A multi-sensor-based fault hidden danger identification method and a suspended rail inspection robot

Abstract

The application provides a multi-sensor-based fault hidden danger identification method and a rail-mounted inspection robot. The method comprises the following steps: the rail-mounted inspection robot obtains the type attribute of a target electrical equipment based on a high-definition image of the target electrical equipment, filters a plurality of detection sensors according to the type attribute, and determines the confidence priority coefficients of the detection sensors; the detection sensors are controlled to detect the target electrical equipment to obtain a plurality of groups of detection data, and the fault probability of the target electrical equipment is determined according to the plurality of groups of detection data and the corresponding confidence priority coefficients; the target electrical equipment with a fault probability higher than a first probability threshold is determined as a fault hidden danger equipment, and is output to relevant personnel. The application can select appropriate types of detection sensors according to the type of the electrical equipment, and can also fuse the plurality of groups of detection data obtained by the detection according to the confidence priority coefficients of the detection sensors to obtain a more accurate fault probability of the target electrical equipment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of inspection robots, in particular to a fault hidden danger identification method based on multiple sensors and a rail-mounted inspection robot. BACKGROUND

[0002] Using a rail-mounted inspection robot to inspect electrical equipment is a new inspection technology, and the rail-mounted inspection robot can conveniently detect faults of electrical equipment below through an air track, and does not interfere with the passage of personnel and equipment below.

[0003] The existing rail-mounted inspection robot is usually configured with only one fault detection sensor, such as an infrared camera, which mainly acquires an infrared image of the electrical equipment and determines whether the electrical equipment has a fault hidden danger based on temperature. However, it has been proved in practice that relying on only one detection sensor to detect fault hidden dangers of electrical equipment cannot ensure the reliability of the fault detection result. How to further improve the accuracy of electrical equipment fault detection is a technical problem to be solved at present. SUMMARY

[0004] In order to solve the technical problems in the background art, the present application provides a fault hidden danger identification method based on multiple sensors, a rail-mounted inspection robot, an electronic device, a computer storage medium and a computer program product.

[0005] The present application provides a fault hidden danger identification method based on multiple sensors, which comprises the following steps:

[0006] The rail-mounted inspection robot shoots a high-definition image of a target electrical equipment, obtains a type attribute of the target electrical equipment based on the high-definition image, filters a plurality of detection sensors according to the type attribute, and determines a confidence priority coefficient of each detection sensor;

[0007] Each detection sensor is controlled to detect the target electrical equipment to obtain a plurality of sets of detection data, and a fault probability of the target electrical equipment is determined according to the plurality of sets of detection data and the corresponding confidence priority coefficients;

[0008] The target electrical equipment with a fault probability higher than a first probability threshold is determined as a fault hidden danger equipment, and is output to relevant personnel.

[0009] Optionally, the rail-mounted inspection robot shoots a high-definition image of a target electrical equipment, which comprises:

[0010] The track-mounted inspection robot receives an inspection scheduling task sent by a server, analyzes the inspection scheduling task to obtain each target electrical equipment that needs to be inspected, and the distribution position and inspection sequence of each target electrical equipment;

[0011] An inspection path is determined according to the distribution position and the inspection sequence, and the inspection path includes each inspection point. When the track-mounted inspection robot runs to the inspection point, a high-definition camera is controlled to capture a high-definition image of the target electrical equipment corresponding to the inspection point.

[0012] Optionally, the type attribute of the target electrical equipment is obtained based on the high-definition image, including:

[0013] A first high-definition image corresponding to the target electrical equipment and a plurality of second high-definition images corresponding to a plurality of associated electrical equipment in an electrical connection relationship with the target electrical equipment are obtained by intercepting the high-definition image;

[0014] First feature data of the target electrical equipment is obtained by intercepting the first high-definition image, and second feature data of each associated electrical equipment is obtained by intercepting the second high-definition image;

[0015] The first feature data is classified using a classification model to obtain a first type attribute of the target electrical equipment and a corresponding confidence probability;

[0016] If the confidence probability is higher than the second probability threshold, the first type attribute is determined as the type attribute of the target electrical equipment;

[0017] If the confidence probability is lower than the second probability threshold, the first feature data and the second feature data are classified again using a classification model to obtain a second type attribute of the target electrical equipment, and the second type attribute is determined as the type attribute of the target electrical equipment.

