Fault locating method, system and electronic equipment for power communication network

Abstract

This invention relates to the technical field of power communication, and more particularly to a fault location method, system, and electronic device for power communication networks. The method includes: training a Bi-GRU neural network model using historical network disturbance alarm data from the power communication network to obtain a trained Bi-GRU neural network model; when network disturbance alarm data is detected in the power communication network, applying the trained Bi-GRU neural network model to obtain the fault location result corresponding to the network disturbance alarm data. This method addresses the problem of increasingly complex power communication network topologies, the large variety and complexity of alarm data generated, and the difficulty for maintenance personnel to accurately locate the root cause alarm fault in a short time.

Description

Technical Field

[0001] This invention relates to the technical field of power communication, and in particular to a fault location method, system, and electronic equipment for power communication networks. Background Technology

[0002] As a crucial component of energy production, transmission, and consumption, the power communication network, with its wide service range, strong configuration capabilities, high security and reliability, and green and low-carbon characteristics, has become a major platform for the energy internet. The power communication network is primarily used to transmit various control signals from power production, providing highly reliable, low-latency, and high-bandwidth transmission channels for dedicated networks such as dispatch data networks and data communication networks. It is an indispensable carrier for ensuring the safe and stable operation of the power system. Therefore, ensuring the normal operation of the power communication network is of great significance to the entire power system.

[0003] In recent years, due to the continuous development and advancement of power communication networks, the topology of power communication networks has become increasingly complex, resulting in a wide variety of alarm data with complex content. This makes it difficult for operation and maintenance personnel to accurately locate the root cause of alarm faults in a short period of time. Summary of the Invention

[0004] This invention provides a fault location method, system, and electronic equipment for power communication networks, which addresses the problem that the increasingly complex topology of power communication networks generates a wide variety of alarm data with complex content, making it difficult for maintenance personnel to accurately locate the root cause alarm fault in a short time.

[0005] This invention provides a fault location method for power communication networks, comprising:

[0006] The Bi-GRU neural network model is trained using historical network disturbance alarm data from the power communication network to obtain the trained Bi-GRU neural network model.

[0007] When network disturbance alarm data is detected in the power communication network, the trained Bi-GRU neural network model is applied to obtain the fault location result corresponding to the network disturbance alarm data.

[0008] According to the present invention, a fault location method for a power communication network includes training a Bi-GRU neural network model using historical network disturbance alarm data in the power communication network to obtain a trained Bi-GRU neural network model, comprising:

[0009] Historical network disturbance alarm data in the power communication network is acquired and processed to obtain an alarm dataset, which is composed of the data characteristics of each historical network disturbance alarm data.

[0010] Obtain the frequent itemsets in the alarm dataset, and based on the frequent itemsets in the alarm dataset, obtain the association rule set of the alarm dataset;

[0011] The association rule set of the alarm dataset is input into the Bi-GRU neural network model for training, resulting in the trained Bi-GRU neural network model.

[0012] A fault location method for a power communication network provided by the present invention further includes:

[0013] When the network disturbance alarm data is detected in the power communication network, the trained Bi-GRU neural network model is trained again according to the data characteristics corresponding to the network disturbance alarm data.

[0014] Obtain the model parameters after retraining to get the fluctuation range of each model parameter before and after retraining;

[0015] The trained Bi-GRU neural network model is updated based on the model parameters whose fluctuation range is less than the preset fluctuation range.

[0016] According to the fault location method for power communication networks provided by the present invention, the Bi-GRU neural network model includes an input layer, an encoding layer, a decoding layer, and a fully connected layer;

[0017] The encoding layer of the Bi-GRU neural network model consists of two layers of bidirectional gated recurrent unit neural networks, and the decoding layer of the Bi-GRU neural network model is structurally symmetrical with the encoding layer.

[0018] The present invention also provides a fault location system for a power communication network, comprising:

[0019] The offline unit is used to train the Bi-GRU neural network model using historical network disturbance alarm data in the power communication network, so as to obtain the trained Bi-GRU neural network model.

[0020] The online unit is used to obtain the fault location result corresponding to the network disturbance alarm data by applying a trained Bi-GRU neural network model when network disturbance alarm data is detected in the power communication network.

