Electric power overhead line fault identification method and system
Through iterative learning and training of effectiveness prediction network methods, a power overhead line fault recognition model with high recognition accuracy is generated, which solves the problems of low efficiency and poor accuracy of traditional methods, and achieves fast and accurate fault recognition and efficient fault recovery.
Patent Information
- Application Number
- CN202411828575.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-12
- Publication Date
- 2025-04-15
AI Technical Summary
The traditional method of fault identification of overhead power lines is inefficient and has poor accuracy, and cannot be monitored in real time, resulting in long fault location and recovery time and large economic losses.
By obtaining the initial line fault identification model and iterative line fault identification model, iterative parameter learning is performed using power overhead line monitoring data, a training effectiveness prediction network is generated, and advanced parameter learning is performed, and target line fault identification model with high recognition accuracy and robustness is output.
It significantly improves the accuracy and efficiency of line fault identification, and can quickly and accurately identify faults of overhead power lines, shorten the fault recovery time, and reduce economic losses.
Smart Images
Figure CN119827898A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of big data technology, and more specifically, to a method and system for identifying faults in power overhead lines. Background Art
[0002] As an important part of the power system, the safe and stable operation of overhead power lines is crucial to ensuring power supply. However, due to long-term exposure to the natural environment, the lines are easily affected by various external factors, such as bad weather, human damage, equipment aging, etc., which lead to frequent line failures. Therefore, timely and accurate identification of overhead power line faults is of great significance for quickly locating the fault point, shortening the fault recovery time, and reducing economic losses.
[0003] Traditional methods for identifying faults in power overhead lines mainly rely on manual inspections and simple fault indicators, which have disadvantages such as low efficiency, poor accuracy, and inability to monitor in real time. Summary of the invention
[0004] In view of this, the purpose of the present application is to provide a method and system for identifying power overhead line faults.
[0005] In conjunction with the first aspect of the present application, a method for identifying an overhead power line fault is provided, which is applied to an overhead power line fault identification system, and the method comprises:
[0006] Acquire an initialization line fault identification model and an iterative line fault identification model, wherein the iterative line fault identification model is generated by performing iterative parameter learning on the initialization line fault identification model by setting power overhead line monitoring data;
[0007] Generate sample line fault feature data using the initialization line fault identification model and / or the iterative line fault identification model, and add training supervision data to the sample line fault feature data based on the training validity of the sample line fault feature data, so as to use the sample line fault feature data after adding the training supervision data as the data to be learned to perform parameter learning on a preset machine learning network, and generate a training validity prediction network;
[0008] Based on the training validity prediction network, a training validity prediction is performed on the line fault identification result generated by the iterative line fault identification model, so as to perform advanced parameter learning on the iterative line fault identification model based on the training validity prediction data, and output the iterative line fault identification model that has completed the advanced parameter learning as a target line fault identification model.
[0009] In a possible implementation manner of the first aspect, obtaining the initialization line fault identification model and the iterative line fault identification model includes:
[0010] Get the basic neural network;
[0011] Optimizing the initialization parameters of the basic neural network based on a pre-collected sample learning sequence to generate an initialized line fault recognition model, wherein the sample learning sequence includes power overhead line monitoring data that does not carry line fault label data;
[0012] Iterative parameter learning is performed on the initialized line fault identification model based on set power overhead line monitoring data to generate an iterative line fault identification model.
[0013] In a possible implementation manner of the first aspect, performing iterative parameter learning on the initialized line fault identification model based on set power overhead line monitoring data to generate an iterative line fault identification model includes:
[0014] defining a plurality of sample labels, wherein the sample labels include line feature labels and / or line area labels;
[0015] Clustering the set power overhead line monitoring data based on each of the sample labels to generate clustered monitoring data corresponding to each of the sample labels;
[0016] Iterative parameter learning is performed on the initialized line fault identification model based on each of the cluster monitoring data to generate iterative line fault identification models corresponding to each of the sample labels.
[0017] In a possible implementation manner of the first aspect, generating sample line fault feature data by using the initialization line fault identification model and / or the iterative line fault identification model includes:
[0018] Get trigger signal sample;
[0019] The trigger signal samples are loaded into the initialization line fault identification model and the iterative line fault identification model respectively, and the identification results of the initialization line fault identification model and the iterative line fault identification model are used as sample line fault feature data respectively.
