Electrical equipment discharge hidden danger identification method and related device

Abstract

The invention provides an electrical equipment discharge hidden danger identification method and a related device. The electrical equipment discharge hidden danger identification method comprises the following steps: acquiring a fire sample image and fault information of target equipment; constructing a deep neural network model of the target equipment, and training the constructed deep neural network model based on the historical fire sample image of the target equipment; and detecting the current fire sample image based on the trained deep neural network model of the target equipment, and determining the position of the fire hazard of the target equipment and a fire hazard identification result. According to the invention, the deep neural network model of the constructed target equipment is trained through the historical fire sample image of the target equipment, so that the position of the fire hazard and the fire hazard identification result are determined, and local overheating in the electrical equipment and spark discharge at a hidden position can be found as early as possible.

Description

Technical Field

[0001] The present disclosure relates to the technical field of fire prevention and control, and in particular to a method for identifying discharge hazards of electric power equipment and a related device. Background Art

[0002] Fires caused by improper use and management of electrical equipment occur from time to time. In the operation of the power system, high-voltage electrical equipment plays a vital role. Substation construction is a complex management process involving many professions. It is inevitable to be affected by various uncertain factors and bring risks. It requires strict scientific management. If the transformer has partial discharge, it is very likely to cause premature damage to the transformer, affecting the service life of the transformer. At the same time, partial discharge also directly affects the normal power supply of the region. Inside the substation, there are a large number of cables, switches, transformers and other equipment, which generate heat when working. Therefore, if the equipment fails or is damaged, it may cause safety accidents such as fire or explosion. Therefore, the detection of partial discharge of the transformer has become an important measure to ensure the safe and reliable operation of the equipment. Partial discharge has now become one of the important causes of transformer failure, the hazards caused by partial discharge and the main discharge forms. When the high-voltage equipment discharges electricity, corona, flashover or arc will be generated according to the different electric field strengths. In the target detection task, traditional methods mostly use manually designed feature algorithms to classify images. However, the target environment has problems such as uneven lighting, complex background and object occlusion, which makes it difficult for manual features to stably describe targets under different conditions, resulting in detection or missed detection. In addition, traditional methods cannot detect hidden dangers such as local overheating and hidden spark discharges in electrical systems, so they have problems such as poor robustness and long time consumption. Summary of the invention

[0003] The present disclosure aims to at least solve the above-mentioned technical problems existing in the prior art.

[0004] To this end, an object of the present disclosure is to provide a method for identifying a potential discharge hazard of an electric device, the method comprising:

[0005] Obtain fire sample images and fault information of target equipment;

[0006] Constructing a deep neural network model of a target device, based on the deep neural network model constructed by training historical fire sample images of the target device;

[0007] Based on the trained deep neural network model of the target device, the current fire sample image is detected to determine the location of the fire hazard of the target device and the fire hazard identification result.

[0008] Furthermore, the obtaining of fire sample images and fault information of the target equipment includes:

[0009] Acquire a real-time environment image and fault information including the target device;

[0010] Acquire arc fault images of target equipment where fire occurs;

[0011] A fire sample image is determined based on the real-time environment image and the fault arc image.

[0012] Furthermore, the construction of the deep neural network model of the target device specifically includes:

[0013] Build a dataset of target equipment based on historical fire sample images and fault information;

[0014] Based on the basic deep neural network model, a deep neural network model of the target device that matches the data set of the target device is obtained.

[0015] Furthermore, the deep neural network model constructed by training based on the historical fire sample images of the target device includes:

[0016] Preprocessing the historical fire sample images of the target equipment;

[0017] A deep neural network model is constructed based on the training of preprocessed historical fire sample images of the target equipment.

[0018] Furthermore, the deep neural network model based on the trained target device detects the current fire sample image to determine the location of the fire hazards of the target device and the fire hazard identification results, which at least include the actual defects of the target device, the hidden defects of the target device and the abnormal heating conditions of the target device.

[0019] Furthermore, the determining the location of the fire hazard of the target device and the fire hazard identification result includes:

[0020] Determine the location of fire hazards based on flame color, flame shape, and flame texture.

[0021] Furthermore, the determining the location of the fire hazard of the target device and the fire hazard identification result also includes:

[0022] The location of the fire hazard is determined based on the flame flickering frequency, flame area growth, flame shape changes and flame movement direction.