[0021] According to the fault location system of the power communication network provided by the present invention, the offline unit specifically includes:

[0022] The data processing subunit is used to acquire historical network disturbance alarm data in the power communication network and process the historical network disturbance alarm data to obtain an alarm dataset, wherein the alarm dataset is composed of the data features of each historical network disturbance alarm data.

[0023] The training set acquisition sub-unit is used to acquire frequent itemsets in the alarm dataset and obtain the association rule set of the alarm dataset based on the frequent itemsets of the alarm dataset.

[0024] The model training subunit is used to input the association rule set of the alarm dataset into the Bi-GRU neural network model for training, so as to obtain the trained Bi-GRU neural network model.

[0025] According to the fault location system of the power communication network provided by the present invention, the offline unit further includes: a model update subunit, used to retrain the trained Bi-GRU neural network model according to the data features corresponding to the network disturbance alarm data when the network disturbance alarm data is detected in the power communication network; obtain the model parameters obtained after retraining to obtain the fluctuation range of each model parameter before and after retraining; and update the trained Bi-GRU neural network model according to the model parameters whose fluctuation range is less than a preset fluctuation range.

[0026] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the fault location method for power communication networks as described above.

[0027] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the fault location method for power communication networks as described above.

[0028] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the fault location method for power communication networks as described above.

[0029] The fault location method, system, and electronic equipment for power communication networks provided by this invention utilize a Bi-GRU neural network model to train on historical network disturbance alarm data in the power communication network. This model automatically learns the complex relationships between alarm data, enabling rapid and accurate fault location when a new network disturbance alarm is detected. This improves the efficiency and accuracy of fault diagnosis and reduces the time and cost of manual analysis. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0031] Figure 1 This is one of the flowcharts illustrating the fault location method for power communication networks provided by the present invention.

[0032] Figure 2 This is the second flowchart of the fault location method for power communication networks provided by the present invention.

[0033] Figure 3 This is a schematic diagram of the principle of a Bi-GRU neural network model provided by the present invention.

[0034] Figure 4 This is the third flowchart of the fault location method for power communication networks provided by the present invention.

[0035] Figure 5 This is one of the simulation results diagrams of various network models provided by this invention.

[0036] Figure 6 This is the second of the simulation results diagrams of various network models provided by this invention.

[0037] Figure 7 This is the third of the simulation results diagrams of various network models provided by this invention.

[0038] Figure 8 This is a schematic diagram of the fault location system for power communication networks provided by the present invention.

[0039] Figure 9 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0041] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such descriptions can be used interchangeably where appropriate to allow embodiments to be implemented in a sequence other than that illustrated or described in this application. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or modules is not necessarily limited to those explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices. The naming or numbering of steps appearing in this application does not imply that the steps in the method flow must be performed in the chronological / logical order indicated by the naming or numbering. The execution order of named or numbered process steps can be changed according to the desired technical purpose, as long as the same or similar technical effect is achieved. The module division described in this application is a logical division. In practical applications, there may be other division methods. For example, multiple modules may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the shown or discussed mutual coupling, direct coupling, or communication connection may be through some interface, and the indirect coupling or communication connection between units may be electrical or other similar forms, none of which are limited in this application. Furthermore, the modules or sub-modules described as separate components may or may not be physically separated, may or may not be physical units, or may be distributed in multiple circuit units. Some or all of the units can be selected to achieve the purpose of the solution in this application according to actual needs.

[0042] The following is combined Figures 1-9 The specific contents of this invention are described below.

[0043] Figure 1 This is one of the flowcharts illustrating the fault location method for power communication networks provided by the present invention, such as... Figure 1 As shown, the method includes steps S101-S102.

[0044] Step S101: Train the Bi-GRU neural network model using historical network disturbance alarm data from the power communication network to obtain the trained Bi-GRU neural network model.

[0045] Step S102: When network disturbance alarm data is detected in the power communication network, the trained Bi-GRU neural network model is applied to obtain the fault location result corresponding to the network disturbance alarm data.