[0018] Optionally, the type attribute is used to filter a plurality of detection sensors, and a confidence priority coefficient of each detection sensor is determined, including:

[0019] A preset lookup table is analyzed according to the type attribute to filter a plurality of corresponding detection sensors;

[0020] Detection historical data of each detection sensor on the target electrical equipment is obtained, a detection accuracy rate is calculated according to the detection historical data, and the confidence priority coefficient of each detection sensor is determined according to the detection accuracy rate.

[0021] Optionally, the determining the failure probability of the target electrical equipment according to the plurality of groups of detection data and the corresponding confidence priority coefficients comprises:

[0022] performing failure analysis on each group of detection data based on a preset evaluation criterion to obtain a corresponding failure sub-probability;

[0023] weighting and fusing each failure sub-probability multiplied by the corresponding confidence priority coefficient to obtain the failure probability of the target electrical equipment.

[0024] Optionally, the target electrical equipment with the failure probability higher than the first probability threshold is determined as a failure hidden danger equipment, and output to relevant personnel, comprising:

[0025] determining the first probability threshold of the target electrical equipment according to the type attribute;

[0026] determining the target electrical equipment with the failure probability higher than the first probability threshold as a failure hidden danger equipment, and output to relevant personnel.

[0027] The application further provides a hanging rail inspection robot, the device comprising a plurality of detection sensors, a processing device and a storage device, the processing device calling and running a computer program in the storage device to realize the following steps:

[0028] The hanging rail inspection robot shoots a high-definition image of a target electrical equipment, obtains a type attribute of the target electrical equipment based on the high-definition image, screens a plurality of detection sensors according to the type attribute, and determines a confidence priority coefficient of each detection sensor;

[0029] controlling each detection sensor to detect the target electrical equipment to obtain a plurality of groups of detection data, and determining a failure probability of the target electrical equipment according to the plurality of groups of detection data and the corresponding confidence priority coefficients;

[0030] determining the target electrical equipment with the failure probability higher than the first probability threshold as a failure hidden danger equipment, and output to relevant personnel.

[0031] The application further provides an electronic device, comprising a memory storing executable program codes, a processor coupled with the memory, and the processor calling the executable program codes stored in the memory to execute the method according to any one of the above.

[0032] The application further provides a computer storage medium, the storage medium storing a computer program, and the computer program being executed by a processor to execute the method according to any one of the above.

[0033] The application further provides a computer program product, which comprises a computer program stored in a computer storage medium, and the computer program is executed by a processor of an electronic device to implement the method according to any one of the above.

[0034] The application configures various detection sensors for the track-mounted inspection robot, and can select appropriate types of detection sensors according to the type of the electrical equipment, and can also fuse a plurality of groups of detection data obtained by the detection sensors according to the confidence priority coefficients of the detection sensors to obtain a more accurate failure probability of the target electrical equipment. BRIEF DESCRIPTION OF DRAWINGS

[0035] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some of the embodiments of the application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor.

[0036] Figure 1 is a schematic diagram of a fault hidden danger identification method based on multiple sensors disclosed by the embodiments of the application.

[0037] Figure 2 is a schematic diagram of a track-mounted inspection robot disclosed by the embodiments of the application. DETAILED DESCRIPTION

[0038] In order to make the purpose, technical solutions and advantages of the embodiments of the application more clear, the following will combine the drawings in the embodiments of the application to clearly and completely describe the technical solutions in the embodiments of the application. Obviously, the described embodiments are some of the embodiments of the application, but not all the embodiments. The components of the embodiments of the application described and shown in the drawings here can be arranged and designed in various different configurations.

