Remote intelligent inspection system for transformer substation

Abstract

The invention discloses a remote intelligent substation inspection system, which relates to the technical field of substation inspection and comprises an inspection module, an inspection management module and an inspection data analysis module. The patrolling module comprises a fixed patrolling module and a mobile patrolling module, the fixed patrolling module is fixedly installed at a preset position of the transformer substation to monitor a fixed area, and the mobile patrolling module carries out patrolling in the transformer substation according to a preset patrolling line to monitor an area on the patrolling line and generates monitoring data; according to the remote intelligent inspection system for the transformer substation, the fixed inspection module and the mobile inspection module are arranged to monitor and inspect the transformer substation, visible light images, infrared thermal imaging and sound data of all targets in the transformer substation are collected and analyzed, and abnormal positions and abnormal items are reflected in real time through a transformer substation data twinborn model; and the maintenance personnel can more intuitively master the abnormal condition of the transformer substation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of substation inspection, in particular to a remote intelligent inspection system for substation. BACKGROUND

[0002] With the rapid development of company power grid, the problems such as heavy workload, long time, declining management lean level and "can't manage" are increasingly prominent, which cannot meet the requirements of lean management of substation. Although some existing technologies improve the inspection efficiency through intelligent inspection of inspection robots, the inspection results are not intuitive enough, the maintenance personnel cannot quickly grasp the overall and abnormal conditions of the substation, and the intelligent inspection generates a large amount of inspection data, which greatly increases the processing pressure of the maintenance personnel. In addition, a large number of light or false alarms can easily lead to fatigue and reduced sensitivity of the maintenance personnel, which is not conducive to the timely discovery and processing of important alarms. SUMMARY

[0003] The present application aims to provide a remote intelligent inspection system for substation to solve the above problems in the prior art.

[0004] In order to achieve the above-mentioned purpose, the present application provides the following technical scheme: a remote intelligent inspection system for substation, comprising a patrol module, a patrol management module and a patrol data analysis module;

[0005] The patrol module comprises a fixed patrol module and a mobile patrol module, the fixed patrol module is fixedly installed at a predetermined position of the substation to monitor the fixed area, and the mobile patrol module patrols along a preset patrol line in the substation to monitor the area on the patrol line and generates monitoring data, the monitoring data comprising visible light image data, infrared thermal imaging data, sound data and monitoring position data;

[0006] The patrol management module is used for acquiring the monitoring data of the patrol module and managing the predetermined position of the fixed patrol module and the patrol line of the mobile patrol module;

[0007] The patrol data analysis module is configured to preprocess and standardize the monitoring data, perform similarity analysis based on the set target normal visible light image sample data, target normal infrared thermal imaging sample data, and normal target sound sample data, and the processed monitoring data; when the similarity analysis result is a target normal, perform target recognition on the visible light image data based on a pre-trained target detection algorithm to generate normal patrol result data; when the similarity analysis result is a target abnormal, identify the target abnormal items of the monitoring data based on a pre-trained abnormal type recognition algorithm to generate abnormal patrol result data; analyze the abnormal processing priority based on the feature deviation degree of the identified abnormal patrol result data and the corresponding target normal visible light image sample data, target normal infrared thermal imaging sample data, or normal target sound sample data, and generate processing priority analysis data.

[0008] Further, the system further comprises an intelligent linkage module and a digital twin model module of a transformer substation;

[0009] The intelligent linkage module is configured to push the monitoring data, normal patrol result data, abnormal patrol result data, and processing priority analysis data to a management personnel and obtain a confirmation instruction of the management personnel, generate corresponding control instruction data after the confirmation instruction, and perform transformer substation management operations according to the control instruction data;

[0010] The digital twin model module of the transformer substation is configured to construct a digital twin model of the transformer substation according to the style and distribution of the transformer substation equipment, and indicate the abnormal position and abnormal items of the transformer substation in real time on the digital twin model of the transformer substation according to the patrol analysis data.

[0011] Further, the monitoring comprises the following steps:

[0012] Visible light information, infrared information, and sound information of a target area are collected to generate visible light image data, infrared thermal imaging data, and sound data, respectively;

[0013] Position information of the corresponding patrol module is obtained to generate monitoring position data;

[0014] The visible light image data, infrared thermal imaging data, sound data, and monitoring position data are collected and combined to generate monitoring data.

[0015] In the process of monitoring the area of the patrol route by the mobile patrol module according to the preset patrol route, the preset patrol route includes a patrol route, a patrol speed, a monitoring azimuth angle, and a pitch angle, and the like.

[0016] Furthermore, the target normal visible light image sample data, target normal infrared thermal imaging sample data, and normal target sound sample data are obtained through the following steps:

[0017] Visible light image data, infrared thermal imaging data, and sound data of equipment, environment, and personnel collected in history and in real time are preprocessed and standardized respectively;

[0018] The visible light image data, infrared thermal imaging data, and sound data collected in the past under normal and abnormal conditions of equipment, environment, and personnel are extracted respectively, and normal visible light image data, normal infrared thermal imaging data, normal sound data, abnormal visible light image data, abnormal infrared thermal imaging data, and abnormal sound data are generated respectively.