[0023] One aspect of the present disclosure provides a device for identifying discharge hazards of electric equipment, comprising:

[0024] An acquisition module, used to acquire fire sample images and fault information of target equipment;

[0025] A construction module, used to construct a deep neural network model of a target device, based on the deep neural network model constructed by training historical fire sample images of the target device;

[0026] The diagnosis module is used to detect the current fire sample image based on the trained deep neural network model of the target device, determine the location of the fire hazard of the target device and the fire hazard identification result.

[0027] One aspect of the present disclosure provides a storage medium storing a computer program, wherein the computer program implements the steps of the above method when executed by a processor.

[0028] One aspect of the present disclosure provides an electronic device, comprising at least a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps of the above method when executing the computer program in the memory.

[0029] The present disclosure provides a method and a related device for identifying discharge hazards of electric equipment, which have the following beneficial effects:

[0030] By acquiring fire sample images and fault information of the target device; constructing a deep neural network model of the target device, and training the constructed deep neural network model based on the historical fire sample images of the target device; and detecting the current fire sample image based on the trained deep neural network model of the target device, the location of the fire hazard of the target device and the fire hazard identification result are determined. The deep neural network model is trained based on the historical fire sample images of the target device, and then the location of the fire hazard and the fire hazard identification result are determined, which is conducive to the early detection of local overheating in electrical equipment and spark discharge in hidden locations. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0032] Figure 1 is a flow chart of a method for identifying discharge hazards of electric power equipment in an embodiment of the present disclosure;

[0033] Figure 2 is a flow chart of a method for identifying discharge hazards of electric power equipment in an embodiment of the present disclosure;

[0034] Figure 3 It is a flow chart of a method for identifying discharge hazards of electric power equipment in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0035] Various aspects and features of the present disclosure are described herein with reference to the drawings.

[0036] It should be understood that various modifications may be made to the embodiments of the present application. Therefore, the above description should not be considered as limiting, but only as an example of an embodiment. Other modifications within the scope and spirit of the present disclosure will occur to those skilled in the art.

[0037] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments of the present disclosure and, together with the general description of the present disclosure given above and the detailed description of the embodiments given below, serve to explain the principles of the present disclosure.

[0038] These and other characteristics of the present disclosure will become apparent from the following description of a preferred form of embodiment given as a non-limiting example, with reference to the attached drawings.

[0039] It should also be understood that, although the present disclosure has been described with reference to some specific examples, those skilled in the art will be able to realize many other equivalent forms of the present disclosure that have the features described in the claims and are therefore within the scope of protection defined thereby.

[0040] The above and other aspects, features and advantages of the present disclosure will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings.

[0041] Specific embodiments of the present disclosure are described hereinafter with reference to the accompanying drawings; however, it should be understood that the embodiments claimed are merely examples of the present disclosure, which may be implemented in a variety of ways. Well-known and / or repeated functions and structures are not described in detail to avoid obscuring the present disclosure with unnecessary or redundant details. Therefore, the specific structural and functional details claimed herein are not intended to be limiting, but merely serve as a basis and representative basis for the claims to teach those skilled in the art to use the present disclosure in a variety of ways with substantially any suitable detailed structure.

[0042] The first embodiment of the present disclosure provides a method for identifying discharge hazards of electric equipment, such as Figure 1-Figure 3 As shown, the method for identifying discharge hazards of electric equipment includes:

[0043] S1: Obtain fire sample images and fault information of target equipment;

[0044] Specifically, the target equipment includes electrical equipment such as transformers, conductors, insulators, mutual inductors, lightning arresters, switchgear, high-voltage insulators, disconnectors and circuit breakers, or at least one of capacitors, bushings, surge arresters, cables, reactors and relay protection devices.

[0045] Specifically, obtaining a fire sample image of a target device includes:

[0046] S101: Acquire a real-time environment image and fault information including a target device;

[0047] Specifically, by monitoring the target device in real time through an image acquisition device, an environmental image including the target device and its surrounding environment can be acquired.

[0048] Fault information refers to electrical contact heating, arc discharge, overheating of electrical equipment, short circuit, poor contact, leakage, overload, improper use and failure of protection device.

