained after normalization is 35%. Then, according to the first weight value, the second weight value, the first classification result and the second classification result, it is determined which classification result has a higher score, that is, it is used as the target partial discharge type corresponding to the partial discharge data of the switchgear. When the first classification result and the second classification result are the same, the partial discharge type pointed to by the same classification result can be directly used as the target partial discharge type corresponding to the partial discharge data of the switchgear.

[0110] Each partial discharge type may correspond to one or more reasons for partial discharge abnormalities. The reasons for partial discharge abnormalities corresponding to each partial discharge type can be collected according to the historical operation and maintenance experience and troubleshooting experience of high-voltage switchgears, so as to predict the reasons for partial discharge abnormalities. The reasons for partial discharge abnormalities corresponding to surface discharge may include: deterioration of the surface of the insulating medium, such as rupture of the cable insulation layer, etc.; the reasons for partial discharge abnormalities corresponding to floating discharge may include cracking of the fuse, etc.; the reasons for partial discharge abnormalities corresponding to corona discharge may include abnormal insulation of components inside the high-voltage switchgear, resulting in tip discharge; the reasons for partial discharge abnormalities corresponding to air gap discharge include hollowing inside the insulating medium and defective quality, etc.; the reasons for partial discharge abnormalities corresponding to particle discharge include excessive dust accumulation inside the high-voltage switchgear, resulting in the dust floating and charging inside the high-voltage switchgear, etc. The present invention does not limit the specific reasons for partial discharge abnormalities corresponding to each partial discharge type, and can be added as needed. After determining the partial discharge type, the pre-stored reasons for partial discharge abnormalities corresponding to the partial discharge type can be obtained and used as the predicted reasons for partial discharge abnormalities; in addition, according to the classified target partial discharge type, combined with the structural characteristics, operating environment and historical fault data of the high-voltage switchgear, the reasons for the abnormalities can be analyzed and determined. The specific process is not elaborated here. By combining the weights of the two classification methods, the present invention can make full use of their respective advantages, reduce the errors that may be brought by a single classification method, and thus improve the accuracy of classification.

[0111] In one embodiment, after determining the reason for the abnormality of the high-voltage switchgear according to the target partial discharge type, it further includes:

[0112] Construct a hidden danger feature set of the high-voltage switchgear; the hidden danger feature set includes high humidity in the substation building, equipment aging and family defects;

[0113] Obtain the device information of the high-voltage switchgear, and when it is determined that the high-voltage switchgear meets the hidden danger feature set according to the device information and the real-time environmental monitoring data, determine the occurrence probability of each hidden danger feature to generate a risk identification list;

[0114] Generate a warning message for the substation building according to the risk identification list to realize the warning of the substation building.

[0115] Since the influence of environmental humidity on the partial discharge of the switchgear is positively correlated, and when the humidity in the substation building reaches more than 50%, the partial discharge signal is significantly enhanced. Therefore, the present invention can set the preset humidity threshold to 50%, that is, when the humidity in the substation building reaches the set preset humidity threshold of 50%, the too high humidity in the substation building is used as an influencing factor for the abnormal partial discharge of the high-voltage switchgear, indicating that the analysis of the abnormal partial discharge of the high-voltage switchgear is caused by the too high humidity in the substation building.

[0116] The present invention pre-collects the device information of the high-voltage switchgear in each substation building, including the operation time and operation duration of the high-voltage switchgear, and obtains the time of historical partial discharge abnormality of each high-voltage switchgear, and stores the time of historical partial discharge abnormality of each high-voltage switchgear and the operation duration in an associated manner to determine the operation duration at the time of historical partial discharge abnormality of each high-voltage switchgear; then statistically analyzes the information of the partial discharge abnormality and operation duration of the high-voltage switchgear, establishes the relationship between the partial discharge abnormality and operation duration of the high-voltage switchgear, and the fitting function between the probability of partial discharge abnormality of the high-voltage switchgear and the operation duration, and determines the operation duration when the partial discharge abnormality of the switchgear is greater than the preset probability threshold according to the fitting function. For example, when the preset probability threshold is set to 50%, the operation duration when the probability of partial discharge abnormality of the switchgear is greater than 50% is 13 - 14 years. And as the operation duration of the high-voltage switchgear increases, the probability of occurrence of faults such as partial discharge abnormality increases accordingly. The operation duration threshold can be determined according to the operation duration when the probability of partial discharge abnormality of the switchgear is greater than the preset probability threshold. For example, 13 years is used as the operation duration threshold. Then, by obtaining the operation duration of the high-voltage switchgear, if the operation duration is greater than the preset operation duration threshold, the aging of the high-voltage switchgear device can be used as an influencing factor for the abnormal partial discharge of the high-voltage switchgear.

