Hybrid model-based power anomaly identification method, apparatus and device, and medium

Abstract

The invention discloses a hybrid model-based power anomaly recognition method, device and equipment and a medium, and the method comprises the steps: obtaining a to-be-recognized power anomaly image, and inputting the power anomaly image into a pre-trained hybrid power anomaly recognition model; respectively performing first image feature extraction and second image feature extraction on the power anomaly image through the first anomaly recognition model and the second anomaly recognition model to generate a first image feature and a second image feature; carrying out feature alignment on the first image feature and the second image feature; and performing feature fusion on the first image feature and the second image feature after feature alignment to generate a fused abnormal feature based on the abnormal level information of the abnormality of the power equipment, identifying the fused abnormal feature, and outputting the abnormal type of the power abnormal image. According to the technical scheme, the recognition precision and robustness of the model are improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method, device, equipment and medium for identifying power anomalies based on a series-parallel model. Background Art

[0002] Fault diagnosis or abnormality identification of power equipment is an important part of ensuring the safe operation of the power system.

[0003] Traditional fault diagnosis or anomaly identification often relies on a single model for prediction or classification, but a single model is difficult to fully handle complex power equipment failure problems. Improving the performance of a single model by increasing the network depth and parameters of the single model will increase the demand for computing power, resulting in increased usage costs and limiting the promotion of the model.

[0004] Therefore, there is an urgent need for a power anomaly identification method that is easy to promote and can comprehensively handle complex power equipment failures. Summary of the invention

[0005] The present invention provides a method, device, equipment and medium for identifying power anomalies based on a series-parallel model, so as to improve the accuracy of identifying power anomalies.

[0006] According to one aspect of the present invention, a method for identifying power anomalies based on a series-parallel model is provided, the method comprising:

[0007] Acquire a power anomaly image to be identified, and input the power anomaly image into a pre-trained hybrid power anomaly identification model; wherein the hybrid power anomaly identification model includes a first anomaly identification model and a second anomaly identification model;

[0008] By using the first abnormality recognition model and the second abnormality recognition model, the first image feature extraction and the second image feature extraction are performed on the power abnormality image respectively to generate a first image feature and a second image feature; wherein the first image feature is used to characterize the fine-grained texture feature information of the power abnormality image, and the second image feature is used to characterize the global feature and spatial feature of the power abnormality image;

[0009] Performing feature alignment for the first image feature and the second image feature; wherein the feature alignment process includes head feature alignment and body feature alignment;

[0010] Based on the abnormality level information of the abnormality occurring in the power equipment, feature fusion is performed on the first image feature and the second image feature after feature alignment to generate a fused abnormality feature, and the fused abnormality feature is identified to output the abnormality type of the power abnormality image.

[0011] According to another aspect of the present invention, there is provided a power anomaly identification device based on a series-parallel model, the device comprising:

[0012] An image acquisition module, used for acquiring an electric power anomaly image to be identified, and inputting the electric power anomaly image into a pre-trained hybrid electric power anomaly recognition model; wherein the hybrid electric power anomaly recognition model includes a first anomaly recognition model and a second anomaly recognition model;

[0013] A feature extraction module, used to perform first image feature extraction and second image feature extraction on the power abnormality image through the first abnormality recognition model and the second abnormality recognition model, respectively, to generate first image features and second image features; wherein the first image features are used to characterize fine-grained texture feature information of the power abnormality image, and the second image features are used to characterize global features and spatial features of the power abnormality image;

[0014] A feature alignment module, used for performing feature alignment on the first image feature and the second image feature; wherein the feature alignment process includes head feature alignment and body feature alignment;

[0015] The abnormality determination module is used to perform feature fusion to generate fused abnormality features based on the abnormality level information of the abnormality occurring in the power equipment and the first image features and the second image features after feature alignment, and to identify the fused abnormality features and output the abnormality type of the power abnormality image.

[0016] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0017] at least one processor; and

[0018] a memory communicatively connected to the at least one processor; wherein,

[0019] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the power anomaly identification method based on the series-parallel model described in any embodiment of the present invention.