[0020] In a possible implementation of the first aspect, the sample line fault feature data after adding the training supervision data is used as the to-be-learned data to perform parameter learning on the preset machine learning network to generate a training effectiveness prediction network by the following method:
[0021] Define a sample size threshold value, and use the sample line fault feature data after adding training supervision data as the data to be learned;
[0022] If the scale of the data to be learned is not less than the sample scale threshold value, parameter learning is performed on a preset machine learning network based on the data to be learned to generate a training validity prediction network, wherein the training validity prediction network is used to extract fault knowledge features from the line fault identification results and generate training validity prediction data based on the fault knowledge features.
[0023] In a possible implementation of the first aspect, performing training validity prediction on the line fault identification result generated by the iterative line fault identification model based on the training validity prediction network, so as to perform advanced parameter learning on the iterative line fault identification model based on the training validity prediction data, includes:
[0024] Loading the same set trigger signal to the initialization line fault identification model and the iterative line fault identification model respectively, generating a line fault identification result corresponding to the initialization line fault identification model and a line fault identification result corresponding to the iterative line fault identification model;
[0025] Based on the training validity prediction network, advanced parameter learning value prediction is performed on the line fault identification results corresponding to the initialization line fault identification model and the line fault identification results corresponding to the iterative line fault identification model, respectively, to generate training validity prediction data corresponding to the initialization line fault identification model and training validity prediction data corresponding to the iterative line fault identification model;
[0026] Determine a training validity error between the initialization line fault identification model and the iterative line fault identification model based on the training validity prediction data corresponding to the initialization line fault identification model and the training validity prediction data generated by the iterative line fault identification model;
[0027] According to the training validity error, the iterative line fault identification model is trained based on a back propagation algorithm.
[0028] In a possible implementation of the first aspect, performing training validity prediction on the line fault identification result generated by the iterative line fault identification model based on the training validity prediction network, so as to perform advanced parameter learning on the iterative line fault identification model based on the training validity prediction data, includes:
[0029] Defining a model performance threshold value and acquiring optimized learning data, wherein the training effectiveness prediction data corresponding to the optimized learning data meets the model performance threshold value;
[0030] Performing training validity prediction on the line fault identification result generated by the iterative line fault identification model based on the training validity prediction network to generate training validity prediction data corresponding to the iterative line fault identification model;
[0031] If the training validity prediction data is less than the model performance threshold, advanced parameter learning is performed on the iterative line fault identification model based on the optimized learning data until the training validity prediction data corresponding to the iterative line fault identification model meets the model performance threshold.
[0032] In a possible implementation of the first aspect, the method further includes:
[0033] Based on the target line fault identification model, line fault identification is performed on any input power overhead line monitoring data to generate a line fault identification result.
[0034] In combination with the second aspect of the present application, a power overhead line fault identification system is provided, the power overhead line fault identification system comprising a machine-readable storage medium and a processor, the machine-readable storage medium storing machine-executable instructions, and when the processor executes the machine-executable instructions, the power overhead line fault identification system implements the aforementioned power overhead line fault identification method.
[0035] In conjunction with the third aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed, the aforementioned method for identifying faults in power overhead lines is implemented.
[0036] In combination with any of the above aspects, the embodiments of the present application significantly improve the accuracy and efficiency of line fault identification. First, the method uses the power overhead line monitoring data to perform iterative parameter learning on the initialized line fault identification model to generate an iterative line fault identification model, thereby enhancing the model's adaptability to actual line fault characteristics. Secondly, by generating sample line fault feature data and adding training supervision data, a high-quality training data set is constructed, and then an effective training effectiveness prediction network is trained. The prediction network can accurately evaluate the training effectiveness of the line fault identification results generated by the iterative line fault identification model, thereby guiding the advanced parameter learning of the iterative line fault identification model. The target line fault identification model that is finally output has undergone multiple optimizations and has higher recognition accuracy and robustness. It can quickly and accurately identify faults of power overhead lines in practical applications, providing a strong guarantee for the safe and stable operation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained by combining these drawings without paying creative work.