[0019] Then, the normal visible light image data, normal infrared thermal imaging data, normal sound data, abnormal visible light image data, abnormal infrared thermal imaging data, and abnormal sound data are classified according to the different monitored targets, and multiple sets of target normal visible light image sample data, multiple sets of target normal infrared thermal imaging sample data, multiple sets of normal target sound sample data, multiple sets of target abnormal visible light image sample data, multiple sets of target abnormal infrared thermal imaging sample data, and multiple sets of abnormal target sound sample data are generated respectively.

[0020] Collect all monitoring location data and merge monitoring location data that are less than the set distance threshold to generate multiple standard monitoring location data;

[0021] Each set of target normal visible light image sample data, target normal infrared thermal imaging sample data, normal target sound sample data, target abnormal visible light image sample data, target abnormal infrared thermal imaging sample data, and abnormal target sound sample data is associated with the standard monitoring location data corresponding to the monitoring location data in the same monitoring data to generate monitoring location and monitoring target association data.

[0022] Search the data associated with the monitoring location and the monitoring target to find the target's normal visible light image sample data, target's normal infrared thermal imaging sample data, and normal target's sound sample data that are closest to the standard monitoring location data in the monitoring data. Generate target's normal visible light image sample data, target's normal infrared thermal imaging sample data, and normal target's sound sample data to be used.

[0023] Furthermore, the patrol data analysis module performs similarity analysis, including the following steps:

[0024] Based on the first DSCAN clustering algorithm, clusters are performed on each group of target normal visible light image sample data, target normal infrared thermal imaging sample data, and target normal sound sample data, and all core points are output during cluster analysis to generate corresponding target normal visible light image sample core point set, target normal infrared thermal imaging sample core point set, and target normal sound sample core point set.

[0025] Calculate the minimum vector distance (such as Euclidean distance, cosine distance, etc.) between all core points in the visible light image data and the corresponding target normal visible light image sample core point set, the infrared thermal imaging data and the corresponding target normal infrared thermal imaging sample core point set, and the sound data and the corresponding target normal sound sample core point set, respectively, and generate the minimum core point distance of visible light, infrared thermal imaging, and sound.

[0026] Determine whether the minimum core point distance for visible light, the minimum core point distance for infrared thermal imaging, and the minimum core point distance for sound are less than the neighborhood radius parameter of the first DSCAN clustering algorithm. If yes, the target is normal; otherwise, the target is abnormal.

[0027] Furthermore, the patrol data analysis module performs target recognition on visible light image data based on a pre-trained target detection algorithm, including the following steps:

[0028] When the similarity analysis result indicates that the target is normal, the target recognition algorithm based on the pre-trained target detection algorithm is used to identify the target in the visible light image data, such as the meter reading, license plate, and face, to generate normal patrol result data.

[0029] Collect and combine normal inspection results data and corresponding monitoring data to generate normal inspection analysis data;

[0030] Whenever the monitoring data corresponding to the normal inspection result data reaches the set update quantity, the visible light image data, infrared thermal imaging data, and sound data in the monitoring data corresponding to the normal inspection result data are used to update the normal visible light image sample data, normal infrared thermal imaging sample data, and normal sound sample data, respectively.

[0031] Furthermore, the patrol data analysis module identifies target anomalies in the monitoring data based on a pre-trained anomaly type recognition algorithm, including the following steps:

[0032] Set up abnormal item labels corresponding to various abnormal items to associate the abnormal visible light image sample data, abnormal infrared thermal imaging sample data, and abnormal target sound sample data with the corresponding data.

[0033] Based on the second DSCAN clustering algorithm, cluster analysis was performed on the target abnormal visible light image sample data, target abnormal infrared thermal imaging sample data and abnormal target sound sample data of various abnormal items, respectively, generating multiple target abnormal visible light image sample data clusters, multiple target abnormal infrared thermal imaging sample data clusters and multiple target abnormal sound sample data clusters;

[0034] The confidence scores of various abnormal item labels appearing in each target abnormal visible light image sample data cluster, target abnormal infrared thermal imaging sample data cluster, and target abnormal sound sample data cluster are calculated, and it is determined whether the highest confidence score is greater than the set confidence threshold.

[0035] If not, adjust the parameters of the second DSCAN clustering algorithm as needed, and then re-perform the clustering analysis after adjusting the clustering features of the target anomalous visible light image sample data, the target anomalous infrared thermal imaging sample data, and the anomalous target sound sample data.

[0036] If so, associate the anomaly item label corresponding to the highest confidence level with the corresponding target anomaly visible light image sample data cluster, target anomaly infrared thermal imaging sample data cluster, or target anomaly sound sample data cluster;

[0037] Calculate the minimum vector distance between the visible light image data and the corresponding abnormal visible light image sample data clusters of each target, the infrared thermal imaging data and the corresponding abnormal infrared thermal imaging sample data clusters of each target, and the sound data and the corresponding abnormal sound sample data clusters of each target in the monitoring data. Output the abnormal item labels of the target abnormal visible light image sample data clusters, target abnormal infrared thermal imaging sample data clusters, and target abnormal sound sample data clusters where the core point corresponding to the minimum vector distance is located, and generate abnormal inspection result data.

[0038] The abnormal inspection results data and corresponding monitoring data are collected and combined to generate abnormal inspection analysis data.