[0046] In embodiments of the present invention, the historical network disturbance alarm data in the power communication network includes raw data collected from various gateway devices without any data processing, including alarm data related to gateways, environment, communication connections, performance, and devices. Because the raw network disturbance alarm data contains a large amount of non-alarm data, invalid data, redundant data, etc., and the raw alarm data is unevenly distributed, data preprocessing is required on the raw alarm data in the data management database to ensure the quality of the alarm data and the efficiency of processing and analysis.

[0047] In embodiments of the present invention, the "multi-source data aggregation" technology is used to import different types of network disturbance alarm data into the data management library through a unified entry point.

[0048] In one possible implementation, such as Figure 2 As shown, the Bi-GRU neural network model is trained using historical network disturbance alarm data in the power communication network to obtain the trained Bi-GRU neural network model, including steps S201-S203.

[0049] Step S201: Obtain historical network disturbance alarm data from the power communication network, process the historical network disturbance alarm data to obtain an alarm dataset, which consists of the data features of each historical network disturbance alarm data.

[0050] Specifically, historical network disturbance alarm data in the power communication network is processed to obtain an alarm dataset, which includes:

[0051] In the data management database, alarm information fields related to fault feature extraction are selected to obtain historical network disturbance alarm data. Basic processing such as data cleaning, deduplication, and noise reduction are performed on this historical data to provide fundamental data support for locating network disturbance sources. An appropriate time window is selected to synchronously process alarm information at the time of the fault, forming an alarm transaction set. The network disturbance alarm information in the alarm transaction set, after real-time aggregation and preprocessing, is then merged. Using data association and overriding techniques, data is associated with specified fields and merged. Similar fields are overridden using specified judgment logic, resulting in a single alarm dataset.

[0052] Step S202: Obtain the frequent itemsets in the alarm dataset, and obtain the association rule set of the alarm dataset based on the frequent itemsets in the alarm dataset.

[0053] Specifically, the data features extracted from the alarm dataset are used as input, and frequent itemsets and association rules are mined based on a weighted association analysis algorithm.

[0054] Step S203: Input the association rule set of the alarm dataset into the Bi-GRU neural network model for training to obtain the trained Bi-GRU neural network model.

[0055] Specifically, frequent itemsets and association rules mined by association analysis will be used as training sample sets and fed into the Bi-GRU neural network model for learning and training. This will locate the fault source from a large number of disturbance alarm features and mine the root alarm that can directly reflect the fault source in the communication network. An optimization algorithm will be used to minimize the loss function to improve the effectiveness and accuracy of fault location association analysis.

[0056] In one possible implementation, the Bi-GRU neural network model includes an input layer, an encoding layer, a decoding layer, and a fully connected layer; the encoding layer of the Bi-GRU neural network model consists of two layers of bidirectional gated recurrent unit neural networks, and the decoding layer of the Bi-GRU neural network model is structurally symmetrical to the encoding layer.

[0057] Specifically, considering the dynamic and random nature of alarm data, in order to ensure the accuracy of the mapping between alarm data and fault location, a bidirectional Bi-GRU neural network is used to continuously update the new network disturbance alarm data and fault location mapping relationship.

[0058] For example, such as Figure 3 As shown, the entire Bi-GRU neural network model includes an input layer, an encoding layer, a decoding layer, a fully connected layer, and an output layer. The input layer receives historical data from various alarm messages in the alarm system and preprocesses the data, including anomaly removal and standardization. The encoding layer consists of two layers of bidirectional gated recurrent unit neural networks, used to compress and reduce the dimensionality of the data to fully extract the inherent information of the time series. The decoding layer is symmetrical to the encoding layer, used to progressively recover the dimensionality-reduced features, and outputs the reconstructed time series through the fully connected layer. All nonlinear activation functions in the network layers use the hyperbolic tangent (tanh) function.

[0059] In an embodiment of the present invention, a loss function is selected to calculate the difference between the localization result of the Bi-GRU neural network model and the actual fault location. By minimizing the loss value, the model convergence and the fault localization error are minimized, thereby accurately locating the fault. By calculating the difference between the localization result and the actual fault location, and feeding the error back to the network, the parameters are adjusted based on the feedback information, so that the error between its output value and the actual fault gradually decreases, assisting in rapid fault localization and preventing misjudgment or misadjustment in communication scheduling.