[0020] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the power anomaly identification method based on the hybrid model described in any embodiment of the present invention when executed.

[0021] According to another aspect of the present invention, a computer program product is provided. The computer program product includes a computer program. When the computer program is executed by a processor, the method for identifying an electric power anomaly based on a series-parallel model described in any embodiment of the present invention is implemented.

[0022] The technical solution of the embodiment of the present invention proposes a multi-model hybrid power equipment anomaly detection method, which performs anomaly recognition on power anomaly images based on a hybrid model that integrates multiple anomaly recognition models, extracts and fuses features of the power anomaly images from multiple feature dimensions, and can extract more abnormal feature information of the power anomaly images, realize network feature compensation and precision alignment, and through the reasoning communication of multiple models, on the one hand, improves the recognition accuracy and robustness of the model, and on the other hand, enables the model to recognize more complex anomaly types, thereby improving the model performance of the model.

[0023] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0025] Figure 1 is a flow chart of a method for identifying power anomalies based on a series-parallel model according to Embodiment 1 of the present invention;

[0026] Figure 2 is a flow chart of a method for identifying power anomalies based on a series-parallel model according to a second embodiment of the present invention;

[0027] Figure 3 2 is a schematic diagram of the structure of a power anomaly identification device based on a series-parallel model according to Embodiment 3 of the present invention;

[0028] Figure 4 It is a structural schematic diagram of an electronic device for implementing the power anomaly identification method based on a series-parallel model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0029] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0030] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0031] Embodiment 1

[0032] Figure 1 A flowchart of a method for identifying power anomalies based on a hybrid model is provided for the first embodiment of the present invention. This embodiment is applicable to the case of identifying anomalies occurring in power equipment. The method can be performed by a power anomaly identification device based on a hybrid model. The power anomaly identification device based on a hybrid model can be implemented in the form of hardware and / or software. The power anomaly identification device based on a hybrid model can be configured in various general computing devices, such as client devices or server devices that integrate a neural network model and a large model. Figure 1 As shown, the method includes:

[0033] S110, obtaining an electric power anomaly image to be identified, and inputting the electric power anomaly image into a pre-trained hybrid electric power anomaly identification model.

[0034] The power abnormality image to be identified may refer to a device image of a power device having an abnormality, and the power abnormality image needs to be asynchronously identified to determine the abnormality type of the abnormality occurring in the power device.

[0035] The hybrid power anomaly recognition model may include a first anomaly recognition model and a second anomaly recognition model. It should be noted that the first anomaly recognition model and the second anomaly recognition model have different model structures, the second anomaly recognition model may be a neural network model, and the second anomaly recognition model may be a visual large model. The hybrid power anomaly recognition model combines the advantages of the neural network model and the visual large model and has better model performance.

[0036] S120 , performing first image feature extraction and second image feature extraction on the power abnormality image by using the first abnormality recognition model and the second abnormality recognition model respectively, to generate first image features and second image features.

[0037] The first image feature can be used to characterize the fine-grained texture feature information of the power anomaly image, and the second image feature can be used to characterize the global feature and spatial feature of the power anomaly image.

[0038] In an embodiment of the present invention, the first abnormality recognition model can extract local fine-grained texture feature information and edge feature information of the power abnormality image in the form of a convolution kernel traversing the image to generate a first image feature; the second abnormality recognition model can extract global feature information and spatial feature information of the power abnormality image through a codec converter to generate a second image feature. Optionally, the first image feature and the second image feature can both be represented in the form of a feature vector.

[0039] Optionally, in an embodiment of the present invention, before feature extraction is performed on the power abnormality image, it is also necessary to perform image preprocessing operations on the power abnormality image, such as image scale normalization, illumination contrast normalization, image feature sharpening and other preprocessing operations, so as to ensure the consistency of the power abnormality image and improve the image quality of the power abnormality image.

[0040] S130: Perform feature alignment on the first image feature and the second image feature.