[0117] In addition, the device information also includes the manufacturer information, brand information and model information of the high-voltage switchgear, and the family defects of the high-voltage switchgear are also statistically analyzed. The family defects include the high-voltage switchgear corresponding to the manufacturer information, brand information and model information. According to the device live detection data, the number of times of partial discharge abnormality and other faults of a certain component is relatively large, and all high-voltage switchgear in the substation buildings belonging to the same manufacturer information, brand information and model information are marked according to the family defects. When the high-voltage switchgear has a partial discharge abnormality, it is determined whether the high-voltage switchgear is marked. If the high-voltage switchgear is marked, the family defect is used as an influencing factor for the abnormal partial discharge of the high-voltage switchgear.

[0118] Based on this, the present invention constructs a hidden danger feature set of high-voltage switch cabinets from the high humidity in the substation building, equipment aging, and family defects; and uses the high humidity in the substation building as a hidden danger feature for early warning to handle the high humidity in the substation building and avoid abnormal partial discharge of the switch cabinet and more serious faults; uses the aging of high-voltage switch cabinet equipment as a hidden danger feature for early warning to increase the maintenance frequency of high-voltage switch cabinets with equipment aging and avoid abnormal partial discharge of the switch cabinet and greater faults; uses family defects as a hidden danger feature for early warning to increase the maintenance frequency of marked high-voltage switch cabinets or replace components prone to problems, and avoid abnormal partial discharge of the switch cabinet and greater faults.

[0119] The present invention obtains the equipment information of the high-voltage switch cabinet, and when it is determined according to the equipment information and real-time environmental monitoring data that the high-voltage switch cabinet conforms to the hidden danger features in the hidden danger feature set, that is, when it is determined according to the equipment information that the high-voltage switch cabinet is marked, it is judged that the high-voltage switch cabinet in the distribution substation building has the hidden danger feature of family defects, and the occurrence probability of family defects is determined according to the probability values of faults such as abnormal partial discharge occurring in the high-voltage switch cabinet corresponding to the manufacturer information, brand information, and signal information, and added to the risk identification list; when it is determined according to the equipment information that the high-voltage switch cabinet reaches the operation duration threshold, it is judged that the high-voltage switch cabinet in the distribution substation building has the hidden danger feature of high-voltage switch cabinet equipment aging, and according to the fitting function of abnormal partial discharge of the high-voltage switch cabinet and the operation duration, the probability of abnormal partial discharge occurring in the high-voltage switch cabinet at this operation duration is determined and used as the occurrence probability of high-voltage switch cabinet equipment aging; when it is judged according to the substation building humidity uploaded by the temperature and humidity sensor that the substation building humidity is greater than the preset humidity threshold, it is judged that the high-voltage switch cabinet in the distribution substation building has the hidden danger feature of high substation building humidity, and the higher the substation building temperature, the higher the occurrence probability of this hidden danger feature, and it is added to the risk identification list. The occurrence probability of high substation building humidity can be determined according to the probability and frequency of faults such as abnormal partial discharge caused by historical substation building humidity. The specific determination method of the occurrence probability of high substation building humidity in the present invention is not limited.

[0120] Finally, early warning information for the distribution substation building is generated according to the risk identification list. The early warning information includes multiple hidden danger features listed in the risk identification list and the occurrence probability of each hidden danger feature, and the early warning information can sort the hidden danger features according to the occurrence probability of the hidden danger features, so that the operation and maintenance personnel can handle one or more hidden danger features with the highest risk and realize the early warning of the distribution substation building.