[0049] S102: Acquire a fault arc image of a target device where a fire occurs;

[0050] Fault arc refers to the phenomenon of gas ionization discharge caused by air breakdown due to insulation aging and damage in electrical lines or equipment, loose electrical connections, humid air, and sharp increase in voltage and current. Through multi-circuit fault monitoring of the fault arc detector, the fault arc image of the target equipment where the fire occurred can be obtained.

[0051] S103: Determine a fire sample image based on the real-time environment image and the fault arc image.

[0052] The fire sample image refers to the image data of the target device where the fire occurs, which is obtained by preprocessing the real-time environment image of the target device where the fire occurs (including the target device of the fault arc and the background environment image around the target image).

[0053] The present disclosure obtains multiple real-time environment images containing the target device, and uses a fault arc detector to obtain a fault arc image of the target device where a fire occurs, thereby obtaining a fire sample image of the target device. In the real-time environment image, the image with the fire source is screened out, which is conducive to using the fire sample image to provide a high-accuracy sample for subsequent deep learning, thereby improving the accuracy of deep learning.

[0054] S2: Build a deep neural network model for the target device, based on the historical fire sample images of the target device;

[0055] Building a deep neural network model for the target device specifically includes:

[0056] S201: construct a data set of target equipment based on historical fire sample images and fault information;

[0057] Specifically, the training data set uses part or all of the fire sample images. Feature extraction is performed on the fire sample images to obtain different feature vectors. The category labels of the samples are annotated according to the feature vectors obtained from the fault information records to determine whether a fault has occurred. The obtained feature vector set is normalized and zero-eliminating to obtain the final training data set.

[0058] In feature selection and extraction, redundant features and irrelevant features are removed from the original feature set, and features that play a decisive role in distinguishing data features are retained. Geometric median filter pruning (FPGM) is used to further remove redundant filters from each convolutional layer, and parameters with low importance or strong substitutability in the neural network are discarded to reduce model redundancy. A convolution is performed on the image to extract features, and the feature map is normalized to the same size. The present invention removes redundant features and irrelevant features from the original feature set, retains features that play a decisive role in distinguishing data features, and geometric median filter pruning (FPGM) is used to further remove redundant filters from each convolutional layer, and parameters with low importance or strong substitutability in the neural network are discarded to reduce model redundancy. A convolution is performed on the image to extract features, and the feature map is normalized to the same size. By automatically pruning redundant filters in the convolutional layer, the importance of each Block in MobileNetV2 to the discharge diagnosis task can be autonomously learned under the drive of PRPD data and the redundant parts can be trimmed, thereby achieving structural simplification in the depth direction. In addition, the FPGM-ESA-based automatic pruning method in the breadth direction can adaptively determine the pruning ratio of each layer according to the parameter redundancy of different convolutional layers of the model, and determine the optimal pruning scale through cyclic search. After pruning, the discharge recognition rate is basically the same as that of MobileNetV2. Compared with pruning methods such as pre-training pruning, Slimming, FPGM and AMC, this method has better recoverability and improves the recognition rate by 0.30% to 4.64%.

[0059] S202: Based on the basic deep neural network model, a deep neural network model of the target device that matches the data set of the target device is obtained.

[0060] Taking the insulated enclosed metal switchgear as the target device, partial discharge diagnosis of the target device is explained as an example. After constructing the data set of the insulated enclosed metal switchgear, the basic deep neural network model (DNN) is introduced into the partial discharge diagnosis of the insulated enclosed metal switchgear, and a lightweight algorithm model matching the partial discharge data set is obtained. Specifically, the redundant modules in the model are perceived and pruned end-to-end, and the input feature map (fire sample image after feature extraction) is compressed by point-by-point convolution using the residual block. Then, the features are extracted through deep convolution, and an enhanced random search algorithm is used for cyclic search until a maximum lossless compression model is obtained. The pruning ratio of each layer is autonomously solved to achieve structural compression in the depth direction.

[0061] The present disclosure takes into account the rapid response and easy deployment inside the terminal substation, and adopts a flame recognition algorithm based on large-scale fire sample data. An iterative importance factor is introduced into the fault training model, and the deep neural network (DNN) model is introduced as the basic model into the partial discharge diagnosis of the insulated closed metal switchgear to obtain a lightweight algorithm model that matches the data set, eliminating the adverse effects of the classification network's insensitivity to position. The model scale is reduced to a certain extent, thereby obtaining a lightweight model that matches the partial discharge data set. The detection and recognition speed has been significantly improved, the detection and recognition speed and accuracy have been improved, the model's ability to detect small targets has been enhanced, and the problems of large space occupation and slow detection in the RCNN algorithm have been avoided. The results show that this method uses multiple convolutions to replace large-scale convolution kernels, which plays a role in reducing the amount of calculation without significantly reducing the accuracy. Therefore, the space occupation of the model can be independently designed under data-driven high-precision, lightweight, and low-latency partial discharge deep diagnosis models. Compared with the existing deep models, the resource occupation is greatly reduced.