[0039] Furthermore, normal inspection analysis data and abnormal inspection analysis data can be collected to generate inspection analysis data. Statistical processing of the inspection analysis data can then be performed to generate corresponding inspection analysis data statistical tables and inspection analysis data statistical charts.

[0040] Furthermore, the patrol data analysis module is also used to analyze the anomaly handling priority based on the degree of feature deviation between the abnormal patrol result data and the corresponding normal visible light image sample data, normal infrared thermal imaging sample data, or normal target sound sample data, including the following steps:

[0041] For each abnormal item label, a set of reference weights for visible light image data, infrared thermal imaging data, and sound data are set to generate an abnormal item reference weight vector.

[0042] For each abnormal item label, set an abnormality degree mapping function for the visible light image data, infrared thermal imaging data, and sound data;

[0043] Calculate the center feature vectors of the target normal visible light image sample data, the target normal infrared thermal imaging sample data, and the normal target sound sample data respectively, and generate the center vectors of the target normal visible light image sample, the target normal infrared thermal imaging sample, and the normal target sound sample;

[0044] The vector distances between the feature vectors of the visible light image data and the center vector of the normal visible light image sample of the target, the feature vectors of the infrared thermal imaging data and the center vector of the normal infrared thermal imaging sample of the target, and the feature vectors of the sound data and the center vector of the normal sound sample of the target are calculated and normalized to generate image deviation data, thermal imaging deviation data and sound deviation data.

[0045] The image deviation data, thermal imaging deviation data, and sound deviation data are respectively input into the corresponding anomaly degree mapping function to generate image anomaly degree score, thermal imaging anomaly degree score, and sound anomaly degree score;

[0046] Based on the abnormal item reference weight vector, the image abnormality score, thermal imaging abnormality score, and sound abnormality score are weighted and summed to generate a comprehensive abnormality score.

[0047] Set up score intervals for each processing priority to generate priority score interval data. Search for the processing priority corresponding to the priority score interval where the comprehensive anomaly score is located in the priority score interval data to generate processing priority analysis data.

[0048] 1. Compared with the prior art, the present invention provides a remote intelligent inspection system for substations, which monitors and inspects substations by setting up fixed inspection modules and mobile inspection modules. It collects and analyzes visible light images, infrared thermal imaging and sound data of various targets in the substation, and reflects abnormal locations and abnormal items in real time through the substation data twin model, so that maintenance personnel can more intuitively grasp the abnormal situation of the substation.

[0049] 2. Compared with the prior art, the remote intelligent inspection system for substations provided by the present invention, by setting up an inspection data analysis module, first determines whether the data is normal when analyzing visible light images, infrared thermal imaging and sound data, and then judges the specific abnormal items when abnormality occurs. This simplifies the normal judgment process, greatly improves the efficiency of processing massive normal operation data of substations, and avoids the problem of reduced accuracy of abnormality analysis caused by comparing abnormal items with a large number of normal samples.

[0050] 3. Compared with the prior art, the substation remote intelligent inspection system provided by the present invention, after the inspection data analysis module analyzes the abnormal items, sets different mapping functions to map the degree of abnormality of different abnormal items based on the deviation between the monitoring data corresponding to the abnormal items and the monitoring data of the normal state. In this way, the severity of abnormal items can be distinguished by the degree of abnormality, avoiding important abnormalities from being buried in a large number of unimportant abnormal alarms or reminders, and avoiding the reduction of the sensitivity of maintenance personnel by a large number of unimportant abnormal alarms or reminders over a long period of time. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0052] Figure 1 A system structure block diagram provided for embodiments of the present invention;

[0053] Figure 2 This is a diagram illustrating the data processing steps of the patrol data analysis module provided in an embodiment of the present invention.

[0054] Figure 3 This is a flowchart illustrating the steps for analyzing patrol data and determining whether a target is normal, provided in an embodiment of the present invention.

[0055] Figure 4 This is a diagram illustrating the processing steps when the target is normal according to the similarity analysis result of the patrol data analysis module provided in this embodiment of the invention.

[0056] Figure 5 This is a diagram illustrating the processing steps when the target is abnormal in the similarity analysis result of the patrol data analysis module provided in this embodiment of the invention;

[0057] Figure 6 A flowchart illustrating the priority analysis steps for anomaly handling in the inspection data analysis module provided in this embodiment of the invention. Detailed Implementation

[0058] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0059] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified. Furthermore, the terms "installed," "connected," and "linked" should be interpreted broadly; for example, they may refer to a fixed connection, a detachable connection, or an integral connection; they may refer to a mechanical connection or an electrical connection; they may refer to a direct connection or an indirect connection through an intermediate medium; and they may refer to the internal connection of two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0060] Exemplary embodiments will be described more fully below with reference to the accompanying drawings; however, these exemplary embodiments may be embodied in different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will enable those skilled in the art to fully understand the scope of this disclosure.

[0061] Where there is no conflict, the various embodiments of this disclosure and the features thereof in the embodiments may be combined with each other.

[0062] As used herein, the term “and / or” includes any and all combinations of one or more related enumerated entries.

[0063] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit this disclosure. As used herein, the singular forms “a” and “the” are also intended to include the plural forms unless the context clearly indicates otherwise. It will also be understood that when the terms “comprising” and / or “made of” are used in this specification, the presence of the stated feature, integral, step, operation, element, and / or component is specified, but the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof is not excluded.