[0121] Through real-time monitoring and data analysis, the present invention can discover potential hidden dangers of high-voltage switchgear in advance, generate early warning information, thereby reducing the risks of equipment failures and safety accidents; the risk identification list provides clear hidden danger characteristics and occurrence probabilities, which helps maintenance personnel quickly locate problems and take targeted measures, improving maintenance efficiency; and based on the hidden danger characteristics and occurrence probabilities in the risk identification list, maintenance and replacement plans can be reasonably arranged, optimizing resource allocation and avoiding unnecessary waste; by timely discovering and handling potential hidden dangers, the service life of the high-voltage switchgear can be extended, and the reliability and stability of the equipment can be improved; the risk identification list and early warning information provide data support for maintenance personnel and management, helping to make more informed decisions and improving the overall management level of the substation building.

[0122] In the embodiment of the present application, based on the problem of how to improve the monitoring efficiency of the substation building, a monitoring method for the substation building is designed, which realizes obtaining real-time partial discharge monitoring data and real-time environmental monitoring data of the high-voltage switchgear in the substation building, and constructing a partial discharge diagnostic waveform atlas and a partial discharge diagnostic auxiliary text according to the real-time partial discharge monitoring data and the real-time environmental monitoring data; inputting the partial discharge diagnostic waveform atlas and the partial discharge diagnostic auxiliary text into a pre-constructed graphic and text multi-modal partial discharge classification model for processing to obtain a first classification result of the corresponding partial discharge type of the high-voltage switchgear; determining a plurality of partial discharge mechanism model characteristic values of the high-voltage switchgear according to the partial discharge diagnostic waveform atlas to match with a pre-constructed partial discharge mechanism model, and determining a second classification result of the corresponding partial discharge type of the high-voltage switchgear according to the matching result; determining the target partial discharge type of the high-voltage switchgear based on the first classification result and the second classification result, and determining the abnormal cause of the high-voltage switchgear according to the target partial discharge type, so as to implement the technical solution for monitoring the substation building; improving the accuracy of discriminating the partial discharge type of the switchgear and the monitoring efficiency of the substation building, so as to facilitate precise maintenance of the high-voltage switchgear.

[0123] It should be noted that although the steps in the above flow chart are sequentially shown according to the indication of the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders.

[0124] In another embodiment, as Figure 3 shown, the second aspect of the present invention provides a monitoring system for a substation building, including:

[0125] A partial discharge information acquisition module 10, configured to obtain real-time partial discharge monitoring data and real-time environmental monitoring data of the high-voltage switchgear in the substation building, and construct a partial discharge diagnostic waveform atlas and a partial discharge diagnostic auxiliary text according to the real-time partial discharge monitoring data and the real-time environmental monitoring data;

[0126] The first result generation module 20 is configured to input the partial discharge diagnosis waveform atlas and the partial discharge diagnosis auxiliary text into a pre-constructed graphic and text multi-modal partial discharge classification model for processing, so as to obtain a first classification result of the corresponding partial discharge type of the high-voltage switchgear;

[0127] The second result generation module 30 is configured to determine a plurality of partial discharge mechanism model characteristic values of the high-voltage switchgear according to the partial discharge diagnosis waveform atlas, so as to match with a pre-constructed partial discharge mechanism model, and determine a second classification result of the corresponding partial discharge type of the high-voltage switchgear according to the matching result;

[0128] The abnormal cause determination module 40 is configured to determine the target partial discharge type of the high-voltage switchgear based on the first classification result and the second classification result, and determine the abnormal cause of the high-voltage switchgear according to the target partial discharge type, so as to realize the monitoring of the substation building.

[0129] It should be noted that each module in the above monitoring system for a substation building can be implemented in whole or in part by software, hardware, and their combination. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above respective modules. For the specific limitations of a monitoring system for a substation building, refer to the limitations of a monitoring method for a substation building in the above text. The two have the same functions and effects, and will not be elaborated here.

[0130] The third aspect of the present invention provides an electronic device, which includes:

[0131] A processor, a memory, and a bus;

[0132] The bus is used to connect the processor and the memory;

[0133] The memory is used to store operation instructions;

[0134] The processor is configured to execute the operations corresponding to a monitoring method for a substation building as shown in the first aspect of the present application by calling the operation instructions.

[0135] In an alternative embodiment, an electronic device is provided, as Figure 4 shown Figure 4The electronic device 5000 shown includes: a processor 5001 and a memory 5003. Among them, the processor 5001 and the memory 5003 are connected, such as connected by a bus 5002. Optionally, the electronic device 5000 may further include a transceiver 5004. It should be noted that in actual applications, the transceiver 5004 is not limited to one, and the structure of the electronic device 5000 does not constitute a limitation on the embodiments of the present application.