[0062] This paper adopts an enhanced random search algorithm to search cyclically until the maximum lossless compression model is obtained, and autonomously solves the pruning ratio of each layer to achieve structural compression in the depth direction; compared with AMC pruning, the parameter compression rate is higher, resource occupation is greatly reduced, and storage and memory consumption are only 0.88MB and 12.91MB. Through fault model training, fault feature identification, fault evolution law, and fault prediction technology based on deep learning, the equipment fault prediction capability can be improved.

[0063] The deep neural network model constructed based on the fire sample image training of the target device includes:

[0064] S203: preprocessing the historical fire sample images of the target equipment;

[0065] Preprocessing refers to identifying the source of danger of the target equipment based on AI recognition technology, automatically identifying fire safety hazards, and inputting the preprocessed image into a lightweight deep convolutional neural network to classify the target. Compared with deep learning models for classification, fewer samples are required, with strong controllability and high accuracy. Solve the problems of high resource usage and reasoning lag of existing partial discharge deep diagnosis models. Convert the fault data in the power equipment into data that can be recognized by the learning algorithm; compared with the partial discharge diagnosis method based on VGG, ResNet, ShuffleNet and GhostNet, it shows similar discharge recognition capabilities under different signal-to-noise ratios. Preprocessing the fire sample image is conducive to simultaneously identifying the target position and locating the defect type, reducing the degree of manual participation, and improving the efficiency and accuracy of defect recognition.

[0066] S204: A deep neural network model is constructed based on training of the preprocessed historical fire sample images of the target equipment.

[0067] After the preprocessed fire sample image features of the target device are extracted, the feature vector is obtained. The deep convolutional neural network is used to learn historical fault samples, and the connection weights of each layer in the neural network model are continuously adjusted. The feature vector of the defect to be identified (corresponding to the re-acquired fire sample image) is similar to the feature vector in the feature matching library, and the feature vector in the feature matching library that is most similar to the feature vector of the defect to be identified is obtained. Finally, the terminal CPU processes it through the connected FPGA used to judge the fire situation. After the FPGA processing is completed, the judgment result is returned, and the trained neural network model is tested to obtain the diagnostic recognition result. It is applied to electrical fire fault identification. Simulation experiments show that it has faster learning speed and higher convergence success rate than traditional BP neural networks, which improves the accuracy of electrical fire fault identification and has obvious advantages.

[0068] The deep neural network CNN model is based on the fault theory model of deep learning. According to the weights of all connections of multi-layer perceptrons in multiple hidden layers, the deep learning algorithm is used to calculate the hierarchical feature observation data composed of high-level feature learning obtained from the lower layers, and the target features are learned hierarchically. Automatically learning features in the multi-level abstract process can enable the machine learning system to learn from the raw data to the complex function mapping from input to output. It solves the arduous and difficult process of manually extracting features.

[0069] S3: Detect the current fire sample image based on the trained deep neural network model of the target device to determine the location of the fire hazard of the target device and the fire hazard identification result.

[0070] Specifically include:

[0071] S301: Extract features from the current fire sample image;

[0072] S302: The fire sample image after feature extraction is input into the constructed deep neural network model of the target device;

[0073] S303: Based on the dynamic prediction and static prediction of the system fault, determine the location of the target device and the defect identification result.

[0074] The fire hazard identification result includes actual defects of the target device, hidden defects of the target device and abnormal heating conditions of the target device.

[0075] Static prediction refers to determining the location of fire hazards based on flame color, flame shape, and flame texture.

[0076] Dynamic prediction refers to determining the location of fire hazards based on the flickering frequency of the flame, the area growth of the flame, the change in the shape of the flame and the direction of the flame's movement.