[0060] During training, the softmax function is used as the loss function to evaluate the model's performance. The softmax function maps each component of a vector to the interval [0,1], ensuring that the sum of all component outputs is 1. Assume the original neural network output is... The output after the softmax function, as shown in the equation, can be viewed as the probability distribution of the predicted class. For the i-th category.

[0061]

[0062] In one possible implementation, the fault location method for the power communication network, such as Figure 4 As shown, it also includes steps S401-S403.

[0063] Step S401: When network disturbance alarm data is detected in the power communication network, the trained Bi-GRU neural network model is trained again based on the data characteristics corresponding to the network disturbance alarm data.

[0064] Step S402: Obtain the model parameters after retraining to obtain the fluctuation range of each model parameter before and after retraining.

[0065] Step S403: Update the trained Bi-GRU neural network model based on the model parameters whose fluctuation range is less than the preset fluctuation range.

[0066] Specifically, the Bi-GRU neural network model is trained using historical network disturbance alarm data from the power communication network. When a new alarm event occurs, it is preprocessed, its associated features are extracted, and then fed into the initial neural network model for training, resulting in an updated neural network model. Considering that the updated neural network is trained using new alarm events, its parameters may be more biased towards the new alarm data. To prevent the neural network from forgetting previously learned knowledge, parameters are selectively chosen. For example, parameters with a variation range less than λ (in this embodiment, the preset fluctuation range λ is 0-10%) are selected as the parameters of the original neural network model, while parameters with a variation greater than 10% are discarded.

[0067] The most detailed embodiments of the present invention are given below by way of example.

[0068] In the embodiments of the present invention, the Adam gradient descent algorithm is used, the learning rate of the Bi-GRU neural network model is set to 0.005, the maximum number of training iterations is 1000, and the number of forward and backward GRU layers is set to 20. The results of the squared absolute error, mean square error and root mean square error of the training set and the test set can be obtained, as shown in Table 1.

[0069] Table 1

[0070]

[0071] Wherein, MAE represents Mean Absolute Error, MSE represents Mean Squared Error, and RMSE represents Root Mean Squared Error. In fault location models, these three evaluation metrics quantify the error between the model's predicted values ​​and the actual values. RMSE, however, assigns a higher weight to large errors, better reflecting the model's performance in extreme situations.

[0072] Root mean square error (RMSE) is used as the primary criterion for evaluating fault location performance. RMSE is defined as the square root of the mean of the squared prediction errors and is a commonly used metric for measuring model prediction accuracy. By comparing the RMSE values ​​of different models on the same dataset, the performance of each model in accurately locating system faults is evaluated. Bi-GRU network models, RNN network models, and LSTM network models were trained on the same dataset, and the simulation results are shown below. Figure 5 As shown in the figure, the RMSE value of the trained Bi-GRU neural network model in this invention is significantly lower than that of other neural network models, indicating its effectiveness and accuracy in handling complex fault localization.

[0073] To verify the impact of the learning rate on model performance, the root mean square error of fault localization for each neural network model was compared by varying the learning rate. The comparison results are as follows: Figure 6 As shown. By Figure 6 As can be seen, the RMSE values ​​of all three models decrease with the continuous increase of the learning rate. This is because the RMSE value decreases continuously with the increase of the learning rate, and the accuracy of fault location also increases. Compared with the LSTM model and the RNN model, the Bi-GRU neural network model has a lower RMSE for fault location using alarm data from the power communication network, indicating that the method proposed in this invention is superior in the process of accurate fault location.

[0074] By selecting a basic dataset and gradually changing the dataset size, the execution time of the training and learning process is compared. The comparison results are as follows: Figure 7As shown in the figure, the execution time increases with the increase of the dataset. Compared with the other two neural network models, the Bi-GRU neural network model proposed in this invention has a lower execution time, ensuring the real-time performance of the fault localization process in complex networks.

[0075] By employing the above method, the present invention can achieve at least one of the following beneficial effects:

[0076] 1. By training historical network disturbance alarm data in the power communication network using a Bi-GRU neural network model, the complex relationships between alarm data are automatically learned. This allows for rapid and accurate fault location when new network disturbance alarms are detected. This improves the efficiency and accuracy of fault diagnosis and reduces the time and cost of manual analysis.