[0039] Please refer to Figure 1 The embodiments of the application disclose a fault hidden danger identification method based on multiple sensors, and the method comprises the following steps:

[0040] The track-mounted inspection robot shoots a high-definition image of a target electrical equipment, obtains a type attribute of the target electrical equipment based on the high-definition image, selects a plurality of detection sensors according to the type attribute, and determines a confidence priority coefficient of each detection sensor;

[0041] The detection sensors detect the target electrical equipment to obtain a plurality of groups of detection data, and determine a failure probability of the target electrical equipment according to the plurality of groups of detection data and the corresponding confidence priority coefficients;

[0042] determining the target electrical equipment with the failure probability higher than the first probability threshold as a failure hidden danger equipment, and outputting to relevant personnel.

[0043] The hanging rail inspection robot is configured with various detection sensors, including but not limited to a high-definition camera, an infrared camera, a current transformer, a voltage transformer, a temperature sensor, a reluctance current sensor (AMR, GMR and TMR sensors, etc.). The hanging rail inspection robot first controls the high-definition camera to shoot a high-definition image of the target electrical equipment, and obtains the type attribute of the target electrical equipment from the shot high-definition image, the target electrical equipment being, for example, a power transformer, a motor, an electrical control equipment, etc.; according to the type attribute, a plurality of detection sensors suitable for the target electrical equipment are screened out, and the confidence priority coefficients of the detection sensors are determined. Then, the detection sensors are controlled to detect the target electrical equipment to obtain a plurality of groups of detection data, and the failure probability of the target electrical equipment is determined according to the plurality of groups of detection data and the confidence priority coefficients determined in the foregoing. If the failure probability of the target electrical equipment is higher than a probability threshold, the target electrical equipment is determined as a failure hidden danger equipment, and is output to relevant personnel. Accordingly, the relevant personnel can remotely or on-site handle the failure hidden danger equipment in advance, thereby reducing the loss caused by equipment failure.

[0044] The hanging rail inspection robot and the basic structure of its track in the present application can completely adopt the existing technology, and the present application does not limit this.

[0045] Optionally, the hanging rail inspection robot shoots a high-definition image of the target electrical equipment, including:

[0046] The hanging rail inspection robot receives a server-sent inspection scheduling task, and analyzes the target electrical equipment to be inspected, the distribution position of each target electrical equipment and the inspection sequence of each target electrical equipment from the inspection scheduling task;

[0047] The inspection path is determined according to the distribution position and the inspection sequence, and the inspection path includes each inspection point. When the hanging rail inspection robot runs to the inspection point, the high-definition camera is controlled to shoot a high-definition image of the target electrical equipment corresponding to the inspection point.

[0048] In this embodiment, a track inspection robot docking station can be arranged in the area to be inspected, and multiple track inspection robots can be on standby in the docking station and can also be charged in the docking station. When there is an inspection requirement, the server generates a corresponding inspection scheduling task and sends it to the most suitable track inspection robot. The track inspection robot analyzes the target electrical equipment to be inspected this time from the received inspection scheduling task, as well as the distribution positions and inspection sequences of the target electrical equipment in the area to be inspected. The track inspection robot can determine the optimal inspection path according to the distribution positions and inspection sequences of the target electrical equipment. Of course, the determination of the inspection path also needs to consider the layout of the track, which will not be described in detail. When the track inspection robot runs to each inspection point, the high-definition camera can be controlled to shoot a high-definition image of the target electrical equipment corresponding to the inspection point.

[0049] Optionally, the type attribute of the target electrical equipment is obtained based on the high-definition image, including:

[0050] The first high-definition image corresponding to the target electrical equipment and the second high-definition image corresponding to the associated electrical equipment in an electrically connected relationship with the target electrical equipment are obtained by cutting the high-definition image.