[0038] Figure 1 A schematic flow chart of a method for identifying a fault in an overhead power line provided in an embodiment of the present application. DETAILED DESCRIPTION
[0039] 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 are within the scope of protection of the present invention.
[0040] The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, device, product or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units inherent to these processes, methods, products or ends.
[0041] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.
[0042] Figure 1 The flowchart of the method for identifying a fault in an overhead power line provided by an embodiment of the present application is shown. It should be understood that in other embodiments, the order of some steps in the method for identifying a fault in an overhead power line of the present embodiment can be shared with each other based on actual needs, or some steps can be omitted or maintained. The method for identifying a fault in an overhead power line includes:
[0043] Step S110, obtaining an initialization line fault identification model and an iterative line fault identification model, wherein the iterative line fault identification model is generated by performing iterative parameter learning on the initialization line fault identification model by setting power overhead line monitoring data.
[0044] In this embodiment, the server first obtains the basic neural network from its repository. This basic neural network can be a multilayer perceptron (MLP) structure, which has an input layer, several hidden layers and an output layer. For example, the number of neurons in the input layer is determined according to the number of features of the power overhead line monitoring data. Assuming that the power overhead line monitoring data contains 10 feature quantities such as voltage, current, and power factor, then the input layer has 10 neurons. The number of hidden layers and the number of neurons in each hidden layer can be set according to experience or some predefined rules, such as setting it to 3 hidden layers, each with 20, 15, and 10 neurons respectively. The output layer determines the number of neurons according to the number of fault types to be identified. If 5 different line fault types are to be identified, then the output layer has 5 neurons.
[0045] Then, a large amount of power overhead line monitoring data is collected from the database of the power system, and these data constitute the sample learning sequence. However, these data do not carry line fault label data. For example, the server collected power overhead line monitoring data of different time periods and different line sections in the past month, with a total of about 100,000 data records. The server inputs this data into the basic neural network. In this process, the server uses the Stochastic Gradient Descent (SGD) algorithm to optimize the parameters of the basic neural network. It will calculate the error between the output result and the expected result (since there is no fault label, this can be some expectations based on data statistical characteristics) based on the propagation results of the input data in the network, and then adjust the parameters such as weights and biases in the network based on this error. After multiple iterations, for example, after 1,000 iterations, when the error converges to an acceptable range, an initialized line fault recognition model is generated.
[0046] The server defines multiple sample tags, including line feature tags and line area tags. Line feature tags can be divided according to the physical characteristics of the power overhead line, such as "high voltage line", "low voltage line", "high current line", etc.; line area tags can be divided according to the geographical location of the line, such as "city A line", "city B line", etc.
[0047] Assume that the server defines three line feature tags (high voltage line, low voltage line, high current line) and two line area tags (city A line, city B line). The server obtains the set power overhead line monitoring data from the power system database. These data are data with complete label information, totaling about 50,000 data records.
[0048] For each sample label, the server clusters the set power overhead line monitoring data. Taking the line feature label "high voltage line" as an example, the server will cluster the data using the K-means clustering algorithm based on the features related to the high voltage line in the data (such as high voltage value, specific power factor range, etc.). Assume that after clustering, 3 cluster monitoring data subsets are obtained.
[0049] Then, the server performs iterative parameter learning on the initialized line fault identification model based on each clustered monitoring data subset. For the clustered monitoring data subset corresponding to the "high voltage line" label, the server inputs it into the initialized line fault identification model. The stochastic gradient descent algorithm is used again to adjust the parameters of the model according to the input data and the corresponding fault labels (such as the fault type labels that may occur in the high voltage line). After multiple iterations, for example, 500 iterations, an iterative line fault identification model corresponding to the "high voltage line" label is generated. In the same way, the server processes the clustered monitoring data corresponding to other sample labels to generate the corresponding iterative line fault identification model.
[0050] Step S120, using the initialized line fault identification model and / or the iterative line fault identification model to generate sample line fault feature data, and adding training supervision data to the sample line fault feature data based on the training validity of the sample line fault feature data, so as to use the sample line fault feature data after adding the training supervision data as the data to be learned to perform parameter learning on the preset machine learning network, and generate a training validity prediction network.
