Electric power communication network fault positioning analysis method and system based on deep learning

Abstract

The invention relates to the technical field of electric power communication, and discloses an electric power communication network fault positioning analysis method and system based on deep learning, and the method comprises the steps: generating a collection and extraction scheme, carrying out the preprocessing, generating a fault positioning model, carrying out the fault prediction and risk positioning, and constructing an electric power communication network reliability operation index system. The system corresponds to the method. According to the invention, the method achieves the comprehensive and targeted obtaining of the operation data and fault information through the collection and extraction scheme generated based on the data collection and fault positioning records of the electric power communication network. By constructing a fault positioning model in which a convolutional neural network and a recurrent neural network are combined, effective feature extraction and accurate classification of power communication network faults are realized; and fault prediction and risk positioning are carried out by using the trained model, and a fault occurrence probability and a risk level are calculated based on the fault prediction and risk positioning, so that a reliable power communication network reliability operation index system is finally constructed, and network operation and maintenance decision is assisted.

Description

Technical Field

[0001] The present application relates to the field of power communication technology, and specifically to a method and system for locating and analyzing faults in a power communication network based on deep learning. Background Art

[0002] At present, the fault diagnosis of the power backbone communication network mainly relies on the company's communication dispatch at all levels to regularly / periodically inspect the important alarms and important performance data of the equipment network management and integrated network management. Fault diagnosis is too dependent on the experience of operation and maintenance personnel, and the efficiency of fault diagnosis is not high. Especially in an era when the scale of network operation and maintenance data is growing exponentially, it is no longer realistic to rely solely on manpower or experience to complete network fault diagnosis. It is urgent to develop automatic and efficient data mining and analysis tools to realize the fault diagnosis of the power backbone communication network, and combine a large amount of key information such as equipment operation history data and external data to carry out network fault prediction and risk location.

[0003] Conducting research on data collection, fault diagnosis, risk prediction and network health management technologies for power backbone communication networks based on deep learning will promote the transformation of fault handling from relying on experience to intelligent diagnosis, and from passive handling during daily operation and maintenance to active defense beforehand. It is of great significance to improve the efficiency of power grid communication dispatching fault handling and improve the safe operation level of power backbone communication networks and power grids. Traditional fault location methods often rely on manual experience and simple rule matching, which is difficult to cope with massive and complex and changeable operation data. In terms of data collection, it lacks systematicity and pertinence, and cannot accurately obtain comprehensive and effective operation data and fault information, resulting in insufficient basis for fault analysis. In the data processing link, due to the lack of a complete data cleaning, standardization and normalization process, the data quality is uneven, which interferes with the accuracy of fault location. In the construction of fault location models, a single neural network architecture or traditional algorithm is mostly used, which cannot fully mine the spatial and temporal series characteristics in the operation data of the power communication network, the feature extraction capability is limited, and the classification effect is not ideal. It is difficult to achieve real-time and accurate monitoring and analysis for fault prediction and risk location. It is impossible to deeply correlate fault characteristics based on the network topology structure, resulting in inaccurate risk location and lack of reliability in fault probability calculation, which in turn affects the construction of the reliability operation indicator system of the power communication network and cannot provide a comprehensive and accurate decision-making basis for network operation and maintenance.

[0004] Chinese patent number CN202411036477.5 discloses a method and system for predicting faults in a power backbone communication network based on deep learning, which includes: building and training a neural network model, constructing a deep learning model, and the model includes at least three hidden layers and one output layer; loading training data and test data, and preprocessing the data; using the training data set to train the constructed deep learning model; testing the model to verify its performance, and using the test data set to test the trained deep learning model; using the trained model to predict faults for input IDs, and outputting the prediction results. However, the invention fails to achieve automatic and efficient data mining analysis.

[0005] In summary, a new technical solution for power communication network fault location analysis based on deep learning is urgently needed to solve the above problems. Summary of the invention

[0006] The purpose of this application is to provide a method and system for fault location analysis in a power communication network based on deep learning to solve the technical problems raised in the above background technology.