[0051] The first feature data of the target electrical equipment is obtained by cutting the first high-definition image, and the second feature data of each associated electrical equipment is obtained by cutting the second high-definition image.

[0052] The first feature data is classified using a classification model to obtain a first type attribute of the target electrical equipment and a corresponding confidence probability.

[0053] If the confidence probability is higher than the second probability threshold, the first type attribute is determined as the type attribute of the target electrical equipment.

[0054] If the confidence probability is lower than the second probability threshold, the first feature data and each second feature data are classified again using a classification model to obtain a second type attribute of the target electrical equipment, and the second type attribute is determined as the type attribute of the target electrical equipment.

[0055] In this embodiment, the high-definition image shot by the high-definition camera contains the target electrical equipment and the associated electrical equipment in an electrically connected relationship with the target electrical equipment. The first high-definition image and the second high-definition image corresponding to the target electrical equipment and each associated electrical equipment are obtained by cutting the high-definition image. The first feature data of the target electrical equipment and the second feature data of each associated electrical equipment are obtained by extracting features from the first high-definition image and each second high-definition image, respectively.

[0056] Meanwhile, the application constructs two types of attribute classification models, i.e. the classification model and the classification large model. The classification model is a small model, which uses common algorithms such as SVM and Naive Bayes to classify the first feature data. The classification calculation of the small model is relatively simple and fast. The classification large model is based on a general large model, such as ChatGPT, BERT large model, KIMI, and general thousand questions. The classification large model can comprehensively analyze multiple data, especially semantic analysis, and has higher classification accuracy, but the calculation amount will increase significantly. It should be noted that the general large model needs to be lightly trained using the sorted small sample data, i.e. the general large model is fine-tuned to obtain the classification large model.

[0057] The first feature data of the target electrical equipment is first classified using the small model, i.e. the classification model, to obtain the first type attribute and the corresponding confidence probability. If the confidence probability is higher than the second probability threshold, it means that the characteristics of the target electrical equipment are obvious, and the credibility of the classification result of the classification model is high enough. At this time, the first type attribute can be directly determined as the type attribute of the target electrical equipment. If the confidence probability is lower than the second probability threshold, it means that the characteristics of the target electrical equipment are not obvious, and the credibility of the classification result of the classification model is not high enough. At this time, the classification large model is used to perform secondary classification processing on the first feature data of the target electrical equipment and the second feature data of each associated electrical equipment to obtain the second type attribute. Since the second type attribute is determined by the classification large model based on the first feature data of the target electrical equipment and the second feature data of each associated electrical equipment, its credibility will be higher than the first type attribute obtained by the classification model. Therefore, the second type attribute can be determined as the type attribute of the target electrical equipment.

[0058] Therefore, the application determines the confidence probability obtained by analyzing the small model, i.e. the classification model, to decide whether to further use the large model, i.e. the classification large model, for secondary classification, so as to significantly improve the overall classification efficiency of the type attribute of the target electrical equipment.

[0059] It should be noted that the second probability threshold is preferably a dynamic value. Specifically, the number of electrical equipment around the target electrical equipment is first detected based on the first high-definition image. When there are more electrical equipment around the target electrical equipment, the area of the target electrical equipment is more likely to be blocked, and the first feature data obtained by the target electrical equipment is less likely to be obtained. At this time, the probability of the confidence probability being high is larger, and correspondingly, the second probability threshold is larger. On the contrary, when there are fewer electrical equipment around the target electrical equipment, the second probability threshold is smaller.

[0060] Optionally, the type attribute is used to filter out a plurality of detection sensors, and a confidence priority coefficient of each detection sensor is determined, comprising:

[0061] A preset lookup table is analyzed according to the type attribute, and a plurality of corresponding detection sensors are filtered out;

[0062] Detection historical data of each detection sensor on the target electrical equipment is obtained, a detection accuracy is calculated according to the detection historical data, and the confidence priority coefficient of each detection sensor is determined according to the detection accuracy.