[0041] The feature alignment process may include head feature alignment and body feature alignment. It should be noted that the head feature may refer to spatial position feature information. The body feature may refer to the feature ontology of the first image feature and the second image feature (ie, the body itself is the feature ontology).

[0042] Specifically, the first image features and the second image features extracted by the first abnormality recognition model and the second abnormality recognition model can be aligned so that the first image features and the second image features are in the same feature space, so that the two are comparable in the same feature space, laying a foundation for subsequent feature fusion of the two.

[0043] S140, based on the abnormality level information of the abnormality occurring in the power equipment, perform feature fusion on the first image feature and the second image feature after feature alignment to generate a fused abnormality feature, identify the fused abnormality feature, and output the abnormality type of the power abnormality image.

[0044] The abnormality level information can be used to characterize the critical degree of the abnormality of the power equipment, and the abnormality level information can be determined according to the on-site conditions of the abnormality of the power equipment. Optionally, the abnormality level information of the abnormality of the power equipment can be determined according to a pre-specified on-site inspection system of the power grid.

[0045] Specifically, weights can be assigned to the first image features and the second image features after feature alignment based on the abnormality level information of the abnormality occurring in the power equipment, and feature fusion can be performed according to the weights assigned to the two to generate a higher-dimensional fused abnormal feature. The fused abnormal feature can then be identified to determine the abnormal type of the power abnormality image.

[0046] Optionally, in an embodiment of the present invention, based on the abnormality level information of the abnormality occurring in the power equipment, feature fusion is performed on the first image feature and the second image feature after feature alignment to generate a fused abnormality feature, including: determining the confidence weights of the first image feature and the second image feature according to the abnormality level information of the abnormality occurring in the power equipment; and performing feature fusion on the first image feature and the second image feature after feature alignment according to the confidence weight to generate a fused abnormality feature.

[0047] Specifically, based on the abnormality level information of the abnormality occurring in the power equipment and the mapping relationship between the pre-set abnormality level and the confidence weights of the first image feature and the second image feature, the respective confidence weights of the first image feature and the second image feature under the abnormality level information can be determined, and based on their respective confidence weights, the first image feature and the second image feature are feature-connected in the same feature space to generate a fused abnormality feature.

[0048] In an optional implementation of an embodiment of the present invention, before the first image feature and the second image feature are subjected to a feature fusion process, the first image feature or the second image feature may be subjected to a dimensionality reduction process according to the device type of the power equipment. This is to take into account that when the device type of the power equipment is different, the image features of different dimensions in the power anomaly image corresponding to the power equipment have different degrees of influence on the abnormal feature. Exemplarily, feature analysis may be performed by a principal component analysis method to obtain the main feature components in the first image feature and the second image feature. When the device type of the power equipment to which the power anomaly image belongs is a large-size hardware-type power equipment, the main feature components in the first image feature are subjected to a dimensionality reduction process, and the main feature components of the first image feature after the dimensionality reduction process are subjected to feature fusion with the second image feature; when the device type of the power equipment to which the power anomaly image belongs is a small-size hardware-type power equipment, the main feature components in the second image feature are subjected to a dimensionality reduction process, and the main feature components of the second image feature after the dimensionality reduction process are subjected to feature fusion with the first image feature.

[0049] By performing dimensionality reduction processing on the first image feature and the second image feature, it is ensured that the data processing volume and memory overhead are reduced while not affecting the data quality of the feature data, thereby improving the recognition efficiency of the model.

[0050] Optionally, in an embodiment of the present invention, the fused abnormal features are identified and the abnormal type of the power abnormal image is output, including: establishing an optimal mapping relationship between the defect type and the optimal recognition model based on the historical model data of the first abnormal recognition model and the second abnormal recognition model and the model evaluation index; according to the abnormal type of the power abnormal image and the optimal mapping relationship, using the output result of the first abnormal recognition model or the second abnormal recognition model on the fused abnormal features as the abnormal type of the power abnormal image output by the hybrid power abnormal recognition model.