[0007] To achieve the above objectives, this application discloses the following technical solutions:

[0008] In a first aspect, the present application discloses a method for fault location analysis of a power communication network based on deep learning, the method comprising:

[0009] S1: Based on the records of power communication network data collection and fault location, obtain the communication network fault location requirements based on deep learning, and generate a collection and extraction plan for operation data and fault information; wherein the operation data includes network flow data, equipment operation status data, and link status data, and the fault information includes fault occurrence time, fault type, and fault location information;

[0010] S2: Based on the acquisition and extraction scheme, obtain corresponding operation data and fault information, and perform preprocessing; wherein the preprocessing method includes data cleaning, data standardization and data normalization;

[0011] S3: Based on the fault characteristics of the power communication network, a deep learning-based fault location model is generated; wherein the fault location model adopts an architecture combining a convolutional neural network and a recurrent neural network, and uses the acquired operation data and fault information to complete model training. The fault location model is used for feature extraction and classification of faults in the power communication network, and the feature extraction includes extracting network traffic characteristics, equipment operation characteristics, and link characteristics related to the fault, and the classification is used to classify the fault into different types of communication faults;

[0012] S3: Collect typical communication network operation and alarm data, input them into the trained fault location model, and perform network fault prediction and risk location;

[0013] S4: Based on the communication network fault prediction and risk location, the fault occurrence probability and risk level are calculated, and a reliability operation index system for the power communication network is constructed. The reliability operation index system includes network connectivity index, network delay index, fault recovery time index and equipment failure rate index.

[0014] Preferably, the data cleaning step includes removing duplicate data, processing missing values ​​and processing outliers.

[0015] Preferably, the convolutional neural network is used to extract the spatial features of the communication network operation data, and the recurrent neural network is used to extract the time series features of the communication network operation data. The architecture combining the two performs feature fusion and classification through a fully connected layer.

[0016] Preferably, the model training uses a back-propagation algorithm to adjust model parameters.

[0017] Preferably, the fault prediction is based on real-time monitoring and analysis of the communication network operation data by a deep learning model. When the fault probability output by the model exceeds a preset fault probability threshold, it is determined that a fault may occur, and risk location is performed. The relationship between the fault characteristics and the network topology is analyzed using a risk location formula, and the network topology is represented as G=(V, E), where V is a node set and E is an edge set. The risk location formula is specifically:

[0018]

[0019] Among them, ω i is the weight of the i-th feature, node v∈V, x i is the i-th component of the fault feature vector x, v i is the i-th component of the eigenvector associated with node v, k(x i ,v i ) is the preset feature similarity measurement function, n is the number of features, P E is the proportion of faults in the edge set to which node v belongs, R(x,v) is the calculated risk level of node v, and when R(x,v)≥R τ (x,v), it is determined that node v may fail and its position is G=(V,E), R τ (x,v) is the failure probability threshold.

[0020] Preferably, in the network operation health assessment method, the probability of a fault occurring is calculated through the prediction results of a deep learning model, and the risk level is determined based on the probability of a fault occurring and the degree of impact that the fault may cause.

[0021] Preferably, the network connectivity index in the power communication network reliability operation index system is determined by calculating the proportion of reachable nodes in the network, the network delay index is measured by the average delay time, the fault recovery time index records the time from the occurrence of the fault to the restoration of normal operation, and the equipment failure rate index counts the number of equipment failures per unit time.

[0022] Preferably, the method further comprises generating a visual report based on the network operation health assessment results, wherein the visual report comprises a network health status score, fault prediction information, risk location information, and a graphical display of reliability operation indicator data.

[0023] Preferably, the deep learning-based fault location model is updated and trained regularly, and the updated training data includes newly added operating data and fault information as well as resampling of historical data.