[0051] In this embodiment, the server obtains a trigger signal sample. This trigger signal sample can be a small fragment of real-time monitoring data from a specific line segment in the power system, such as the voltage, current, temperature and other monitoring data of a high-voltage line at a certain moment. The server loads this trigger signal sample into the initialization line fault identification model and the iterative line fault identification model respectively. For the initialization line fault identification model, when the trigger signal sample is input, the model will calculate according to its internal parameters and algorithms and output an identification result. Assume that this identification result is a vector, in which each element represents the possibility of different fault types, for example [0.1, 0.3, 0.2, 0.1, 0.3], which respectively represent the possibility of 5 different fault types. Similarly, for the iterative line fault identification model, a similar identification result will also be output, such as [0.2, 0.25, 0.15, 0.2, 0.2]. These identification results are respectively used as sample line fault feature data by the server.
[0052] Then, define a sample size threshold value. Assume that this sample size threshold value is set to 100 sets of sample line fault feature data. The server will continuously collect sample line fault feature data, and when the size of the collected sample line fault feature data is not less than this sample size threshold value, it will start the subsequent operation.
[0053] For each set of sample line fault feature data, the server will evaluate its training effectiveness. The evaluation of training effectiveness can be based on a variety of factors, such as the degree of conformity of the recognition results with the actual fault conditions (if there is reference data of the actual fault conditions), the stability of the recognition results, etc. Assuming that for a certain set of sample line fault feature data, its recognition results deviate greatly from the known actual fault conditions, the server considers that its training effectiveness is low. Based on the evaluation results of training effectiveness, the server adds training supervision data to the sample line fault feature data. For example, if the training effectiveness of a certain set of sample line fault feature data is low, the server may add more supervision information to it, such as adjusting the weight of the fault type.
[0054] The server uses the sample line fault feature data after adding the training supervision data as the data to be learned. Assume that 100 sets of sample line fault feature data after adding the training supervision data have been collected. The server uses these data to be learned to learn the parameters of the preset machine learning network. This preset machine learning network can be a convolutional neural network (CNN) structure, which is specifically used to learn the prediction mode of training effectiveness from the sample line fault feature data. The server inputs the data to be learned into this CNN, calculates the output result through forward propagation, and then uses the back propagation algorithm to adjust the parameters in the CNN according to the error between the output result and the expected training effectiveness result (which can be determined based on the previous evaluation of the sample line fault feature data). After multiple iterations, for example, 500 iterations, when the error converges to an acceptable range, a training effectiveness prediction network is generated. This training effectiveness prediction network can extract fault knowledge features from the line fault identification results and generate training effectiveness prediction data based on these fault knowledge features.
[0055] Step S130, performing training validity prediction on the line fault identification result generated by the iterative line fault identification model based on the training validity prediction network, so as to perform advanced parameter learning on the iterative line fault identification model based on the training validity prediction data, and outputting the iterative line fault identification model that has completed the advanced parameter learning as a target line fault identification model.
[0056] In this embodiment, the server first loads the same set trigger signal to the initialization line fault identification model and the iterative line fault identification model respectively. The set trigger signal can be complete monitoring data from a key line in the power system, including continuous data of multiple characteristic quantities such as voltage, current, temperature, etc. over a period of time.
[0057] For the initialization line fault identification model, when the set trigger signal is input, the corresponding line fault identification result will be generated. Assume that this result is a vector, such as [0.15, 0.25, 0.1, 0.2, 0.3], representing the possibility of 5 different fault types. For the iterative line fault identification model, similar line fault identification results will also be generated, such as [0.2, 0.2, 0.15, 0.2, 0.25].
[0058] The server then predicts the advanced parameter learning value of the line fault identification results corresponding to the initialization line fault identification model and the line fault identification results corresponding to the iterative line fault identification model based on the training effectiveness prediction network. The training effectiveness prediction network analyzes these identification results and extracts the fault knowledge features therein. For example, it may find that the probability change trend of certain fault types in the identification results of the iterative line fault identification model does not conform to the normal operation law of the power system, thereby judging that there may be problems with its training effectiveness. Based on these analyses, the training effectiveness prediction network generates corresponding training effectiveness prediction data. Assume that for the initialization line fault identification model, the generated training effectiveness prediction data is 0.8 (the value here can be a relative indicator obtained based on the calculation mechanism inside the network, indicating the level of training effectiveness), and for the iterative line fault identification model, the generated training effectiveness prediction data is 0.7.