[0063] In this embodiment, different types of electrical equipment have different fault types, and different fault types can be detected and identified by corresponding several types of detection sensors. For example:

[0064] Power transformer: fault types include insulation aging, partial discharge, overheating, etc. Detection methods include oil gas chromatography analysis (DGA), furfural content (HPLC), polarization current (PDC), recovery voltage (RVM), etc. diagnostic rules.

[0065] Distribution network and industrial system: fault types include electrical abnormalities such as partial discharge, breakdown, etc. Monitoring technology involves current sensors and voltage sensors.

[0066] Internal electrical equipment: fault types may include temperature abnormally high due to conduction loop failure, overload, etc. It can be detected by infrared detection technology.

[0067] Electrical control equipment: fault types may include signal display abnormalities, button, contactor, thermal relay and fuse conditions, etc. Image recognition technology can be used for detection.

[0068] Therefore, a lookup table is constructed in advance, which includes a plurality of detection sensors corresponding to different types of electrical equipment. After the type attribute of the target electrical equipment is determined, the lookup table can be used to determine a plurality of detection sensors.

[0069] Meanwhile, detection historical data of each detection sensor on the target electrical equipment is obtained. The detection historical data can be recorded during actual inspection or during the test phase. The detection historical data includes a plurality of successful records and failure records of using corresponding types of detection sensors to detect faults of the target electrical equipment. According to this, a detection accuracy can be calculated, and a confidence priority coefficient of each detection sensor is determined according to the detection accuracy.

[0070] Obviously, the higher the detection accuracy, the higher the corresponding confidence priority coefficient.

[0071] Optionally, the determining the failure probability of the target electrical equipment according to the plurality of groups of detection data and the corresponding confidence priority coefficients comprises:

[0072] performing failure analysis on each group of detection data based on a preset evaluation criterion to obtain a corresponding failure sub-probability;

[0073] weighting and fusing the failure sub-probabilities after being multiplied by the corresponding confidence priority coefficients to obtain the failure probability of the target electrical equipment.

[0074] In this embodiment, after obtaining the detection data of the target electrical equipment transmitted by each detection sensor, failure analysis can be performed on each group of detection data according to a preset evaluation criterion, so as to obtain a failure sub-probability based on the detection result of the detection sensor. The preset evaluation criterion includes a determination criterion set for various failure types of the target electrical equipment, for example, a temperature exceeding 70°C is determined as a high failure probability, and a temperature not exceeding 35°C is determined as a low failure probability. Finally, weighting and fusing calculation are performed after each failure sub-probability is multiplied by the corresponding confidence priority coefficient, and the failure probability of the target electrical equipment is obtained.

[0075] Optionally, the target electrical equipment with a failure probability higher than a first probability threshold is determined as a failure hidden danger equipment, and output to relevant personnel, comprising:

[0076] determining the first probability threshold of the corresponding target electrical equipment according to the type attribute;

[0077] determining the target electrical equipment with a failure probability higher than a first probability threshold as a failure hidden danger equipment, and output to relevant personnel.

[0078] In this embodiment, the first probability threshold of different types of target electrical equipment in the present application is different, and the first probability threshold can be preset. Specifically, the importance of the target electrical equipment in the entire electrical network in the region and the self-adjustment / self-correction capability of the target electrical equipment are comprehensively evaluated, and a higher first probability threshold is set for those target electrical equipment with low importance and / or high self-adjustment / self-correction capability, and a lower first probability threshold is set for those target electrical equipment with high importance and / or low self-adjustment / self-correction capability. In this way, unnecessary early warning can be reduced.

[0079] Referring to Figure 2 the embodiment of the present application also provides a hanging rail inspection robot, the device comprises a plurality of detection sensors, a processing device and a storage device, the processing device calls and runs a computer program in the storage device to realize the following steps:

[0080] The hanging rail inspection robot shoots a high-definition image of the target electrical equipment, obtains a type attribute of the target electrical equipment based on the high-definition image, filters a plurality of detection sensors according to the type attribute, and determines a confidence priority coefficient of each detection sensor;

[0081] The detection sensors control each detection sensor to detect the target electrical equipment to obtain a plurality of sets of detection data, and determine a fault probability of the target electrical equipment according to the plurality of sets of detection data and the corresponding confidence priority coefficients.