[0024] In the second aspect, the present application discloses a power communication network fault location analysis system based on deep learning, the system is applicable to the power communication network fault location analysis method based on deep learning as described above, the system includes a collection and extraction scheme generation module, a data preprocessing module, a fault location model generation module, a fault prediction and location module and an indicator system construction module connected in sequence;

[0025] The acquisition and extraction scheme generation module is configured to: based on the records of power communication network data acquisition and fault location, obtain the communication network fault location requirements based on deep learning, and generate an acquisition and extraction scheme for operation data and fault information; wherein the operation data includes network flow data, equipment operation status data, and link status data, and the fault information includes fault occurrence time, fault type, and fault location information;

[0026] The data preprocessing module is configured to: based on the acquisition and extraction scheme, obtain corresponding operation data and fault information, and perform preprocessing; wherein the preprocessing method includes data cleaning, data standardization and data normalization;

[0027] The fault location model generation module is configured to generate a deep learning-based fault location model based on the fault characteristics of the power communication network; wherein the fault location model adopts an architecture combining a convolutional neural network and a recurrent neural network, and uses the acquired operation data and fault information to complete model training; the fault location model is used for feature extraction and classification of faults in the power communication network, wherein the feature extraction includes extracting network traffic characteristics, equipment operation characteristics, and link characteristics related to the fault, and the classification is used to classify the fault into different types of communication faults;

[0028] The fault prediction and location module is configured to: collect typical communication network operation and alarm data, input it into the trained fault location model, and perform network fault prediction and risk location;

[0029] The indicator system construction module is configured as follows: based on communication network fault prediction and risk location, the fault probability and risk level are calculated to construct a reliability operation indicator system for the power communication network. The reliability operation indicator system includes network connectivity indicators, network delay indicators, fault recovery time indicators and equipment failure rate indicators.

[0030] Beneficial effects: The deep learning-based power communication network fault location analysis method and system of the present application utilizes an acquisition and extraction scheme based on power communication network data acquisition and fault location record generation to achieve comprehensive and targeted acquisition of operation data and fault information, and uses a preprocessing method including data cleaning, standardization and normalization to ensure data quality; by constructing a fault location model that combines convolutional neural networks and recurrent neural networks, it achieves effective feature extraction and accurate classification of power communication network faults; it also uses the trained model to predict faults and locate risks, and based on this, calculates the probability of fault occurrence and risk level, and ultimately achieves the construction of a reliable power communication network reliability operation indicator system to assist in network operation and maintenance decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without paying any creative work.

[0032] Figure 1 A flowchart of a method for locating and analyzing faults in a power communication network based on deep learning provided in an embodiment of the present application;

[0033] Figure 2 A structural block diagram of a deep learning-based power communication network fault location and analysis system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0034] The technical solutions in the embodiments of the present application are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application.

[0035] In this article, the term "comprising" is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of more restrictions, the elements defined by the sentence "comprising..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0036] The first aspect of this embodiment discloses Figure 1 A method for locating and analyzing faults in a power communication network based on deep learning is shown, the method comprising:

[0037] S1: Based on the records of power communication network data collection and fault location, obtain the communication network fault location requirements based on deep learning, and generate a collection and extraction plan for operation data and fault information; where the operation data includes network traffic data, equipment operation status data, and link status data, and the fault information includes the time of fault occurrence, fault type, and fault location information;

[0038] S2: Based on the collection and extraction scheme, the corresponding operation data and fault information are obtained and preprocessed; the preprocessing method includes data cleaning, data standardization and data normalization;

[0039] S3: Based on the fault characteristics of the power communication network, a deep learning-based fault location model is generated; wherein the fault location model adopts an architecture that combines a convolutional neural network and a recurrent neural network, and uses the acquired operation data and fault information to complete the model training. The fault location model is used for feature extraction and classification of faults in the power communication network. The feature extraction includes extracting fault-related network traffic characteristics, equipment operation characteristics, and link characteristics. The classification is used to classify the fault into different types of communication faults;

[0040] S3: Collect typical communication network operation and alarm data, input them into the trained fault location model, and perform network fault prediction and risk location;

[0041] S4: Based on the prediction of communication network faults and risk location, the probability of fault occurrence and risk level are calculated, and a reliability operation index system for the power communication network is constructed. The reliability operation index system includes network connectivity index, network delay index, fault recovery time index and equipment failure rate index.