[0136] The processor 5001 can be a CPU, a general-purpose processor, a DSP, an ASIC, an FPGA or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It can implement or execute various exemplary logic blocks, modules and circuits described in connection with the disclosure of the present application. The processor 5001 can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0137] The bus 5002 may include a path for transmitting information between the above components. The bus 5002 can be a PCI bus or an EISA bus, etc. The bus 5002 can be divided into an address bus, a data bus, a control bus, etc. For the sake of representation, Figure 4 only a thick line is used to represent it in the figure, but it does not mean that there is only one bus or one type of bus.

[0138] The memory 5003 can be a ROM or other types of static storage devices that can store static information and instructions, a RAM or other types of dynamic storage devices that can store information and instructions, or it can also be an EEPROM, a CD-ROM or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.

[0139] The memory 5003 is used to store the application program code for implementing the solution of the present application and is controlled by the processor 5001 to execute. The processor 5001 is used to execute the application program code stored in the memory 5003 to implement the content shown in any of the foregoing method embodiments.

[0140] Among them, the electronic device includes but is not limited to: mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle terminals (such as vehicle navigation terminals), etc. and fixed terminals such as digital TVs, desktop computers, etc.

[0141] In the fourth aspect of the present invention, a computer-readable storage medium is provided. A computer program is stored on the computer-readable storage medium, and when the program is executed by a processor, it implements a monitoring method for a distribution substation shown in the first aspect of the present application.

[0142] Another embodiment of the present application provides a computer-readable storage medium. A computer program is stored on the computer-readable storage medium, and when it runs on a computer, it enables the computer to execute the corresponding content in the foregoing method embodiments.

[0143] In addition, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the steps of the above method.

[0144] In summary, the present invention relates to the technical field of distribution substation monitoring. It discloses a monitoring method, system, device and medium for a distribution substation. By using the real-time partial discharge monitoring data and real-time environmental monitoring data of high-voltage switchgears in the distribution substation, a partial discharge diagnosis waveform map and a partial discharge diagnosis auxiliary text are constructed, and they are input into a pre-constructed text-image multi-modal partial discharge classification model for processing to obtain a first classification result of the partial discharge type to which the high-voltage switchgear belongs; multiple partial discharge mechanism model characteristic values of the partial discharge diagnosis waveform map are determined and matched with a pre-constructed partial discharge mechanism model, and then a second classification result is determined according to the matching result; based on the first and second classification results, the target partial discharge type of the high-voltage switchgear is determined, and then the abnormal cause of the high-voltage switchgear is determined to realize the monitoring of the distribution substation, improve the accuracy of the discrimination of the partial discharge type of the switchgear and the monitoring efficiency of the distribution substation, so as to facilitate the precise maintenance of the high-voltage switchgear.

[0145] Each embodiment in this specification is described in a progressive manner. For parts that are the same or similar in each embodiment, they can be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0146] The above-described embodiments merely represent several preferred embodiments of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the technical principles of the present invention, several improvements and substitutions can be made, and these improvements and substitutions should also be regarded as the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the protection scope of the claims.