[0077] 2. When network disturbance alarm data is detected in the power communication network, the trained Bi-GRU neural network model is retrained based on the data characteristics corresponding to the alarm data. The fluctuation range of each model parameter before and after retraining is obtained. Model parameters with fluctuation ranges smaller than a preset range are used to update the trained Bi-GRU neural network model. This avoids a decrease in model accuracy due to parameter updates.

[0078] The fault location system for power communication networks provided by the present invention will be described below. The fault location system for power communication networks described below can be referred to in correspondence with the fault location method for power communication networks described above.

[0079] like Figure 8 As shown, the present invention also provides a fault location system for a power communication network, comprising:

[0080] The offline unit is used to train the Bi-GRU neural network model using historical network disturbance alarm data in the power communication network, so as to obtain the trained Bi-GRU neural network model.

[0081] The online unit is used to obtain the fault location result corresponding to the network disturbance alarm data by applying the trained Bi-GRU neural network model when network disturbance alarm data is detected in the power communication network.

[0082] In one possible implementation, the offline unit specifically includes:

[0083] The data processing subunit is used to acquire historical network disturbance alarm data in the power communication network and process the historical network disturbance alarm data to obtain an alarm dataset. The alarm dataset consists of the data characteristics of each historical network disturbance alarm data.

[0084] The training set obtains a sub-unit, which is used to obtain frequent itemsets in the alarm dataset, and obtains the association rule set of the alarm dataset based on the frequent itemsets in the alarm dataset;

[0085] The model training subunit is used to input the association rule set of the alarm dataset into the Bi-GRU neural network model for training, and obtain the trained Bi-GRU neural network model.

[0086] In an embodiment of the present invention, the alarm dataset consists of data features of various historical network disturbance alarm data. The offline unit automatically divides the alarm dataset into training and test datasets for training the Bi-GRU neural network model.

[0087] In one possible implementation, the offline unit further includes:

[0088] The model update subunit is used to retrain the trained Bi-GRU neural network model based on the data characteristics corresponding to the network disturbance alarm data when network disturbance alarm data is detected in the power communication network; obtain the model parameters obtained after retraining to obtain the fluctuation range of each model parameter before and after retraining; and update the trained Bi-GRU neural network model based on the model parameters whose fluctuation range is less than the preset fluctuation range.

[0089] In embodiments of the present invention, historical normal operation status data includes historical network disturbance alarm data in the power communication network. An offline unit is used to train and update the Bi-GRU neural network model, and load the trained and updated Bi-GRU neural network model into the online unit. In the online unit, when the location result does not match the actual fault location, fault location will be re-performed.

[0090] Figure 9 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 9 As shown, the electronic device may include: a processor 910, a communications interface 920, a memory 930, and a communication bus 940, wherein the processor 910, the communications interface 920, and the memory 930 communicate with each other via the communication bus 940. The processor 910 can call logical instructions in the memory 930 to execute a fault location method for a power communication network, the method including:

[0091] The Bi-GRU neural network model is trained using historical network disturbance alarm data from the power communication network to obtain the trained Bi-GRU neural network model.

[0092] When network disturbance alarm data is detected in the power communication network, the trained Bi-GRU neural network model is applied to obtain the fault location result corresponding to the network disturbance alarm data.

[0093] Furthermore, the logical instructions in the aforementioned memory 930 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0094] On the other hand, the present invention also provides a computer program product, the computer program product comprising a computer program that can be stored on a non-transitory computer-readable storage medium, wherein when the computer program is executed by a processor, the computer is able to execute a fault location method for a power communication network provided by the above methods, the method comprising:

[0095] The Bi-GRU neural network model is trained using historical network disturbance alarm data from the power communication network to obtain the trained Bi-GRU neural network model.

[0096] When network disturbance alarm data is detected in the power communication network, the trained Bi-GRU neural network model is applied to obtain the fault location result corresponding to the network disturbance alarm data.

[0097] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform a fault location method for a power communication network provided by the methods described above, the method comprising:

[0098] The Bi-GRU neural network model is trained using historical network disturbance alarm data from the power communication network to obtain the trained Bi-GRU neural network model.