[0042] Based on the above, this embodiment utilizes a collection and extraction scheme based on the generation of power communication network data collection and fault location records to achieve comprehensive and targeted acquisition of operation data and fault information, and uses a preprocessing method including data cleaning, standardization and normalization to ensure data quality; by constructing a fault location model that combines convolutional neural networks and recurrent neural networks, it achieves effective feature extraction and accurate classification of power communication network faults; it also uses the trained model to perform fault prediction and risk location, and calculates the probability of fault occurrence and risk level based on this, and finally realizes the construction of a reliable power communication network reliability operation indicator system to assist network operation and maintenance decision-making.

[0043] Specifically, the data cleaning step includes removing duplicate data, processing missing values ​​and processing outliers; wherein, the missing value processing in this embodiment adopts a mean filling method or a model-based filling method, and the outlier value processing adopts a statistical method or a cluster analysis method.

[0044] Based on the above, this embodiment utilizes detailed data cleaning steps, including removing duplicate data, missing value processing and outlier processing, to achieve the effect of improving data quality, providing a more accurate and clean data foundation for subsequent deep learning-based fault location analysis, reducing fault location deviations caused by data problems, and improving the accuracy and reliability of the overall analysis process.

[0045] Specifically, the convolutional neural network is used to extract the spatial features of the communication network operation data, and the recurrent neural network is used to extract the time series features of the communication network operation data. The architecture combining the two performs feature fusion and classification through a fully connected layer.

[0046] Based on the above, this embodiment utilizes a convolutional neural network to extract spatial features, a recurrent neural network to extract time series features, and a fully connected layer fusion classification architecture to achieve a more comprehensive and in-depth mining of communication network operation data features. Compared with the traditional single architecture or method, it can better grasp the data characteristics and enhance the feature extraction capability of the fault location model, thereby improving the accuracy of classification of different types of communication faults.

[0047] Specifically, the model training uses the back-propagation algorithm to adjust the model parameters.

[0048] Based on the above, this embodiment uses the back propagation algorithm to adjust the model parameters, so that the fault location model based on deep learning can continuously optimize its own parameters according to the training data, so that the model can fit the data characteristics more accurately and converge to a better state quickly, thereby improving the accuracy and effectiveness of the fault location model in practical applications in judging and classifying power communication network faults, and ensuring the reliable implementation of fault location analysis.

[0049] Specifically, fault prediction is based on real-time monitoring and analysis of communication network operation data by deep learning models. When the fault probability output by the model exceeds the preset fault probability threshold, it is determined that a fault may occur and risk location is performed. The relationship between fault characteristics and network topology is analyzed using the risk location formula. The network topology is represented as G = (V, E), where V is the node set and E is the edge set. The risk location formula is as follows:

[0050]

[0051] Among them, ω i is the weight of the i-th feature, node v∈V, x i is the i-th component of the fault feature vector x, v i is the i-th component of the eigenvector associated with node v, k(x i ,v i ) is the preset feature similarity measurement function, n is the number of features, P E is the proportion of faults in the edge set to which node v belongs, R(x,v) is the calculated risk level of node v, and when R(x,v)≥R τ (x,v), it is determined that node v may fail and its position is G=(V,E), R τ (x, v) is a fault probability threshold. It should be noted that this embodiment uses any existing similarity measurement to perform similarity analysis, for example, Euclidean distance, cosine similarity, and the like.

[0052] Based on the above, this embodiment utilizes real-time monitoring and analysis of operating data based on a deep learning model, and combines risk location formulas and fault probability thresholds to determine faults and locate risks, thereby achieving real-time and accurate risk location based on fault characteristics associated with network topology structures. It overcomes the problems of inaccurate risk location and difficulty in real-time monitoring in existing technologies, making it more accurate to judge nodes and locations where failures may occur, and providing a strong basis for subsequent operation and maintenance measures.

[0053] Specifically, in the network operation health assessment method, the probability of a fault occurring is calculated through the prediction results of the deep learning model, and the risk level is determined based on the probability of a fault occurring and the degree of impact that the fault may cause.