Claims

1. A method for monitoring a power distribution station, characterized in that: include: Acquire real-time partial discharge monitoring data and real-time environmental monitoring data of the high-voltage switchgear in the power distribution station, and construct a partial discharge diagnosis waveform spectrum and a partial discharge diagnosis auxiliary text according to the real-time partial discharge monitoring data and the real-time environmental monitoring data; Inputting the partial discharge diagnosis waveform graph and the partial discharge diagnosis auxiliary text into a pre-constructed graphic and text multi-modal partial discharge classification model for processing, to obtain a first classification result of the partial discharge type corresponding to the high-voltage switchgear; Determine multiple partial discharge mechanism model characteristic values ​​of the high-voltage switchgear according to the partial discharge diagnostic waveform spectrum to match with the pre-constructed partial discharge mechanism model, and determine the second classification result of the partial discharge type corresponding to the high-voltage switchgear according to the matching result; the partial discharge mechanism model characteristic values ​​include amplitude, number of discharges, discharge time interval, 50Hz and 100Hz frequency component comparison, positive and negative semi-axis symmetry, single peak or double peak characteristics, classification diagram, ultrasonic detection probability and PRPS characteristics; Determining a target partial discharge type of the high-voltage switchgear based on the first classification result and the second classification result, and determining an abnormal cause of the high-voltage switchgear according to the target partial discharge type, so as to realize monitoring of the distribution station; The constructing of a partial discharge diagnosis waveform graph and a partial discharge diagnosis auxiliary text according to the real-time partial discharge monitoring data and the real-time environmental monitoring data includes: The partial discharge diagnostic waveform spectrum is constructed according to the real-time partial discharge monitoring data, and the mechanism characteristics of the real-time partial discharge monitoring data are extracted to obtain a plurality of partial discharge mechanism characteristic components; the partial discharge mechanism characteristic components include amplitude dispersion, phase aggregation, polarity effect, flight diagram characteristics, pulse balance, 50 Hz frequency component and 100 Hz frequency component; Quantifying the correlation between the real-time partial discharge monitoring data and the real-time environmental monitoring data, and generating a partial discharge external influencing factor according to the correlation quantification result; Performing data discretization and text escape processing on each of the partial discharge mechanism characteristic components and the partial discharge external influencing factors, and expressing the processing results in a key-value pair manner to obtain the partial discharge diagnosis auxiliary text; The real-time environmental monitoring data includes real-time temperature data and real-time humidity data; wherein, The quantifying the correlation between the real-time partial discharge monitoring data and the real-time environmental monitoring data, and generating a partial discharge external influencing factor according to the correlation quantification result, comprises: Performing Pearson correlation analysis and calculation on the real-time partial discharge monitoring data and the real-time temperature data in the same time period to obtain a partial discharge-temperature correlation coefficient; Performing Pearson correlation analysis and calculation on the real-time partial discharge monitoring data and the real-time humidity data in the same time period to obtain a partial discharge-humidity correlation coefficient; The partial discharge temperature correlation coefficient and the partial discharge humidity correlation coefficient are weighted according to a preset ratio to obtain the partial discharge external influencing factor.

2. A method for monitoring a power distribution station according to claim 1, characterized in that: The image-text multimodal partial discharge classification model includes an image encoder, a text encoder, an image-text multimodal encoder and an image-text multimodal decoder; wherein, The inputting of the partial discharge diagnosis waveform graph and the partial discharge diagnosis auxiliary text into a pre-built graphic and text multi-modal partial discharge classification model for processing to obtain a first classification result of the partial discharge type corresponding to the high-voltage switchgear includes: Encoding the partial discharge diagnosis waveform spectrum through the image encoder to obtain a partial discharge spectrum feature vector, and encoding the partial discharge diagnosis auxiliary text based on the text encoder to obtain a partial discharge text feature vector; Based on the cross-modal feature alignment loss function and the image-text matching loss function, the partial discharge spectrum feature vector and the partial discharge text feature vector are input into the image-text multimodal encoder for processing to obtain an image-text encoding result; The image-text encoding result is input into the image-text multimodal decoder for processing, and a first classification result of the partial discharge type corresponding to the high-voltage switch cabinet is output.

3. A method for monitoring a power distribution station according to claim 1, characterized in that: Determining multiple partial discharge mechanism model characteristic values ​​of the high-voltage switchgear according to the partial discharge diagnosis waveform spectrum to match with a pre-constructed partial discharge mechanism model, and determining a second classification result of the partial discharge type corresponding to the high-voltage switchgear according to the matching result, includes: quantifying multiple partial discharge mechanism model characteristic values ​​of the partial discharge diagnostic waveform spectrum; The characteristic values ​​of the multiple partial discharge mechanism models are matched with the partial discharge mechanism models corresponding to the multiple pre-constructed partial discharge types to obtain a plurality of matching degrees, and the partial discharge type corresponding to the partial discharge mechanism model with the highest matching degree is taken as the second classification result.

4. A method for monitoring a power distribution station according to claim 3, characterized in that: The first classification result includes the partial discharge type and its corresponding probability value; The determining a target partial discharge type of the high-voltage switchgear based on the first classification result and the second classification result includes: Determine a first weight value of the first classification result according to the probability value, and determine a second weight value of the second classification result based on the matching degree; The first weight value and the second weight value are normalized to obtain a normalized weight value, and a target partial discharge type of the high-voltage switchgear is determined according to the normalized weight value, the first classification result, and the second classification result.