[0077] Based on the dynamic prediction and static prediction of system faults, the association rules of multiple identified fault features are analyzed to form the time evolution law of single fault features and the propagation evolution law model of multiple features, to explore the change law of performance parameters or fault features, to infer the change trend of fault features, to estimate the propagation, development and change trend of power equipment performance of faults, to conduct fault trend prediction and evolution analysis to build training data sets, so as to achieve the role of fault prevention.

[0078] According to the attribute value of each feature vector and the prediction rules generated by the deep neural network model of the constructed target device, the sample category is labeled for the sample to be predicted. The attribute value refers to the time, location, casualties, fire type, fire level, and cause of the fire. The attribute value represents the characteristic attribute of the case, so as to obtain the fire case with the highest similarity to the target fire case. The solution of this case can quickly guide the target case to solve the current problem. It can preliminarily determine the location and size of the fire, and determine whether the fire isolation measures involved can control the fire within the set fire protection area.

[0079] In the normal state where no fire is detected, the FPGA detects the abnormal state of fire with single-core processing, quickly reconstructs and expands multi-core parallel processing, and quickly completes the full-site scanning detection FPGA. After amplification and analog-to-digital A / D conversion, the CPU analyzes and distinguishes the conversion intensity and compares it with the alarm preset value. When it exceeds the preset value, an alarm signal is issued and transmitted to the monitoring device, which is then further identified and resolved by the monitoring device. For example, when the smoke concentration reaches the set threshold, such as when the fire smoke reaches 5% OBS / M-15% OBS / M, an alarm will be issued.

[0080] For example, the lightweight deep neural network model computing software connects the input layer with the intermediate layer with a weight w i , the connection weight between the hidden layer and the output layer is w h , the hidden layer neuron threshold θ and the output layer neuron threshold γ are assigned a random number between (-1,1), given the calculation accuracy value ε and the maximum number of learning times M, calculate the input u of each neuron in the hidden layer h (k), and then use the activation function to calculate the hidden layer output v h (k), find the output layer input y. (k) and output y. (k); use the output layer neuron δ. (k) and the hidden layer neuron output v h (k) Modify the connection weight w ho And threshold γ, then calculate the global error E, and determine whether E is full. If E < ε or the number of learning times exceeds the set maximum number M, the algorithm ends, otherwise the next learning sample and the corresponding expected output are randomly selected.

[0081] Based on the same inventive concept as the first embodiment above, the second embodiment of the present disclosure provides a device for identifying discharge hazards of electric equipment, comprising:

[0082] An acquisition module, used to acquire fire sample images and fault information of target equipment;

[0083] A construction module is used to construct a deep neural network model of a target device, and the deep neural network model is constructed based on historical fire sample images of the target device;

[0084] The diagnosis module is used to detect the current fire sample image based on the trained deep neural network model of the target device, determine the location of the fire hazard of the target device and the fire hazard identification result.

[0085] Furthermore, the acquisition module includes:

[0086] A first image acquisition module, used to acquire a real-time environment image and fault information including a target device;

[0087] A second image acquisition module is used to acquire a fault arc image of a target device where a fire occurs;

[0088] The sample image determination module is used to determine the fire sample image based on the real-time environment image and the fault arc image.

[0089] Furthermore, the building blocks include:

[0090] A dataset construction module is used to construct a dataset of target equipment based on historical fire sample images and fault information;

[0091] A model building module, used for obtaining a deep neural network model of a target device that matches a data set of the target device based on a basic deep neural network model;

[0092] A preprocessing module, used for preprocessing fire sample images of historical target equipment;

[0093] A training module is used to train a deep neural network model based on preprocessed fire sample images of historical target equipment.

[0094] The fire hazard identification result includes at least actual defects of the target equipment, hidden defects of the target equipment and abnormal heating conditions of the target equipment.

[0095] Determine the location of fire hazards in the target equipment and the fire hazard identification results, including:

[0096] Determine the location of fire hazards based on flame color, flame shape, and flame texture.

[0097] The location of the fire hazard is determined based on the flame flickering frequency, flame area growth, flame shape changes and flame movement direction.

[0098] A third embodiment of the present disclosure provides a storage medium storing a computer program. When the computer program is executed by a processor, the steps of implementing the above method include:

[0099] S11: Acquire fire sample images and fault information of target equipment;

[0100] S12: construct a deep neural network model of the target device, based on the historical fire sample images of the target device.

[0101] S13: Detect the current fire sample image based on the trained deep neural network model of the target device to determine the location of the fire hazard of the target device and the fire hazard identification result.