[0099] When network disturbance alarm data is detected in the power communication network, the trained Bi-GRU neural network model is applied to obtain the fault location result corresponding to the network disturbance alarm data.

[0100] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0101] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0102] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method of fault location for a power communication network, characterized by, include: The Bi-GRU neural network model is trained using historical network disturbance alarm data from the power communication network to obtain the trained Bi-GRU neural network model. When network disturbance alarm data is detected in the power communication network, the trained Bi-GRU neural network model is applied to obtain the fault location result corresponding to the network disturbance alarm data. The process of training a Bi-GRU neural network model using historical network disturbance alarm data from the power communication network to obtain a trained Bi-GRU neural network model includes: Historical network disturbance alarm data in the power communication network is acquired and processed to obtain an alarm dataset, which is composed of the data characteristics of each historical network disturbance alarm data. Obtain the frequent itemsets in the alarm dataset, and based on the frequent itemsets in the alarm dataset, obtain the association rule set of the alarm dataset; The association rule set of the alarm dataset is input into the Bi-GRU neural network model for training to obtain the trained Bi-GRU neural network model. The Bi-GRU neural network model includes an input layer, an encoding layer, a decoding layer, and a fully connected layer. The input layer is used to receive historical data of various alarm information in the alarm system and preprocess the data. The encoding layer consists of two layers of bidirectional gated recurrent unit neural networks and is used to compress and reduce the dimensionality of the data. The decoding layer is symmetrical to the encoding layer and is used to gradually recover the features after dimensionality reduction. The fully connected layer is used to output the reconstructed time series.

2. The fault location method of a power communication network according to claim 1, characterized by, Also includes: When the network disturbance alarm data is detected in the power communication network, the trained Bi-GRU neural network model is trained again according to the data characteristics corresponding to the network disturbance alarm data. Obtain the model parameters after retraining to get the fluctuation range of each model parameter before and after retraining; The trained Bi-GRU neural network model is updated based on the model parameters whose fluctuation range is less than the preset fluctuation range.

3. A fault location system for a power communication network, characterized by include: The offline unit is used to train the Bi-GRU neural network model using historical network disturbance alarm data in the power communication network, so as to obtain the trained Bi-GRU neural network model. An online unit is used to obtain the fault location result corresponding to the network disturbance alarm data by applying a trained Bi-GRU neural network model when network disturbance alarm data is detected in the power communication network. The offline unit specifically includes: The data processing subunit is used to acquire historical network disturbance alarm data in the power communication network and process the historical network disturbance alarm data to obtain an alarm dataset, wherein the alarm dataset is composed of the data features of each historical network disturbance alarm data. The training set acquisition sub-unit is used to acquire frequent itemsets in the alarm dataset and obtain the association rule set of the alarm dataset based on the frequent itemsets of the alarm dataset. The model training subunit is used to input the association rule set of the alarm dataset into the Bi-GRU neural network model for training, so as to obtain the trained Bi-GRU neural network model. The Bi-GRU neural network model includes an input layer, an encoding layer, a decoding layer, and a fully connected layer. The input layer is used to receive historical data of various alarm information in the alarm system and preprocess the data. The encoding layer consists of two layers of bidirectional gated recurrent unit neural networks and is used to compress and reduce the dimensionality of the data. The decoding layer is symmetrical to the encoding layer and is used to gradually recover the features after dimensionality reduction. The fully connected layer is used to output the reconstructed time series.

4. The fault location system of a power communication network according to claim 3, characterized in that, The offline unit further includes: The model update subunit is used to retrain the trained Bi-GRU neural network model based on the data characteristics corresponding to the network disturbance alarm data when the network disturbance alarm data is detected in the power communication network; obtain the model parameters obtained after retraining to obtain the fluctuation range of each model parameter before and after retraining; and update the trained Bi-GRU neural network model based on the model parameters whose fluctuation range is less than the preset fluctuation range.

5. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the fault location method for the power communication network as described in claim 1 or 2. 6.A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, When the computer program is executed by the processor, it implements the fault location method for the power communication network as described in claim 1 or 2.

7. A computer program product comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in claim 1 or 2.

Images