5. A method for monitoring a power distribution station according to claim 1, characterized in that: After determining the abnormal cause of the high-voltage switchgear according to the target partial discharge type, the method further includes: Constructing a hidden danger feature set of the high-voltage switch cabinet; the hidden danger feature set includes excessive humidity in the station room, equipment aging and family defects; Acquire equipment information of the high-voltage switch cabinet, and when it is determined that the high-voltage switch cabinet meets the hidden danger feature set according to the equipment information and the real-time environmental monitoring data, determine the occurrence probability of each hidden danger feature to generate a risk identification list; Early warning information of the power distribution station is generated according to the risk identification list to achieve early warning of the power distribution station.

6. A monitoring system for a power distribution station, characterized in that: include: A partial discharge information acquisition module is used to acquire real-time partial discharge monitoring data and real-time environmental monitoring data of the high-voltage switchgear in the power distribution station, and to construct a partial discharge diagnosis waveform spectrum and a partial discharge diagnosis auxiliary text according to the real-time partial discharge monitoring data and the real-time environmental monitoring data; A first result generating module is used to input the partial discharge diagnosis waveform spectrum and the partial discharge diagnosis auxiliary text into a pre-built graphic and text multi-modal partial discharge classification model for processing, so as to obtain a first classification result of the partial discharge type corresponding to the high-voltage switchgear; A second result generating module is used to determine multiple partial discharge mechanism model characteristic values ​​of the high-voltage switchgear according to the partial discharge diagnostic waveform spectrum, so as to match them with the pre-constructed partial discharge mechanism model, and determine the second classification result of the partial discharge type corresponding to the high-voltage switchgear according to the matching result; the partial discharge mechanism model characteristic values ​​include amplitude, number of discharges, discharge time interval, 50Hz and 100Hz frequency component comparison, positive and negative semi-axis symmetry, single peak or double peak characteristics, classification diagram, ultrasonic detection probability and PRPS characteristics; an abnormality cause determination module, used to determine a target partial discharge type of the high-voltage switchgear based on the first classification result and the second classification result, and determine the abnormality cause of the high-voltage switchgear according to the target partial discharge type, so as to realize monitoring of the distribution station; The constructing of a partial discharge diagnosis waveform graph and a partial discharge diagnosis auxiliary text according to the real-time partial discharge monitoring data and the real-time environmental monitoring data includes: The partial discharge diagnostic waveform spectrum is constructed according to the real-time partial discharge monitoring data, and the mechanism characteristics of the real-time partial discharge monitoring data are extracted to obtain a plurality of partial discharge mechanism characteristic components; the partial discharge mechanism characteristic components include amplitude dispersion, phase aggregation, polarity effect, flight diagram characteristics, pulse balance, 50 Hz frequency component and 100 Hz frequency component; Quantifying the correlation between the real-time partial discharge monitoring data and the real-time environmental monitoring data, and generating a partial discharge external influencing factor according to the correlation quantification result; Performing data discretization and text escape processing on each of the partial discharge mechanism characteristic components and the partial discharge external influencing factors, and expressing the processing results in a key-value pair manner to obtain the partial discharge diagnosis auxiliary text; The real-time environmental monitoring data includes real-time temperature data and real-time humidity data; wherein, The quantifying the correlation between the real-time partial discharge monitoring data and the real-time environmental monitoring data, and generating a partial discharge external influencing factor according to the correlation quantification result, comprises: Performing Pearson correlation analysis and calculation on the real-time partial discharge monitoring data and the real-time temperature data in the same time period to obtain a partial discharge-temperature correlation coefficient; Performing Pearson correlation analysis and calculation on the real-time partial discharge monitoring data and the real-time humidity data in the same time period to obtain a partial discharge-humidity correlation coefficient; The partial discharge temperature correlation coefficient and the partial discharge humidity correlation coefficient are weighted according to a preset ratio to obtain the partial discharge external influencing factor.

7. An electronic device, characterized in that: The system comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements the method for monitoring a power distribution station as claimed in any one of claims 1 to 5 when executing the computer program.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the device where the computer-readable storage medium is located executes the computer program, the method for monitoring a power distribution station according to any one of claims 1 to 5 is implemented.

Images