[0051] Specifically, the optimal mapping relationship between each abnormality type and the optimal recognition model can be generated in advance based on the prior knowledge such as the historical training data, historical recognition data and model evaluation indicators (accuracy and recall rate, etc.) of the first abnormality recognition model and the second abnormality recognition model, that is, the first abnormality recognition model or the second abnormality recognition model with the best model quality (high recognition accuracy) under each abnormality type is used as the optimal recognition model, and a mapping relationship is established with the abnormality type. According to the optimal mapping relationship and the abnormality recognition results of the fused abnormality type output by the first abnormality recognition model or the second abnormality recognition model, the optimal recognition model is determined, and then the abnormality recognition results of the optimal recognition model are used as the abnormality type of the power abnormality image output by the hybrid power abnormality recognition model. It should be noted that the optimal mapping relationship can be dynamically updated according to the periodic model performance evaluation results of the hybrid power abnormality recognition model.

[0052] By establishing the optimal mapping relationship, the recognition accuracy of the model for anomaly identification is further improved.

[0053] The technical solution of the embodiment of the present invention performs anomaly recognition on the power anomaly image based on a hybrid model that integrates multiple anomaly recognition models, extracts and fuses features of the power anomaly image from multiple feature dimensions, and can extract more abnormal feature information of the power anomaly image. On the one hand, it improves the recognition accuracy and robustness of the model, and on the other hand, it enables the model to recognize more complex anomaly types, thereby improving the model performance of the model.

[0054] Embodiment 2

[0055] Figure 2 This is a flowchart of a method for identifying power anomalies based on a hybrid model provided in the second embodiment of the present invention. This embodiment further refines the above embodiment and provides specific steps for aligning the features of the first image feature and the second image feature. It should be noted that for the undetailed parts of the embodiment of the present invention, please refer to the relevant descriptions of other embodiments, which will not be repeated here. Figure 2 As shown, the method includes:

[0056] S210, obtaining an electric power anomaly image to be identified, and inputting the electric power anomaly image into a pre-trained hybrid electric power anomaly identification model.

[0057] S220 , performing first image feature extraction and second image feature extraction on the power abnormality image by using the first abnormality recognition model and the second abnormality recognition model respectively, to generate first image features and second image features.

[0058] S230, down-sampling and up-sampling the first image feature and the second image feature respectively to perform resolution alignment; and mapping the first image feature after resolution alignment to the same codec dimension as the second image feature to perform main feature alignment.

[0059] S240: After the first image feature and the second image feature are aligned for main body features, feature dimensions of the head features of the first image feature and the second image feature are enhanced respectively, and the head features are aligned according to a cosine similarity feature matching algorithm.

[0060] S250, based on the abnormality level information of the abnormality occurring in the power equipment, perform feature fusion on the first image feature and the second image feature after feature alignment to generate a fused abnormality feature, identify the fused abnormality feature, and output the abnormality type of the power abnormality image.

[0061] In an embodiment of the present invention, in order to ensure the alignment accuracy of the first image feature and the second image feature after feature alignment, two alignment processes of main feature alignment and head feature alignment are performed. First, the first image feature and the second image feature need to be downsampled and upsampled respectively, and the resolutions of the first image feature and the second image feature are sampled to the same resolution dimension through interpolation or transposed convolution to ensure the consistency of resolution size, and then the first image feature is mapped to the same codec dimension as the second image feature for main feature fusion, so that the two are in the same feature space and have the same feature dimension, which is convenient for subsequent feature fusion.

[0062] After the main features of the first image feature and the second image feature are aligned, the feature dimensions of the head features of the first image feature and the second image feature in the same feature space are respectively enhanced through point convolution sampling, and the head features are aligned according to the cosine similarity feature matching algorithm. The high-dimensional head features after the feature dimension enhancement are aligned and compensated for the alignment accuracy through the codec converter.

[0063] In another optional implementation of the embodiment of the present invention, the accuracy of the first image feature extracted by the first abnormality recognition model may be checked according to the device type of the power device identified by the second abnormality recognition model.