[0082] The target electrical equipment with the fault probability higher than a first probability threshold is determined as a fault hidden danger equipment, and is output to relevant personnel.

[0083] The embodiment of the present application also provides an electronic device, including: a memory storing executable program codes; a processor coupled with the memory; the processor invokes the executable program codes stored in the memory to execute the method according to any one of the above embodiments.

[0084] The embodiment of the present application also provides a computer storage medium, which stores a computer program, and the computer program is executed by a processor to execute the method according to any one of the above embodiments.

[0085] The embodiment of the present application also provides a computer program product, which contains a computer program stored in a computer storage medium, and the computer program is executed by a processor of an electronic device to implement the method according to any one of the above embodiments.

[0086] The present application is described with reference to flowcharts and / or block diagrams of the method, device (apparatus) and computer program product according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing devices to produce a machine, so that the instructions executed by the computer or other programmable data processing devices produce a device implemented in the flowcharts and / or block diagrams. Figure 1 The device for implementing the function specified in one flow or multiple flows and / or blocks Figure 1 The device for implementing the function specified in one flow or multiple flows and / or blocks

[0087] These computer program instructions can also be stored in a computer readable storage medium, which can guide the computer or other programmable data processing devices to work in a specific way, so that the instructions stored in the computer readable storage medium produce a product including instruction devices, which implement the flowcharts and / or block diagrams. Figure 1 The device for implementing the function specified in one flow or multiple flows and / or blocksFigure 1 the function specified in the one or more blocks.

[0088] These computer program instructions can also be loaded into computer or other programmable data processing devices, so that a series of operation steps are performed on the computer or other programmable data processing devices to generate computer-implemented processes, thus the instructions executed on the computer or other programmable data processing devices provide processes for implementing the flows Figure 1 the flows or the plurality of flows and / or blocks Figure 1 the steps of the function specified in the one or more blocks.

[0089] The above description is merely one specific implementation of the application. However, one of ordinary skill in the art should, in light of the above description, appreciate changes and modifications made in the scope of the technology disclosed by the present application. Therefore, the protective scope of the present application should be subject to the protection scope of the claims.

Claims

1. A multi-sensor based fault potential identification method, characterized by: The method comprises the following steps: The rail inspection robot captures a high-definition image of the target electrical equipment, extracts a type attribute of the target electrical equipment based on the high-definition image, selects a number of detection sensors based on the type attribute, and determines a confidence priority coefficient for each of the detection sensors; Controlling each of the detection sensors to detect the target electrical equipment to obtain multiple sets of detection data, and determining a failure probability of the target electrical equipment based on the multiple sets of detection data and the corresponding confidence priority coefficients; Determine the target electrical equipment whose failure probability is higher than a first probability threshold as a potential fault device, and output this information to relevant personnel; The method of extracting the type attribute of the target electrical equipment based on the high-definition image includes: Extracting from the high-definition image a first high-definition image corresponding to the target electrical device and second high-definition images corresponding to a plurality of associated electrical devices electrically connected to the target electrical device; Extracting first feature data of the target electrical device from the first high-definition image, and extracting second feature data of each of the associated electrical devices from the second high-definition image; Using a classification model to perform classification processing on the first feature data to obtain a first type attribute of the target electrical equipment and a corresponding confidence probability; If the confidence probability is higher than a second probability threshold, determining the first type attribute as the type attribute of the target electrical device; If the confidence probability is lower than a second probability threshold, a classification model is used to perform secondary classification processing on the first feature data and each of the second feature data to obtain a second type attribute of the target electrical equipment, and the second type attribute is determined as the type attribute of the target electrical equipment.