[0064] The embodiments described herein can be described with reference to plan views and / or cross-sectional views using the ideal schematic diagrams of this disclosure. Therefore, the example illustrations can be modified according to manufacturing techniques and / or tolerances. Therefore, the embodiments are not limited to those shown in the drawings, but include modifications to configurations formed based on manufacturing processes. Therefore, the areas illustrated in the drawings are schematic in nature, and the shapes of the areas shown in the figures illustrate specific shapes of areas of an element, but are not intended to be limiting.

[0065] Unless otherwise specified, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art.

[0066] Please see Figures 1-6 A remote intelligent inspection system for substations includes an inspection module, an inspection management module, an inspection data analysis module, an intelligent linkage module, and a substation digital twin model module.

[0067] The inspection module includes a fixed inspection module and a mobile inspection module. The fixed inspection module is installed at a predetermined location in the substation to monitor a fixed area. The mobile inspection module patrols the substation according to a preset inspection route to monitor the area along the route and generate monitoring data, including visible light image data, infrared thermal imaging data, sound data, and monitoring location data.

[0068] Monitoring includes the following steps:

[0069] A1. Collect visible light, infrared, and sound information of the target area and generate visible light image data, infrared thermal imaging data, and sound data respectively.

[0070] A2. Obtain the location information of the corresponding inspection module and generate monitoring location data;

[0071] A3. Collect and combine visible light image data, infrared thermal imaging data, sound data, and monitoring location data to generate monitoring data.

[0072] During the monitoring process by the mobile patrol module, which follows a preset patrol route, including the patrol path, patrol speed, and monitoring azimuth and elevation angles, a positioning module is also installed on the mobile patrol module. This module enables real-time location tracking and generates corresponding monitoring location data.

[0073] In one embodiment, the fixed inspection module can employ high-definition cameras, infrared cameras, sound sensors, etc. The mobile inspection module can be a wheeled robot, a tracked robot, or a drone, and monitoring is achieved by mounting high-definition cameras, infrared cameras, and sound sensors on the module.

[0074] The patrol management module is used to acquire monitoring data from the patrol modules and manage the predetermined locations of fixed patrol modules and the patrol routes of mobile patrol modules. For example, management includes setting the monitoring locations of fixed patrol modules; adjusting the monitoring angles of fixed patrol modules; and adjusting the time series of the patrol routes of mobile patrol modules, such as setting the travel speed, azimuth angle, and elevation angle of mobile patrol modules on different route segments.

[0075] The patrol data analysis module is used to preprocess and standardize the monitoring data. Based on the set target normal visible light image sample data, target normal infrared thermal image sample data, and normal target sound sample data, as well as the processed monitoring data, it performs similarity analysis, including the following steps:

[0076] B1. Visible light image data, infrared thermal imaging data, and sound data from equipment, environment, and personnel, collected historically and in real-time, are preprocessed and standardized respectively. Preprocessing of visible light image data may include, for example, size normalization, color space conversion, color / contrast enhancement, noise reduction, and smoothing. Preprocessing of infrared thermal imaging data may include, for example, non-uniformity correction, point replacement, radiation calibration and temperature conversion, and image enhancement. Preprocessing of sound data may include, for example, noise reduction, pre-emphasis, framing, windowing, and feature extraction (such as Mel-frequency cepstral coefficients, linear prediction coefficients, and spectral centroids). All data collected in this invention undergoes corresponding preprocessing and standardization to ensure data standardization and eliminate the influence of dimensions.

[0077] B2. Extract the visible light image data, infrared thermal imaging data and sound data collected in the normal and abnormal states of the equipment, environment and personnel respectively, and generate normal visible light image data, normal infrared thermal imaging data, normal sound data, abnormal visible light image data, abnormal infrared thermal imaging data and abnormal sound data respectively.

[0078] B3. Then, the normal visible light image data, normal infrared thermal imaging data, normal sound data, abnormal visible light image data, abnormal infrared thermal imaging data, and abnormal sound data are classified according to the different monitored targets, and multiple sets of target normal visible light image sample data, multiple sets of target normal infrared thermal imaging sample data, multiple sets of normal target sound sample data, multiple sets of target abnormal visible light image sample data, multiple sets of target abnormal infrared thermal imaging sample data, and multiple sets of abnormal target sound sample data are generated respectively.

[0079] B4. Collect all monitoring location data and merge monitoring location data that are less than the set distance threshold to generate multiple standard monitoring location data;

[0080] B5. Associate each set of target normal visible light image sample data, target normal infrared thermal imaging sample data, normal target sound sample data, target abnormal visible light image sample data, target abnormal infrared thermal imaging sample data, and abnormal target sound sample data with the standard monitoring location data corresponding to the monitoring location data in the same monitoring data to generate monitoring location and monitoring target association data.

[0081] B6. Search for target normal visible light image sample data, target normal infrared thermal imaging sample data, and normal target sound sample data that are closest to the standard monitoring location data in the monitoring data in the monitoring data in the monitoring location and monitoring target association data, and generate target normal visible light image sample data, target normal infrared thermal imaging sample data, and normal target sound sample data to be used.