[0064] In an embodiment of the present invention, since the second abnormality recognition model adopts the model structure of the visual large model, the second abnormality recognition large model can more accurately identify the equipment type of the power equipment to which the power abnormality image belongs, and then, based on the equipment type of the power abnormality image to which the power abnormality image identified by the second abnormality recognition model belongs, the first image feature extracted by the first abnormality recognition model from the power abnormality image can be detected, and the feature information in the first image feature that does not conform to the equipment type can be screened out.

[0065] By verifying the accuracy of the first image feature, the accuracy of the image feature information is ensured, thereby improving the model performance of the model.

[0066] The technical solution of the embodiment of the present invention improves the alignment accuracy of the first image feature and the second image feature by aligning the main features and the head features of the first image feature and the second image feature, ensures the data quality after the fusion of the first image feature and the second image feature, and improves the recognition accuracy of the model in the past two years.

[0067] Embodiment 3

[0068] Figure 3 This is a schematic diagram of the structure of a power anomaly identification device based on a series-parallel model provided in the third embodiment of the present invention. Figure 3 As shown, the device comprises:

[0069] The image acquisition module 310 is used to acquire the power anomaly image to be identified, and input the power anomaly image into a pre-trained hybrid power anomaly recognition model; wherein the hybrid power anomaly recognition model includes a first anomaly recognition model and a second anomaly recognition model;

[0070] A feature extraction module 320 is used to perform first image feature extraction and second image feature extraction on the power abnormality image through the first abnormality recognition model and the second abnormality recognition model, respectively, to generate first image features and second image features; wherein the first image features are used to characterize the fine-grained texture feature information of the power abnormality image, and the second image features are used to characterize the global features and spatial features of the power abnormality image;

[0071] A feature alignment module 330 is used to perform feature alignment on the first image feature and the second image feature; wherein the feature alignment process includes head feature alignment and body feature alignment;

[0072] The abnormality determination module 340 is used to perform feature fusion to generate a fused abnormality feature based on the abnormality level information of the abnormality occurring in the power equipment, for the first image feature and the second image feature after feature alignment, and to identify the fused abnormality feature, and output the abnormality type of the power abnormality image.

[0073] The technical solution of the embodiment of the present invention performs anomaly recognition on the power anomaly image based on a hybrid model that integrates multiple anomaly recognition models, extracts and fuses features of the power anomaly image from multiple feature dimensions, and can extract more abnormal feature information of the power anomaly image. On the one hand, it improves the recognition accuracy and robustness of the model, and on the other hand, it enables the model to recognize more complex anomaly types, thereby improving the model performance of the model.

[0074] Optionally, the feature alignment module 330 includes:

[0075] A subject alignment unit is used to downsample and upsample the first image feature and the second image feature respectively to perform resolution alignment; and map the first image feature after resolution alignment to the same codec dimension as the second image feature to perform subject feature alignment;

[0076] A head alignment unit is used to enhance the feature dimensions of the head features of the first image feature and the second image feature respectively after the main features of the first image feature and the second image feature are aligned, and to align the head features according to the cosine similarity feature matching algorithm; wherein the head feature refers to the spatial position feature information.

[0077] Optionally, the abnormality determination module 340 includes:

[0078] A weight determination unit, configured to determine the confidence weights of the first image feature and the second image feature according to the abnormality level information of the abnormality occurring in the power equipment;

[0079] The feature fusion unit is used to fuse the first image feature and the second image feature after feature alignment according to the confidence weight to generate a fused abnormal feature.

[0080] Optionally, the abnormality determination module 340 further includes:

[0081] A dimension reduction unit is used to perform dimension reduction processing on the first image feature or the second image feature according to the device type of the power device.

[0082] Optionally, the abnormality determination module 340 includes:

[0083] A mapping relationship establishing unit, used to establish an optimal mapping relationship between a defect type and an optimal recognition model according to the historical model data and model evaluation indicators of the first abnormality recognition model and the second abnormality recognition model;

[0084] The output unit is used to use the output result of the first abnormality recognition model or the second abnormality recognition model on the fused abnormality feature as the abnormality type of the power abnormality image output by the hybrid power abnormality recognition model according to the abnormality type of the power abnormality image and the optimal mapping relationship.