[0054] Based on the above, this embodiment uses the deep learning model prediction results to determine the probability of fault occurrence, and combines the degree of impact that the fault may cause to determine the risk level, thereby achieving a more scientific and reasonable assessment of the fault risk situation in the power communication network. Compared with the previous rough assessment method, it can more comprehensively consider multiple factors, thereby providing operation and maintenance personnel with more valuable reference risk level information to assist in accurate decision-making.

[0055] Specifically, the network connectivity index in the reliability operation index system of the power communication network is determined by calculating the proportion of reachable nodes in the network, the network delay index is measured by the average delay time, the fault recovery time index records the time from the occurrence of the fault to the restoration of normal operation, and the equipment failure rate index counts the number of equipment failures per unit time.

[0056] Based on the above, this embodiment uses a specific method to calculate the network connectivity index, network delay index, fault recovery time index and equipment failure rate index to construct a reliability operation index system, which realizes the quantification of the operation status of the power communication network from multiple dimensions, and more intuitively and accurately reflects the reliability of the network, which is convenient for operation and maintenance personnel to fully grasp the health of the network. Compared with the existing lack of comprehensive indicator measurement, it can better guide network maintenance and optimization work.

[0057] Specifically, the method also includes generating a visualization report based on the network operation health assessment results, the visualization report including a network health status score, fault prediction information, risk location information, and a graphical display of reliability operation indicator data.

[0058] Through the above, this embodiment utilizes the operation of generating a visual report based on the network operation health assessment results to display complex assessment data in the form of intuitive and easy-to-understand charts, etc., which changes the previous situation where operation and maintenance personnel found it difficult to quickly understand data information, allowing them to more conveniently obtain key information such as fault prediction, risk location, and reliability indicators, thereby efficiently carrying out corresponding operation and maintenance management work.

[0059] Specifically, the deep learning-based fault location model is updated and trained regularly, and the updated training data includes newly added operating data and fault information as well as resampling of historical data.

[0060] Based on the above, this embodiment utilizes the method of regularly updating and training the fault location model and incorporating new and resampled data to enable the model to adapt to the ever-changing situation of the power communication network, such as structural changes, business growth, the emergence of new fault types, etc., and solves the problem of poor adaptability to new situations caused by the inability to update traditional models in a timely manner, ensuring that the model always maintains good fault location analysis performance in long-term applications.

[0061] The second aspect of this embodiment discloses Figure 2 A power communication network fault location analysis system based on deep learning is shown, and the system is applicable to the power communication network fault location analysis method based on deep learning as described above, and the system includes a collection and extraction scheme generation module, a data preprocessing module, a fault location model generation module, a fault prediction and location module, and an indicator system construction module connected in sequence;

[0062] The collection and extraction scheme generation module is configured as follows: based on the records of power communication network data collection and fault location, the communication network fault location requirements based on deep learning are obtained, and a collection and extraction scheme for operation data and fault information is generated; wherein the operation data includes network flow data, equipment operation status data, and link status data, and the fault information includes fault occurrence time, fault type, and fault location information;

[0063] The data preprocessing module is configured to: based on the acquisition and extraction scheme, obtain the corresponding operation data and fault information, and perform preprocessing; wherein the preprocessing method includes data cleaning, data standardization and data normalization;

[0064] The fault location model generation module is configured as follows: based on the fault characteristics of the power communication network, a fault location model based on deep learning is generated; wherein the fault location model adopts an architecture combining a convolutional neural network and a recurrent neural network, and uses the acquired operation data and fault information to complete model training, and the fault location model is used for feature extraction and classification of faults in the power communication network, the feature extraction includes extracting network traffic characteristics, equipment operation characteristics and link characteristics related to the fault, and the classification is used to classify the fault into different types of communication faults;

[0065] The fault prediction and location module is configured as follows: Collect typical communication network operation and alarm data, input it into the trained fault location model, and perform network fault prediction and risk location;

[0066] The indicator system construction module is configured as follows: based on the communication network fault prediction and risk location, the fault probability and risk level are calculated, and the reliability operation indicator system of the power communication network is constructed. The reliability operation indicator system includes network connectivity indicators, network delay indicators, fault recovery time indicators and equipment failure rate indicators.