[0082] By comparing the target types corresponding to the monitoring location data in the monitoring data with the standard monitoring location data through steps B1-B3, the target category within the monitoring data can be quickly identified. This allows for finding normal sample data of the corresponding target category for comparative analysis to determine whether it is normal. This reduces the amount of data that needs to be compared and improves the efficiency and accuracy of the comparison.

[0083] C1. Based on the first DSCAN clustering algorithm, cluster the target normal visible light image sample data, target normal infrared thermal imaging sample data and target normal sound sample data for each group, and output all the core points during the clustering analysis to generate the corresponding target normal visible light image sample core point set, target normal infrared thermal imaging sample core point set and target normal sound sample core point set.

[0084] C2. Calculate the minimum vector distance (such as Euclidean distance, cosine distance, etc.) between the visible light image data and the core point set of the corresponding normal visible light image sample of the target, the infrared thermal imaging data and the core point set of the corresponding normal infrared thermal imaging sample of the target, and the core point set of the sound data and the core point set of the corresponding normal sound sample of the target, respectively, and generate the minimum core point distance of visible light, the minimum core point distance of infrared thermal imaging, and the minimum core point distance of sound.

[0085] C3. Determine whether the minimum core point distance for visible light, infrared thermal imaging, and sound is less than the neighborhood radius parameter of the first DSCAN clustering algorithm. If yes, the target is normal; otherwise, the target is abnormal.

[0086] In other embodiments, the visible light image data, infrared thermal imaging data, and sound data from the monitoring data can be added to the corresponding target normal visible light image waiting sample data, target normal infrared thermal imaging waiting sample data, and target normal sound waiting sample data, respectively. Then, DSCAN clustering analysis can be performed on the target normal visible light image waiting sample data, target normal infrared thermal imaging waiting sample data, and target normal sound waiting sample data to determine whether the visible light image data, infrared thermal imaging data, and sound data in the monitoring data are anomalies. Alternatively, the K-nearest neighbor algorithm can be used to analyze whether there are at least K sample data in the target normal visible light image waiting sample data / target normal infrared thermal imaging waiting sample data / target normal sound waiting sample data that are closest to the feature vector of the corresponding visible light image data / infrared thermal imaging data / sound data and whose distance is less than a set distance threshold. If so, the target is normal; otherwise, the target is abnormal.

[0087] When the similarity analysis result indicates that the target is normal, target recognition is performed on the visible light image data based on a pre-trained target detection algorithm to generate normal inspection result data, including the following steps:

[0088] D1. When the similarity analysis result indicates that the target is normal, target recognition is performed on the visible light image data based on the pre-trained target detection algorithm. For example, meter readings, license plates, and faces are identified to generate normal inspection result data. The pre-trained target detection algorithm is common knowledge in the existing technology and is directly applied without modification. Therefore, it will not be described in detail in this technical solution, and it will not cause any problems for the field. By recognizing meter readings, license plates, and faces from the visible light image data, manual meter reading is eliminated, greatly improving the efficiency of meter reading. Furthermore, the recognition of license plates and faces can directly identify external vehicles and personnel, making it more convenient for personnel management in substations.

[0089] D2. Collect and combine normal inspection result data and corresponding monitoring data to generate normal inspection analysis data;

[0090] D3. Whenever the monitoring data corresponding to the normal inspection result data reaches the set update quantity, use the visible light image data, infrared thermal imaging data, and sound data in the monitoring data corresponding to the normal inspection result data to update the normal visible light image sample data, normal infrared thermal imaging sample data, and normal sound sample data respectively.

[0091] When the similarity analysis result indicates a target anomaly, the target anomaly item in the monitoring data is identified based on a pre-trained anomaly type recognition algorithm, generating anomaly inspection result data, including the following steps:

[0092] E1. Set abnormal item labels corresponding to various abnormal items to associate the abnormal visible light image sample data, abnormal infrared thermal imaging sample data, and abnormal target sound sample data of the target;

[0093] E2. Based on the second DSCAN clustering algorithm, cluster analysis is performed on the target abnormal visible light image sample data, target abnormal infrared thermal imaging sample data, and abnormal target sound sample data of various abnormal items, respectively, to generate multiple target abnormal visible light image sample data clusters, multiple target abnormal infrared thermal imaging sample data clusters, and multiple target abnormal sound sample data clusters; the parameters of the second DSCAN clustering algorithm are different from those of the first DSCAN clustering algorithm, and the first and second are used to distinguish them;

[0094] E3. Calculate the confidence level of each type of abnormal item label in each target abnormal visible light image sample data cluster, target abnormal infrared thermal imaging sample data cluster, and target abnormal sound sample data cluster, and determine whether the highest confidence level is greater than the set confidence threshold.

[0095] E4. If not, adjust the parameters of the second DSCAN clustering algorithm as needed, and then re-perform the clustering analysis after adjusting the clustering characteristics of the target abnormal visible light image sample data, the target abnormal infrared thermal imaging sample data, and the abnormal target sound sample data.

[0096] E5. If so, associate the anomaly item label corresponding to the highest confidence level with the corresponding target anomaly visible light image sample data cluster, target anomaly infrared thermal imaging sample data cluster, or target anomaly sound sample data cluster.