[0102] Of course, other steps of the control method can also be used to implement the above embodiment.

[0103] In this embodiment, the deep neural network model is trained by using historical fire sample images of the target equipment to determine the location of the fire hazard and the fire hazard identification results, which is conducive to early detection of local overheating in electrical equipment and spark discharge phenomena in hidden locations.

[0104] A fourth embodiment of the present disclosure provides an electronic device, the electronic device comprising at least a memory and a processor, the memory storing a computer program, and the processor implementing the steps of the above method when executing the computer program on the memory, specifically including:

[0105] S21: Acquire fire sample images and fault information of target equipment;

[0106] S22: construct a deep neural network model of the target device, based on the historical fire sample images of the target device for training and construction;

[0107] S23: Detect the current fire sample image based on the trained lightweight deep neural network model of the target device to determine the location of the fire hazard of the target device and the fire hazard identification result.

[0108] In this embodiment, a deep neural network model is trained based on historical fire sample images of the target equipment to determine the location of fire hazards and the fire hazard identification results, which is conducive to early detection of local overheating in electrical equipment and spark discharge phenomena in hidden locations.

[0109] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0110] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0111] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0112] In the embodiments provided in the present application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0113] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0114] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0115] If the integrated module is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned motor torque control method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0116] In addition, the features of the embodiments shown in the drawings of the present application or the various embodiments mentioned in this specification are not necessarily understood as independent embodiments. Instead, each feature described in one example of an embodiment can be combined with one or more other desired features from other embodiments to produce other embodiments not described in words or with reference to the drawings.

[0117] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for identifying discharge hazards of electric power equipment, characterized in that: include: Obtain fire sample images and fault information of target equipment; Constructing a deep neural network model of a target device, based on the deep neural network model constructed by training historical fire sample images of the target device; Based on the trained deep neural network model of the target device, the current fire sample image is detected to determine the location of the fire hazard of the target device and the fire hazard identification result.

2. The method for identifying discharge hazards of electric power equipment according to claim 1, characterized in that: The obtaining of fire sample images and fault information of target equipment includes: Acquire a real-time environment image and fault information including the target device; Acquire arc fault images of target equipment where fire occurs; A fire sample image is determined based on the real-time environment image and the fault arc image.

3. The method for identifying discharge hazards of electric power equipment according to claim 2, characterized in that: The construction of the deep neural network model of the target device specifically includes: Build a dataset of target equipment based on historical fire sample images and fault information; Based on the basic deep neural network model, a deep neural network model of the target device that matches the data set of the target device is obtained.

4. The method for identifying discharge hazards of electric power equipment according to claim 1, characterized in that: The deep neural network model constructed based on the historical fire sample image training of the target device includes: Preprocessing the historical fire sample images of the target equipment; A deep neural network model is constructed based on the training of preprocessed historical fire sample images of the target equipment.

5. The method for identifying discharge hazards of electric power equipment according to claim 1, characterized in that: The deep neural network model based on the trained target device detects the current fire sample image to determine the location of the fire hazard of the target device and the fire hazard identification result, wherein the fire hazard identification result at least includes the actual defects of the target device, the hidden defects of the target device and the abnormal heating conditions of the target device.

6. The method for identifying discharge hazards of electric power equipment according to claim 1, characterized in that: The determining of the location of the fire hazard of the target device and the fire hazard identification result includes: Determine the location of fire hazards based on flame color, flame shape, and flame texture.

7. The method for identifying discharge hazards of electric power equipment according to claim 6, characterized in that: The determining of the location of the fire hazard of the target device and the fire hazard identification result also includes: The location of the fire hazard is determined based on the flame flickering frequency, flame area growth, flame shape changes and flame movement direction.

8. A device for identifying potential discharge hazards of electric power equipment, characterized in that: include: An acquisition module, used to acquire fire sample images and fault information of target equipment; A construction module, used to construct a deep neural network model of a target device, based on the deep neural network model constructed by training historical fire sample images of the target device; The diagnosis module is used to detect the current fire sample image based on the trained deep neural network model of the target device, determine the location of the fire hazard of the target device and the fire hazard identification result.

9. A storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. An electronic device, comprising at least a memory and a processor, wherein a computer program is stored in the memory, wherein: The processor implements the steps of the method of any one of claims 1 to 7 when executing the computer program on the memory.

Images