2. The multi-sensor based fault potential identification method according to claim 1, characterized in that: The rail inspection robot captures high-definition images of target electrical equipment, including: The rail-mounted inspection robot receives the inspection scheduling task sent by the server, and obtains the target electrical equipment that needs to be inspected, as well as the distribution location and inspection order of each target electrical equipment from the inspection scheduling task; An inspection path is determined according to the distribution position and the inspection order, and the inspection path includes various inspection points. When running to the inspection point, the high-definition camera is controlled to capture a high-definition image of the target electrical equipment corresponding to the inspection point.

3. The multi-sensor based fault potential identification method according to claim 1, characterized in that: A plurality of detection sensors are obtained by screening according to the type attributes, and a confidence priority coefficient of each of the detection sensors is determined, including: According to the type attribute, a preset comparison table is analyzed to screen out the corresponding detection sensors; Acquire detection history data of each detection sensor on the target electrical device, calculate the detection accuracy rate based on the detection history data, and determine the confidence priority coefficient of each detection sensor based on the detection accuracy rate.

4. The multi-sensor based fault potential identification method according to claim 3, characterized in that: Determining the failure probability of the target electrical equipment according to the multiple sets of detection data and the corresponding confidence priority coefficients includes: Performing fault analysis on each set of detection data based on preset evaluation criteria to obtain corresponding fault sub-probabilities; Each fault sub-probability is multiplied by the corresponding confidence priority coefficient and then weighted fusion is performed to obtain the fault probability of the target electrical equipment.

5. The multi-sensor based fault potential identification method according to any one of claims 1 to 4, characterized in that: Determining the target electrical equipment whose failure probability is higher than a first probability threshold as a potential fault device and outputting the information to relevant personnel includes: Determining the first probability threshold of the corresponding target electrical device according to the type attribute; The target electrical equipment whose failure probability is higher than a first probability threshold is determined as a potential failure equipment, and a message is output to relevant personnel.

6. A rail inspection robot, comprising a plurality of detection sensors, a processing device, and a storage device, characterized in that: The processing device calls and runs the computer program in the storage device to implement the following steps: The rail inspection robot captures a high-definition image of the target electrical equipment, extracts a type attribute of the target electrical equipment based on the high-definition image, selects a number of detection sensors based on the type attribute, and determines a confidence priority coefficient for each of the detection sensors; Controlling each of the detection sensors to detect the target electrical equipment to obtain multiple sets of detection data, and determining a failure probability of the target electrical equipment based on the multiple sets of detection data and the corresponding confidence priority coefficients; Determine the target electrical equipment whose failure probability is higher than a first probability threshold as a potential fault device, and output this information to relevant personnel; The method of extracting the type attribute of the target electrical equipment based on the high-definition image includes: Extracting from the high-definition image a first high-definition image corresponding to the target electrical device and second high-definition images corresponding to a plurality of associated electrical devices electrically connected to the target electrical device; Extracting first feature data of the target electrical device from the first high-definition image, and extracting second feature data of each of the associated electrical devices from the second high-definition image; Using a classification model to perform classification processing on the first feature data to obtain a first type attribute of the target electrical equipment and a corresponding confidence probability; If the confidence probability is higher than a second probability threshold, determining the first type attribute as the type attribute of the target electrical device; If the confidence probability is lower than a second probability threshold, a classification model is used to perform secondary classification processing on the first feature data and each of the second feature data to obtain a second type attribute of the target electrical equipment, and the second type attribute is determined as the type attribute of the target electrical equipment.

7. An electronic device comprising: a memory storing executable program code; A processor coupled to the memory; characterized in that: the processor calls the executable program code stored in the memory to execute the method according to any one of claims 1-5.

8. A computer storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 5 is executed.

9. A computer program product comprising a computer program stored in a computer storage medium, characterized in that: When the computer program is executed by a processor of an electronic device, the method according to any one of claims 1 to 5 is implemented.

Images