[0059] The server determines the training validity error between the training validity prediction data corresponding to the initialization line fault identification model and the training validity prediction data generated by the iteration line fault identification model. In this example, the training validity error is 0.8-0.7=0.1.
[0060] Based on this training validity error, the server trains the iterative line fault identification model based on the back propagation algorithm. It adjusts the parameters in the iterative line fault identification model according to the error, such as adjusting the weights and biases in the neural network. Assume that after 100 iterations of adjustment, the parameters of the iterative line fault identification model are optimized.
[0061] In another way, the server defines a model performance threshold, assuming that this threshold is set to 0.85. The server obtains optimized learning data, which has been previously screened and whose corresponding training effectiveness prediction data meets this model performance threshold. For example, from a large number of previous line fault identification results and corresponding training effectiveness prediction data, those data with training effectiveness prediction data greater than or equal to 0.85 are screened as optimized learning data.
[0062] The server performs training validity prediction on the line fault identification result generated by the iterative line fault identification model based on the training validity prediction network, and generates training validity prediction data corresponding to the iterative line fault identification model. Assume that this data is 0.8, which is less than the model performance threshold value of 0.85.
[0063] The server then performs advanced parameter learning on the iterative line fault identification model based on the optimized learning data. It inputs the optimized learning data into the iterative line fault identification model and uses the back propagation algorithm again to adjust the model's parameters based on the information in the optimized learning data. This process is repeated until the training validity prediction data corresponding to the iterative line fault identification model meets the model performance threshold. For example, after multiple adjustments, the training validity prediction data corresponding to the iterative line fault identification model reaches 0.85.
[0064] When the iterative line fault identification model completes advanced parameter learning and its training validity prediction data meets the requirements (such as meeting the model performance threshold), the server outputs the iterative line fault identification model as the target line fault identification model. This target line fault identification model can be used to identify line faults for any input power overhead line monitoring data and generate line fault identification results. For example, when new power overhead line monitoring data is input into the power system, this target line fault identification model can accurately analyze the data and determine whether there is a line fault and the type of fault.
[0065] Based on the above steps, the embodiment of the present application significantly improves the accuracy and efficiency of line fault identification. First, the method uses the power overhead line monitoring data to perform iterative parameter learning on the initialized line fault identification model to generate an iterative line fault identification model, thereby enhancing the model's adaptability to actual line fault characteristics. Secondly, by generating sample line fault feature data and adding training supervision data, a high-quality training data set is constructed, and then an effective training effectiveness prediction network is trained. The prediction network can accurately evaluate the training effectiveness of the line fault identification results generated by the iterative line fault identification model, thereby guiding the advanced parameter learning of the iterative line fault identification model. The target line fault identification model that is finally output has undergone multiple optimizations and has higher recognition accuracy and robustness. It can quickly and accurately identify faults of power overhead lines in practical applications, providing a strong guarantee for the safe and stable operation of the power system.
[0066] In a possible implementation manner, obtaining an initialization line fault identification model and an iteration line fault identification model includes:
[0067] Get the base neural network.
[0068] The initialization parameters of the basic neural network are optimized based on a pre-collected sample learning sequence to generate an initialized line fault recognition model, wherein the sample learning sequence includes power overhead line monitoring data that does not carry line fault label data.
[0069] Iterative parameter learning is performed on the initialized line fault identification model based on set power overhead line monitoring data to generate an iterative line fault identification model.
[0070] In a possible implementation, performing iterative parameter learning on the initialized line fault identification model based on set power overhead line monitoring data to generate an iterative line fault identification model includes:
[0071] A plurality of sample labels are defined, wherein the sample labels include line feature labels and / or line area labels.
[0072] The power overhead line monitoring data is clustered based on each of the sample labels to generate clustered monitoring data corresponding to each of the sample labels.
[0073] Iterative parameter learning is performed on the initialized line fault identification model based on each of the cluster monitoring data to generate iterative line fault identification models corresponding to each of the sample labels.