[0097] E6. Calculate the minimum vector distance between the visible light image data and the corresponding abnormal visible light image sample data clusters of each target, the infrared thermal imaging data and the corresponding abnormal infrared thermal imaging sample data clusters of each target, and the sound data and the corresponding abnormal sound sample data clusters of each target in the monitoring data. Output the abnormal item labels of the target abnormal visible light image sample data cluster, target abnormal infrared thermal imaging sample data cluster, and target abnormal sound sample data cluster where the core point corresponding to the minimum vector distance is located, and generate abnormal inspection result data. Similarly, whenever the monitoring data corresponding to the abnormal inspection result data reaches the set update quantity, use the visible light image data, infrared thermal imaging data, and sound data in the monitoring data corresponding to the abnormal inspection result data to update the abnormal visible light image sample data, abnormal infrared thermal imaging sample data, and abnormal sound sample data, respectively.

[0098] E7. Collect and combine abnormal inspection results data and corresponding monitoring data to generate abnormal inspection analysis data.

[0099] Furthermore, it can collect normal and abnormal inspection analysis data, generate inspection analysis data, perform statistical processing on the inspection analysis data, and generate corresponding inspection analysis data statistical tables, inspection analysis data statistical charts, and analysis reports.

[0100] Based on the degree of deviation between the identified abnormal inspection results data and the corresponding normal visible light image sample data, normal infrared thermal image sample data, or normal target sound sample data, the anomaly processing priority is analyzed, and processing priority analysis data is generated, including the following steps:

[0101] F1. Set a set of reference weights for visible light image data, infrared thermal imaging data, and sound data for each abnormal item label, generating an abnormal item reference weight vector. The data corresponding to each dimension in the abnormal item parameter weight vector is the reference weight value of an abnormal item. For example, if an abnormal low temperature area is detected by infrared thermal imaging, and the same area is detected by visible light image to determine whether it is a shadow or oil stain, then both visible light image data and infrared thermal imaging data are needed to determine the abnormal item. Sound data has no effect on determining whether the low temperature area is a shadow or oil stain, so the sound data can be set to 0, and the visible light image data and infrared thermal imaging data can be set to 0.4-0.6 (ensuring that the total weight is equal to 1).

[0102] F2. Set an anomaly degree mapping function for the visible light image data, infrared thermal imaging data, and sound data of each abnormal item label. For data where the anomaly degree gradually increases or decreases as the data deviates from the corresponding normal state data, piecewise functions, linear functions, or saturation functions can be selected for mapping. For data where the anomaly degree increases or decreases sharply as the data deviates significantly from the corresponding normal state data, an exponential function can be selected to amplify the data. For example, if high-frequency sound waves generated by discharge are captured by sound data, it indicates the presence of corona discharge or partial discharge. At the same time, if infrared thermal imaging detects a slight temperature rise at the corresponding location, it indicates severe heat accumulation from the discharge. An exponential function can be used to map the small value of the slight temperature rise to a larger value when the discharge phenomenon occurs, thereby rapidly increasing the anomaly (severity) of the discharge phenomenon. If the anomaly itself has a significant impact, the minimum value and rate of change of the function mapping can be increased by setting a bias term or control coefficient.

[0103] F3. Calculate the center feature vectors of the target normal visible light image sample data, the target normal infrared thermal imaging sample data, and the normal target sound sample data respectively, and generate the center vectors of the target normal visible light image sample, the target normal infrared thermal imaging sample, and the normal target sound sample.

[0104] F4. Calculate the vector distance between the feature vector of the visible light image data and the center vector of the normal visible light image sample of the target, the feature vector of the infrared thermal imaging data and the center vector of the normal infrared thermal imaging sample of the target, and the feature vector of the sound data and the center vector of the normal sound sample of the target in the monitoring data corresponding to the abnormal inspection results data, and perform normalization processing to generate image deviation data, thermal imaging deviation data and sound deviation data.

[0105] F5. Input the image deviation data, thermal imaging deviation data, and sound deviation data into the corresponding anomaly degree mapping function to generate image anomaly degree score, thermal imaging anomaly degree score, and sound anomaly degree score;

[0106] F6. Based on the abnormal item reference weight vector, the image abnormality score, thermal imaging abnormality score, and sound abnormality score are weighted and summed to generate a comprehensive abnormality score.

[0107] F7. Set the score intervals corresponding to each processing priority to generate priority score interval data. Search for the processing priority corresponding to the priority score interval where the comprehensive abnormality score is located in the priority score interval data to generate processing priority analysis data.

[0108] The intelligent linkage module is used to push monitoring data, normal inspection result data, abnormal inspection result data and processing priority analysis data to the management personnel and obtain the confirmation instructions from the management personnel. After the confirmation instructions are received, the corresponding control instruction data is generated, and the substation management operation is executed according to the control instruction data. The control instructions can be pre-set to be associated with the normal inspection result data, abnormal inspection result data and processing priority analysis data of various situations, so that when the corresponding normal inspection result data, abnormal inspection result data and processing priority analysis data appear, the corresponding control instruction data will be automatically searched out, and these data and the corresponding monitoring data will be pushed to the maintenance personnel for confirmation. After the maintenance personnel confirm, the control instruction data will be automatically executed, realizing "one-click sequential control", which greatly improves the operation efficiency, reduces the operation risk, and improves the emergency response capability. For example, the linkage between the main equipment SCADA monitoring system change signal and protection alarm signal is realized; (2) the linkage function with the auxiliary control system for weather, fire protection, security, access control and other statuses is realized.