[0067] It should be noted that the deep learning-based power communication network fault location analysis system of the present embodiment corresponds to the aforementioned deep learning-based power communication network fault location analysis method. Therefore, the contents not specifically described in the deep learning-based power communication network fault location analysis system of the present embodiment may include but are not limited to function definitions, working principles, and technical effects, etc., and may refer to the records of the aforementioned deep learning-based power communication network fault location analysis method, and this text will not elaborate on them here.

[0068] In summary, the deep learning-based power communication network fault location analysis method and system of this embodiment, using the collection and extraction scheme based on power communication network data collection and fault location record generation, realizes comprehensive and targeted acquisition of operation data and fault information, and ensures data quality with the help of preprocessing methods including data cleaning, standardization and normalization; by constructing a fault location model that combines convolutional neural networks and recurrent neural networks, it realizes effective feature extraction and accurate classification of power communication network faults; it also uses the trained model to perform fault prediction and risk location, and calculates the probability of fault occurrence and risk level based on this, and finally realizes the construction of a reliable power communication network reliability operation indicator system to assist network operation and maintenance decision-making.

[0069] In the embodiments provided in the present application, it should be understood that the embodiments described herein can be implemented in hardware, software, firmware, middleware, code or any appropriate combination thereof. For hardware implementation, the processor can be implemented in one or more of the following units: application specific integrated circuit (ASIC), digital signal processor (DSP), digital signal processing device (DSPD), programmable logic device (PLD), field programmable gate array (FPGA), processor, controller, microcontroller, microprocessor, other electronic units designed to implement the functions described herein or their combination. For software implementation, part or all of the flow of the embodiment can be completed by instructing the relevant hardware through a computer program. When implemented, the above program can be stored in a computer-readable storage medium or transmitted as one or more instructions or codes on a computer-readable storage medium. Computer-readable storage media include computer storage media and communication media, wherein the communication medium includes any medium that is convenient for transmitting a computer program from one place to another. The storage medium can be any available medium that a computer can access. The computer-readable storage medium can include but is not limited to RAM, ROM, EEPROM, CD-ROM or other optical disk storage, disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of an instruction or data structure and can be accessed by a computer.

[0070] Finally, it should be noted that the above is only a preferred embodiment of the present application and is not intended to limit the present application. Although the present application has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for fault location analysis of power communication network based on deep learning, characterized in that: The method includes: S1: Based on the records of power communication network data collection and fault location, obtain the communication network fault location requirements based on deep learning, and generate a collection and extraction plan for operation data and fault information; wherein the operation data includes network flow data, equipment operation status data, and link status data, and the fault information includes fault occurrence time, fault type, and fault location information; S2: Based on the acquisition and extraction scheme, corresponding operation data and fault information are obtained and preprocessed; wherein the preprocessing method includes data cleaning, data standardization and data normalization; S3: Based on the fault characteristics of the power communication network, a deep learning-based fault location model is generated; wherein the fault location model adopts an architecture combining a convolutional neural network and a recurrent neural network, and uses the acquired operation data and fault information to complete model training. The fault location model is used for feature extraction and classification of faults in the power communication network, and the feature extraction includes extracting network traffic characteristics, equipment operation characteristics, and link characteristics related to the fault, and the classification is used to classify the fault into different types of communication faults; S3: Collect typical communication network operation and alarm data, input them into the trained fault location model, and perform network fault prediction and risk location; S4: Based on the communication network fault prediction and risk location, the fault occurrence probability and risk level are calculated, and a reliability operation index system for the power communication network is constructed. The reliability operation index system includes network connectivity index, network delay index, fault recovery time index and equipment failure rate index.

2. The method for locating and analyzing faults in a power communication network based on deep learning according to claim 1, characterized in that: The data cleaning step includes removing duplicate data, processing missing values ​​and processing outliers.

3. The method for locating and analyzing faults in a power communication network based on deep learning according to claim 1, characterized in that: The convolutional neural network is used to extract the spatial features of the communication network operation data, and the recurrent neural network is used to extract the time series features of the communication network operation data. The architecture combining the two performs feature fusion and classification through a fully connected layer.