[0085] Optionally, the device further comprises:

[0086] The feature verification module is used to verify the accuracy of the first image feature extracted by the first abnormality recognition model according to the equipment type of the power equipment identified by the second abnormality recognition model.

[0087] The power anomaly identification device based on the series-parallel model provided in the embodiment of the present invention can execute the power anomaly identification method based on the series-parallel model provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0088] Embodiment 4

[0089] Figure 4A schematic diagram of an electronic device 410 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0090] like Figure 4 As shown, the electronic device 410 includes at least one processor 411, and a memory connected to the at least one processor 411 in communication, such as a read-only memory (ROM) 412, a random access memory (RAM) 413, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 411 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 412 or the computer program loaded from the storage unit 418 to the random access memory (RAM) 413. In the RAM 413, various programs and data required for the operation of the electronic device 410 can also be stored. The processor 411, the ROM 412, and the RAM 413 are connected to each other via a bus 414. An input / output (I / O) interface 415 is also connected to the bus 414.

[0091] Multiple components in the electronic device 410 are connected to the I / O interface 415, including: an input unit 416, such as a keyboard, a mouse, etc.; an output unit 417, such as various types of displays, speakers, etc.; a storage unit 418, such as a disk, an optical disk, etc.; and a communication unit 419, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 419 allows the electronic device 410 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0092] The processor 411 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 411 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 411 executes the various methods and processes described above, such as the power anomaly identification method based on the hybrid model.

[0093] In some embodiments, the power anomaly identification method based on the series-parallel model may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 418. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 410 via the ROM 412 and / or the communication unit 419. When the computer program is loaded into the RAM 413 and executed by the processor 411, one or more steps of the power anomaly identification method based on the series-parallel model described above may be performed. Alternatively, in other embodiments, the processor 411 may be configured to execute the power anomaly identification method based on the series-parallel model in any other appropriate manner (e.g., by means of firmware).

[0094] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0095] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0096] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, device, or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0097] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0098] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0099] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0100] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0101] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for identifying power anomalies based on a series-parallel model, characterized in that: include: Acquire a power anomaly image to be identified, and input the power anomaly image into a pre-trained hybrid power anomaly identification model; wherein the hybrid power anomaly identification model includes a first anomaly identification model and a second anomaly identification model; By using the first abnormality recognition model and the second abnormality recognition model, the first image feature extraction and the second image feature extraction are performed on the power abnormality image respectively to generate a first image feature and a second image feature; wherein the first image feature is used to characterize the fine-grained texture feature information of the power abnormality image, and the second image feature is used to characterize the global feature and spatial feature of the power abnormality image; Performing feature alignment for the first image feature and the second image feature; wherein the feature alignment process includes head feature alignment and body feature alignment; Based on the abnormality level information of the abnormality occurring in the power equipment, feature fusion is performed on the first image feature and the second image feature after feature alignment to generate a fused abnormality feature, and the fused abnormality feature is identified to output the abnormality type of the power abnormality image.

2. The method according to claim 1, characterized in that Performing feature alignment on the first image feature and the second image feature includes: Downsampling and upsampling the first image feature and the second image feature respectively to perform resolution alignment; and mapping the first image feature after resolution alignment to the same codec dimension as the second image feature to perform subject feature alignment; After the main features of the first image feature and the second image feature are aligned, the feature dimensions of the head features of the first image feature and the second image feature are respectively enhanced, and the head features are aligned according to the cosine similarity feature matching algorithm; wherein the head feature refers to spatial position feature information.

3. The method according to claim 1, characterized in that Based on the abnormality level information of the abnormality occurring in the power equipment, feature fusion is performed on the first image feature and the second image feature after feature alignment to generate a fused abnormality feature, including: Determining confidence weights of the first image feature and the second image feature according to abnormality level information of the abnormality occurring in the power equipment; According to the confidence weight, the first image feature and the second image feature after feature alignment are fused to generate a fused abnormal feature.

4. The method according to claim 3, characterized in that Also includes: The first image feature or the second image feature is subjected to dimensionality reduction processing according to the device type of the electric power device.