[0074] In a possible implementation manner, generating sample line fault feature data using the initialization line fault identification model and / or the iterative line fault identification model includes:
[0075] Get a sample trigger signal.
[0076] The trigger signal samples are loaded into the initialization line fault identification model and the iterative line fault identification model respectively, and the identification results of the initialization line fault identification model and the iterative line fault identification model are used as sample line fault feature data respectively.
[0077] In a possible implementation, the following method is used to perform parameter learning on a preset machine learning network using the sample line fault feature data after adding training supervision data as the data to be learned, so as to generate a training effectiveness prediction network:
[0078] A sample size threshold is defined, and the sample line fault feature data after adding training supervision data is used as the data to be learned.
[0079] If the scale of the data to be learned is not less than the sample scale threshold value, parameter learning is performed on a preset machine learning network based on the data to be learned to generate a training validity prediction network, wherein the training validity prediction network is used to extract fault knowledge features from the line fault identification results and generate training validity prediction data based on the fault knowledge features.
[0080] In a possible implementation, performing training validity prediction on the line fault identification result generated by the iterative line fault identification model based on the training validity prediction network, so as to perform advanced parameter learning on the iterative line fault identification model based on the training validity prediction data, includes:
[0081] The same set trigger signal is loaded into the initialization line fault identification model and the iterative line fault identification model respectively, and a line fault identification result corresponding to the initialization line fault identification model and a line fault identification result corresponding to the iterative line fault identification model are generated.
[0082] Based on the training validity prediction network, advanced parameter learning value prediction is performed on the line fault identification results corresponding to the initialization line fault identification model and the line fault identification results corresponding to the iterative line fault identification model, respectively, to generate training validity prediction data corresponding to the initialization line fault identification model and training validity prediction data corresponding to the iterative line fault identification model.
[0083] A training validity error between the initialization line fault identification model and the iterative line fault identification model is determined based on the training validity prediction data corresponding to the initialization line fault identification model and the training validity prediction data generated by the iterative line fault identification model.
[0084] According to the training validity error, the iterative line fault identification model is trained based on a back propagation algorithm.
[0085] In a possible implementation, performing training validity prediction on the line fault identification result generated by the iterative line fault identification model based on the training validity prediction network, so as to perform advanced parameter learning on the iterative line fault identification model based on the training validity prediction data, includes:
[0086] A model performance threshold is defined, and optimized learning data is obtained, wherein training effectiveness prediction data corresponding to the optimized learning data meets the model performance threshold.
[0087] Based on the training validity prediction network, a training validity prediction is performed on the line fault identification result generated by the iterative line fault identification model to generate training validity prediction data corresponding to the iterative line fault identification model.
[0088] If the training validity prediction data is less than the model performance threshold, advanced parameter learning is performed on the iterative line fault identification model based on the optimized learning data until the training validity prediction data corresponding to the iterative line fault identification model meets the model performance threshold.
[0089] In a possible implementation, the method further includes:
[0090] Based on the target line fault identification model, line fault identification is performed on any input power overhead line monitoring data to generate a line fault identification result.
[0091] In the above embodiments, the power overhead line fault identification system for executing the above method embodiments has at least one processor, a control module (chip set) coupled to at least one of the (at least one) processors, a memory coupled to the control module, a non-volatile memory (NVM) / storage device coupled to the control module, at least one load / output device coupled to the control module, and a network interface coupled to the control module.
[0092] The processor may include at least one single-core or multi-core processor, and the processor may include any combination of general-purpose processors or special-purpose processors (such as graphics processors, application processors, baseband processors, etc.). For some alternative implementations, the power overhead line fault identification system can be based on electronic devices such as the gateway described in the embodiments of the present application.
[0093] For some alternative embodiments, the power overhead line fault identification system may include at least one computer-readable medium (e.g., a memory or NVM / storage device) having instructions and at least one processor integrated with the at least one computer-readable medium and configured to execute the instructions to implement a module to perform the actions described in the present disclosure.
[0094] For one embodiment, the control module may include any suitable interface controller to provide any suitable interface to at least one of the processor(s) and / or any suitable device or component in communication with the control module.
[0095] The control module may include a memory controller module to provide an interface to the memory. The memory controller module may be a hardware module, a software module, and / or a firmware module.