4. The method for locating and analyzing faults in a power communication network based on deep learning according to claim 1, characterized in that: The model training adopts the back propagation algorithm to adjust the model parameters.

5. The method for locating and analyzing faults in a power communication network based on deep learning according to claim 1, characterized in that: The fault prediction is based on real-time monitoring and analysis of communication network operation data by a deep learning model. When the fault probability output by the model exceeds a preset fault probability threshold, it is determined that a fault may occur, and risk location is performed. The relationship between the fault characteristics and the network topology is analyzed using the risk location formula, and the network topology is represented as G = (V, E), where V is a node set and E is an edge set. The risk location formula is specifically: Among them, ω i is the weight of the i-th feature, node v∈V, x i is the i-th component of the fault feature vector x, v i is the i-th component of the eigenvector associated with node v, k(x i ,v i ) is the preset feature similarity measurement function, n is the number of features, P E is the proportion of faults in the edge set to which node v belongs, R(x,v) is the calculated risk level of node v, and when R(x,v)≥R τ (x,v), it is determined that node v may fail and its position is G=(V,E), R τ (x,v) is the failure probability threshold.

6. The method for locating and analyzing faults in a power communication network based on deep learning according to claim 1, characterized in that: In the network operation health assessment method, the probability of failure is calculated through the prediction results of the deep learning model, and the risk level is determined according to the probability of failure and the impact that the failure may cause.

7. The method for locating and analyzing faults in a power communication network based on deep learning according to claim 1, characterized in that: The network connectivity index in the power communication network reliability operation index system is determined by calculating the proportion of reachable nodes in the network, the network delay index is measured by the average delay time, the fault recovery time index records the time from the occurrence of the fault to the restoration of normal operation, and the equipment failure rate index counts the number of equipment failures per unit time.

8. The method for locating and analyzing faults in a power communication network based on deep learning according to claim 1, characterized in that: The method also includes generating a visualization report based on the network operation health assessment results, wherein the visualization report includes a network health status score, fault prediction information, risk location information, and a graphical display of reliability operation indicator data.

9. The method for locating and analyzing faults in a power communication network based on deep learning according to claim 1, characterized in that: The deep learning-based fault location model is updated and trained regularly. The updated training data includes newly added operating data and fault information as well as resampling of historical data.

10. A power communication network fault location analysis system based on deep learning, the system is applicable to the power communication network fault location analysis method based on deep learning as described in any one of claims 1 to 9, characterized in that: The system includes a collection and extraction scheme generation module, a data preprocessing module, a fault location model generation module, a fault prediction and location module and an indicator system construction module which are connected in sequence; The acquisition and extraction scheme generation module is configured to: based on the records of power communication network data acquisition and fault location, obtain the communication network fault location requirements based on deep learning, and generate an acquisition and extraction scheme for operation data and fault information; wherein the operation data includes network flow data, equipment operation status data, and link status data, and the fault information includes fault occurrence time, fault type, and fault location information; The data preprocessing module is configured to: based on the acquisition and extraction scheme, obtain corresponding operation data and fault information, and perform preprocessing; wherein the preprocessing method includes data cleaning, data standardization and data normalization; The fault location model generation module is configured to generate a deep learning-based fault location model based on the fault characteristics of the power communication network; wherein the fault location model adopts an architecture combining a convolutional neural network and a recurrent neural network, and uses the acquired operation data and fault information to complete model training; the fault location model is used for feature extraction and classification of faults in the power communication network, wherein the feature extraction includes extracting network traffic characteristics, equipment operation characteristics, and link characteristics related to the fault, and the classification is used to classify the fault into different types of communication faults; The fault prediction and location module is configured to: collect typical communication network operation and alarm data, input it into the trained fault location model, and perform network fault prediction and risk location; The indicator system construction module is configured as follows: based on communication network fault prediction and risk location, the fault probability and risk level are calculated to construct a reliability operation indicator system for the power communication network. The reliability operation indicator system includes network connectivity indicators, network delay indicators, fault recovery time indicators and equipment failure rate indicators.

Images