5. The method according to claim 1, characterized in that Identifying the fused abnormal features and outputting the abnormal type of the power abnormality image includes: Establishing an optimal mapping relationship between defect type and optimal recognition model according to historical model data and model evaluation indicators of the first abnormality recognition model and the second abnormality recognition model; According to the abnormality type of the power abnormality image and the optimal mapping relationship, the output result of the first abnormality recognition model or the second abnormality recognition model on the fused abnormality feature is used as the abnormality type of the power abnormality image output by the hybrid power abnormality recognition model.

6. The method according to claim 1, characterized in that Also includes: According to the equipment type of the electric power equipment identified by the second abnormality recognition model, the accuracy of the first image feature extracted by the first abnormality recognition model is checked.

7. A power anomaly identification device based on a series-parallel model, characterized in that: include: An image acquisition module, used for acquiring an electric power anomaly image to be identified, and inputting the electric power anomaly image into a pre-trained hybrid electric power anomaly recognition model; wherein the hybrid electric power anomaly recognition model includes a first anomaly recognition model and a second anomaly recognition model; A feature extraction module, used to perform first image feature extraction and second image feature extraction on the power abnormality image through the first abnormality recognition model and the second abnormality recognition model, respectively, to generate first image features and second image features; wherein the first image features are used to characterize fine-grained texture feature information of the power abnormality image, and the second image features are used to characterize global features and spatial features of the power abnormality image; A feature alignment module, used for performing feature alignment on the first image feature and the second image feature; wherein the feature alignment process includes head feature alignment and body feature alignment; The abnormality determination module is used to perform feature fusion to generate fused abnormality features based on the abnormality level information of the abnormality occurring in the power equipment and the first image features and the second image features after feature alignment, and to identify the fused abnormality features and output the abnormality type of the power abnormality image.

8. The device according to claim 7, characterized in that The feature alignment module comprises: A subject alignment unit is used to downsample and upsample the first image feature and the second image feature respectively to perform resolution alignment; and map the first image feature after resolution alignment to the same codec dimension as the second image feature to perform subject feature alignment; A head alignment unit is used to enhance the feature dimensions of the head features of the first image feature and the second image feature respectively after the main features of the first image feature and the second image feature are aligned, and to align the head features according to the cosine similarity feature matching algorithm; wherein the head feature refers to the spatial position feature information.

9. The device according to claim 7, characterized in that The abnormality determination module comprises: A weight determination unit, configured to determine the confidence weights of the first image feature and the second image feature according to the abnormality level information of the abnormality occurring in the power equipment; The feature fusion unit is used to fuse the first image feature and the second image feature after feature alignment according to the confidence weight to generate a fused abnormal feature.

10. The device according to claim 9, characterized in that The abnormality determination module also includes: A dimension reduction unit is used to perform dimension reduction processing on the first image feature or the second image feature according to the device type of the power device.

11. The device according to claim 7, characterized in that The abnormality determination module comprises: A mapping relationship establishing unit, used to establish an optimal mapping relationship between a defect type and an optimal recognition model according to the historical model data and model evaluation indicators of the first abnormality recognition model and the second abnormality recognition model; The output unit is used to use the output result of the first abnormality recognition model or the second abnormality recognition model on the fused abnormality feature as the abnormality type of the power abnormality image output by the hybrid power abnormality recognition model according to the abnormality type of the power abnormality image and the optimal mapping relationship.

12. The device according to claim 7, characterized in that Also includes: The feature verification module is used to verify the accuracy of the first image feature extracted by the first abnormality recognition model according to the equipment type of the power equipment identified by the second abnormality recognition model.

13. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the power anomaly identification method based on the series-parallel model described in any one of claims 1 to 6.

14. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the power anomaly identification method based on a series-parallel model described in any one of claims 1 to 6 when executed.

15. A computer program product, characterized in that It comprises a computer program, which, when executed by a processor, implements the power anomaly identification method based on the series-parallel model as described in any one of claims 1 to 6.

Images