[0109] The substation digital twin model module is used to construct a substation data twin model based on the style and distribution of substation equipment. It then uses inspection and analysis data to indicate the location and status of any anomalies in the substation data twin model in real time. A 3D model can be built first based on the substation structure. Then, based on corresponding monitoring data, anomaly inspection results data, and processing priority analysis data, the module searches for the location of the monitoring data corresponding to the anomaly inspection results data within the monitoring data on the 3D model, highlighting it with different colors. Specific anomaly inspection results data, corresponding monitoring data, and processing priority analysis data can be displayed through settings windows and secondary windows.

[0110] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A remote intelligent inspection system for substations, characterized in that, The system comprises a patrol module, a patrol management module and a patrol data analysis module. The patrol module comprises a fixed patrol module and a mobile patrol module, the fixed patrol module is fixedly installed at a predetermined position of the substation to monitor a fixed area, the mobile patrol module patrols along a preset patrol line in the substation to monitor an area along the patrol line, and generates monitoring data, the monitoring data comprises visible light image data, infrared thermal imaging data, sound data and monitoring position data. The patrol management module is used for acquiring the monitoring data of the patrol module, managing the predetermined position of the fixed patrol module and the patrol line of the mobile patrol module. The patrol data analysis module is used for pre-processing and standardizing the monitoring data, performing similarity analysis based on the set target normal visible light image standby sample data, target normal infrared thermal imaging standby sample data and normal target sound standby sample data, and the processed monitoring data; when the similarity analysis result is target normal, performing target recognition on the visible light image data based on a pre-trained target detection algorithm to generate normal patrol result data; When the similarity analysis result is target abnormal, identifying target abnormal items of the monitoring data based on a pre-trained abnormal type recognition algorithm to generate abnormal patrol result data; Based on the feature deviation degree of the identified abnormal patrol result data from the corresponding target normal visible light image sample data, target normal infrared thermal imaging sample data or normal target sound sample data, analyzing the abnormal processing priority to generate processing priority analysis data.

2. The remote intelligent inspection system for a substation of claim 1, wherein, The system further comprises an intelligent linkage module and a substation digital twin model module. The intelligent linkage module is used for pushing the monitoring data, normal patrol result data, abnormal patrol result data and processing priority analysis data to a management personnel and acquiring a confirmation instruction of the management personnel, generating corresponding control instruction data after the confirmation instruction, and performing substation management operation according to the control instruction data; The substation digital twin model module is used for constructing a substation data twin model according to the style and distribution of substation equipment, and indicating the abnormal position and abnormal items of the substation in real time on the substation data twin model according to the patrol analysis data.

3. The remote intelligent inspection system for a substation of claim 1, wherein, The monitoring comprises the following steps: Collecting visible light information, infrared information and sound information of a target area to generate visible light image data, infrared thermal imaging data and sound data respectively; Acquiring position information of a corresponding patrol module to generate monitoring position data; Collecting and combining the visible light image data, infrared thermal imaging data, sound data and monitoring position data to generate monitoring data.

4. The remote intelligent inspection system for a substation of claim 1, wherein, The target normal visible light image standby sample data, target normal infrared thermal imaging standby sample data and normal target sound standby sample data are obtained by the following steps: Pre-processing and standardizing the visible light image data, infrared thermal imaging data and sound data of devices, environments and personnel collected historically and in real time respectively; Extract visible light image data, infrared thermal imaging data and sound data collected under normal and abnormal states of equipment, environment and personnel respectively, to generate normal visible light image data, normal infrared thermal imaging data, normal sound data, abnormal visible light image data, abnormal infrared thermal imaging data and abnormal sound data respectively; Then, classify the normal visible light image data, normal infrared thermal imaging data, normal sound data, abnormal visible light image data, abnormal infrared thermal imaging data and abnormal sound data according to different monitored targets, to generate multiple sets of target normal visible light image sample data, multiple sets of target normal infrared thermal imaging sample data, multiple sets of normal target sound sample data, multiple sets of target abnormal visible light image sample data, multiple sets of target abnormal infrared thermal imaging sample data and multiple sets of abnormal target sound sample data; Collect all monitoring position data and merge the monitoring position data with a distance less than a set distance threshold to generate multiple standard monitoring position data; Correlate each set of target normal visible light image sample data, target normal infrared thermal imaging sample data, normal target sound sample data, target abnormal visible light image sample data, target abnormal infrared thermal imaging sample data and abnormal target sound sample data with the standard monitoring position data corresponding to the monitoring position data in the same monitoring data, to generate monitoring position and monitoring target correlation data; Search for target normal visible light image sample data, target normal infrared thermal imaging sample data and normal target sound sample data associated with the standard monitoring position data closest to the monitoring position data in the monitoring data from the monitoring position and monitoring target correlation data, to generate target normal visible light image standby sample data, target normal infrared thermal imaging standby sample data and normal target sound standby sample data.