[0096] The memory may be used, for example, to load and store data and / or instructions for the power overhead line fault identification system.For one embodiment, the memory may include any suitable volatile memory, such as a suitable DRAM.
[0097] For one embodiment, the control module may include at least one load-in / load-out controller to provide an interface to the NVM / storage device and the (at least one) load-in / load-out device.
[0098] For example, the NVM / storage device may be used to store data and / or instructions. The NVM / storage device may include any suitable non-volatile memory (e.g., flash memory) and / or may include any suitable (at least one) non-volatile storage device (e.g., at least one hard disk drive (HDD), at least one compact disk (CD) drive, and / or at least one digital versatile disk (DVD) drive).
[0099] The NVM / storage device may include storage resources that are physically part of the device on which the power overhead line fault identification system is installed, or it may be accessible to the device without being part of the device. For example, the NVM / storage device may be accessed via (at least one) load / output device based on a network.
[0100] The (at least one) load-in / output device may provide an interface for the power overhead line fault identification system to communicate with any other appropriate device, and the load-in / output device may include a communication component, a phonetic component, a sensor component, etc. The network interface may provide an interface for the power overhead line fault identification system to communicate based on at least one network, and the power overhead line fault identification system may wirelessly communicate with at least one component of a wireless network based on any of at least one wireless network priors and / or protocols, such as accessing a wireless network based on a communication prior.
[0101] For one embodiment, at least one of the (at least one) processor may be loaded together with the logic of at least one controller of the control module (e.g., a memory controller module). For one embodiment, at least one of the (at least one) processor may be loaded together with the logic of at least one controller of the control module to form a system level load. For one embodiment, at least one of the (at least one) processor may be fused on the same die with the logic of at least one controller of the control module. For one embodiment, at least one of the (at least one) processor may be fused on the same die with the logic of at least one controller of the control module to form a system on chip (SoC).
[0102] The embodiments of the present application are introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. At the same time, for those skilled in the art, based on the idea of the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.
[0103] An embodiment of the present invention discloses a computer-readable storage medium storing a computer program for electronic data exchange, wherein the computer program enables a computer to execute the steps of the method for identifying a fault of an overhead power line described in the above embodiment.
[0104] An embodiment of the present invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to enable a computer to execute the steps in the power overhead line fault identification method described in the above embodiment.
[0105] The device embodiments described above are only illustrative, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, i.e., they may be located in one place, or they may be distributed over multiple modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. Those of ordinary skill in the art may understand and implement the present invention without creative effort.
[0106] Through the specific description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable rewritable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, a magnetic disk storage, a magnetic tape storage, or any other medium that can be used to be computer-readable with or store data.
[0107] Finally, it should be noted that what is disclosed above is only a preferred embodiment of the present invention, which is only used to illustrate the technical solution of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, it should be understood by those skilled in the art that the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.
Claims
1. A method for identifying faults in power overhead lines, characterized in that: The method comprises: Acquire an initialization line fault identification model and an iterative line fault identification model, wherein the iterative line fault identification model is generated by performing iterative parameter learning on the initialization line fault identification model by setting power overhead line monitoring data; Generate sample line fault feature data using the initialization line fault identification model and / or the iterative line fault identification model, and add training supervision data to the sample line fault feature data based on the training validity of the sample line fault feature data, so as to use the sample line fault feature data after adding the training supervision data as the data to be learned to perform parameter learning on a preset machine learning network, and generate a training validity prediction network; Based on the training validity prediction network, a training validity prediction is performed on the line fault identification result generated by the iterative line fault identification model, so as to perform advanced parameter learning on the iterative line fault identification model based on the training validity prediction data, and output the iterative line fault identification model that has completed the advanced parameter learning as a target line fault identification model.
2. The method for identifying a fault in an overhead power line according to claim 1, characterized in that: Obtaining the initialization line fault identification model and the iteration line fault identification model, including: Get the basic neural network; Optimizing the initialization parameters of the basic neural network based on a pre-collected sample learning sequence to generate an initialized line fault recognition model, wherein the sample learning sequence includes power overhead line monitoring data that does not carry line fault label data; Iterative parameter learning is performed on the initialized line fault identification model based on set power overhead line monitoring data to generate an iterative line fault identification model.
3. The method for identifying a fault in an overhead power line according to claim 2, characterized in that: Performing iterative parameter learning on the initialized line fault identification model based on set power overhead line monitoring data to generate an iterative line fault identification model includes: defining a plurality of sample labels, wherein the sample labels include line feature labels and / or line area labels; Clustering the set power overhead line monitoring data based on each of the sample labels to generate clustered monitoring data corresponding to each of the sample labels; Iterative parameter learning is performed on the initialized line fault identification model based on each of the cluster monitoring data to generate iterative line fault identification models corresponding to each of the sample labels.
4. The method for identifying a fault in an overhead power line according to claim 2, characterized in that: Generating sample line fault feature data using the initialization line fault identification model and / or the iterative line fault identification model includes: Get trigger signal sample; The trigger signal samples are loaded into the initialization line fault identification model and the iterative line fault identification model respectively, and the identification results of the initialization line fault identification model and the iterative line fault identification model are used as sample line fault feature data respectively.
5. The method for identifying a fault in an overhead power line according to claim 4, characterized in that: The following method is used to perform parameter learning on the preset machine learning network by adding the sample line fault feature data after the training supervision data is added as the data to be learned, and a training validity prediction network is generated: Define a sample size threshold value, and use the sample line fault feature data after adding training supervision data as the data to be learned; If the scale of the data to be learned is not less than the sample scale threshold value, parameter learning is performed on a preset machine learning network based on the data to be learned to generate a training validity prediction network, wherein the training validity prediction network is used to extract fault knowledge features from the line fault identification results and generate training validity prediction data based on the fault knowledge features.
6. The method for identifying a fault in an overhead power line according to any one of claims 1 to 5, characterized in that: Performing training validity prediction on the line fault identification result generated by the iterative line fault identification model based on the training validity prediction network, so as to perform advanced parameter learning on the iterative line fault identification model based on the training validity prediction data, including: Loading the same set trigger signal to the initialization line fault identification model and the iterative line fault identification model respectively, generating a line fault identification result corresponding to the initialization line fault identification model and a line fault identification result corresponding to the iterative line fault identification model; Based on the training validity prediction network, advanced parameter learning value prediction is performed on the line fault identification results corresponding to the initialization line fault identification model and the line fault identification results corresponding to the iterative line fault identification model, respectively, to generate training validity prediction data corresponding to the initialization line fault identification model and training validity prediction data corresponding to the iterative line fault identification model; Determine a training validity error between the initialization line fault identification model and the iterative line fault identification model based on the training validity prediction data corresponding to the initialization line fault identification model and the training validity prediction data generated by the iterative line fault identification model; According to the training validity error, the iterative line fault identification model is trained based on a back propagation algorithm.
7. The method for identifying a fault in an overhead power line according to any one of claims 1 to 5, characterized in that: Performing training validity prediction on the line fault identification result generated by the iterative line fault identification model based on the training validity prediction network, so as to perform advanced parameter learning on the iterative line fault identification model based on the training validity prediction data, including: Defining a model performance threshold value and acquiring optimized learning data, wherein the training effectiveness prediction data corresponding to the optimized learning data meets the model performance threshold value; Performing training validity prediction on the line fault identification result generated by the iterative line fault identification model based on the training validity prediction network to generate training validity prediction data corresponding to the iterative line fault identification model; If the training validity prediction data is less than the model performance threshold, advanced parameter learning is performed on the iterative line fault identification model based on the optimized learning data until the training validity prediction data corresponding to the iterative line fault identification model meets the model performance threshold.
8. The method for identifying a fault in an overhead power line according to any one of claims 1 to 5, characterized in that: The method further comprises: Based on the target line fault identification model, line fault identification is performed on any input power overhead line monitoring data to generate a line fault identification result.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores machine-executable instructions, and when the machine-executable instructions are executed by a computer, the method for identifying a fault of an overhead power line according to any one of claims 1 to 8 is implemented.
10. A power overhead line fault identification system, characterized in that: It comprises a processor and a computer-readable storage medium, wherein the computer-readable storage medium stores machine-executable instructions, and when the machine-executable instructions are executed by a computer, the method for identifying a fault of an overhead power line according to any one of claims 1 to 8 is implemented.