5. The remote intelligent inspection system for substations of claim 1, wherein, The patrol data analysis module performs similarity analysis, including the following steps: Based on the first DSCAN clustering algorithm, cluster each set of target normal visible light image standby sample data, target normal infrared thermal imaging standby sample data and target normal sound standby sample data respectively, and output all core points in the clustering analysis, to generate a corresponding target normal visible light image sample core point set, a target normal infrared thermal imaging sample core point set and a target normal sound sample core point set; Calculate the minimum vector distance of all core points in the corresponding target normal visible light image sample core point set, the corresponding target normal infrared thermal imaging sample core point set and the corresponding target normal sound sample core point set from the visible light image data, the infrared thermal imaging data and the sound data in the monitoring data respectively, to generate a visible light minimum core point distance, an infrared thermal imaging minimum core point distance and a sound minimum core point distance; Determine whether the visible light minimum core point distance, the infrared thermal imaging minimum core point distance and the sound minimum core point distance are less than the neighborhood radius parameter of the first DSCAN clustering algorithm, if yes, the target is normal, if not, the target is abnormal.

6. The remote intelligent inspection system for substations of claim 1, wherein, The patrol data analysis module performs target recognition on the visible light image data based on a pre-trained target detection algorithm, including the following steps: When the similarity analysis result is normal, a target detection algorithm is used to identify the visible light image data, and normal patrol result data is generated; The normal patrol result data and corresponding monitoring data are collected and combined to generate normal patrol analysis data; When the monitoring data corresponding to the normal patrol result data reaches a set update quantity, the visible light image data, infrared thermal imaging data, and sound data in the monitoring data are used to update the normal visible light image sample data, normal infrared thermal imaging sample data, and normal sound sample data, respectively.

7. The remote intelligent inspection system for substations of claim 1, wherein, The patrol data analysis module identifies target abnormal items in the monitoring data based on a pre-trained abnormal type identification algorithm, including the following steps: An abnormal item label corresponding to each type of abnormal item is set to correspondingly associate the target abnormal visible light image sample data, target abnormal infrared thermal imaging sample data, and abnormal target sound sample data; The target abnormal visible light image sample data, target abnormal infrared thermal imaging sample data, and abnormal target sound sample data of each type of abnormal item are analyzed by clustering based on a second DSCAN clustering algorithm to generate multiple target abnormal visible light image sample data clusters, multiple target abnormal infrared thermal imaging sample data clusters, and multiple target abnormal sound sample data clusters; The confidence of each type of abnormal item label appearing in each target abnormal visible light image sample data cluster, target abnormal infrared thermal imaging sample data cluster, and target abnormal sound sample data cluster is calculated, and it is determined whether the highest confidence is greater than a set confidence threshold; If not, the parameters of the second DSCAN clustering algorithm and the clustering features of the target abnormal visible light image sample data, target abnormal infrared thermal imaging sample data, and abnormal target sound sample data are adjusted as needed, and the clustering analysis is performed again; If yes, the abnormal item label corresponding to the highest confidence is associated with the corresponding target abnormal visible light image sample data cluster, target abnormal infrared thermal imaging sample data cluster, or target abnormal sound sample data cluster; The minimum vector distance of all core points in the visible light image data in the monitoring data and the corresponding target abnormal visible light image sample data cluster, the infrared thermal imaging data and the target abnormal infrared thermal imaging sample data cluster, and the sound data and the corresponding target abnormal sound sample data cluster are calculated, and the abnormal item label of the target abnormal visible light image sample data cluster, target abnormal infrared thermal imaging sample data cluster, and target abnormal sound sample data cluster where the core point corresponding to the minimum vector distance is located is output to generate abnormal patrol result data; The abnormal patrol result data and corresponding monitoring data are collected and combined to generate abnormal patrol analysis data.

8. The remote intelligent inspection system for substations of claim 1, wherein, The patrol data analysis module generates processing priority analysis data, including the following steps: A set of reference weights of visible light image data, infrared thermal imaging data, and sound data is set for each abnormal item label to generate an abnormal item reference weight vector; An abnormal degree mapping function is set for the visible light image data, infrared thermal imaging data, and sound data of each abnormal item label, respectively. The center feature vectors of the target normal visible light image sample data, the target normal infrared thermal imaging sample data and the normal target sound sample data are respectively calculated to generate a target normal visible light image sample center vector, a target normal infrared thermal imaging sample center vector and a normal target sound sample center vector; The vector distances between the feature vectors of the visible light image data in the monitoring data corresponding to the abnormal patrol result data and the target normal visible light image sample center vector, the feature vectors of the infrared thermal imaging data and the target normal infrared thermal imaging sample center vector, and the feature vectors of the sound data and the normal target sound sample center vector are respectively calculated and normalized to generate image deviation degree data, thermal imaging deviation degree data and sound deviation degree data; The image deviation degree data, the thermal imaging deviation degree data and the sound deviation degree data are respectively input into corresponding abnormal degree mapping functions to generate an image abnormal degree score, a thermal imaging abnormal degree score and a sound abnormal degree score; The image abnormal degree score, the thermal imaging abnormal degree score and the sound abnormal degree score are weighted and summed based on an abnormal item reference weight vector to generate a comprehensive abnormal degree score; The score intervals corresponding to the processing priorities are set to generate priority score interval data, and the processing priority corresponding to the priority score interval in which the comprehensive abnormal degree score is located is searched in the priority score interval data to generate processing priority analysis data.

Images