Intelligent operation and maintenance method and device for electric power internet-of-things terminal, storage medium and program product

Abstract

The invention discloses an intelligent operation and maintenance method and device for an electric power Internet of Things terminal, a storage medium and a program product, and relates to the technical field of power grids, and the method comprises the steps: obtaining state detection data of terminal equipment and a knowledge graph generation model issued by a central server; training the knowledge graph generation model based on the state detection data, and obtaining loss function gradient data of the knowledge graph generation model in a training process; uploading the loss function gradient data to a central server to enable the central server to train a global knowledge graph generation model based on the loss function gradient data; and acquiring real-time operation data of the terminal equipment, and uploading the real-time operation data to the central server, so as to determine an abnormal state detection result corresponding to the real-time operation data and a state operation and maintenance knowledge graph of the power Internet of Things terminal through the global knowledge graph generation model. According to the invention, the accuracy of abnormal state detection of the electric power Internet of Things terminal is improved, and the intelligent operation and maintenance efficiency of the electric power Internet of Things terminal is improved.

Description

Technical Field

[0001] The present invention relates to the field of power grid technology, and in particular to an intelligent operation and maintenance method, device, storage medium and program product for a power Internet of Things terminal. Background Art

[0002] In the intelligent operation and maintenance system of the power network, the abnormal detection of the terminal state of the power Internet of Things is a key task. The traditional artificial intelligence method is restricted in the abnormal detection of the terminal state of the power Internet of Things due to the dispersion of data and high transmission costs. Federated learning has attracted much attention as a distributed, efficient and low-cost machine learning method. Federated learning allows data to be trained locally without centralizing the data, which not only reduces the transmission cost but also protects data privacy. However, the current federated learning still faces two major challenges in the application of power networks. First, the current federated learning still has the traditional supervised learning model driven by local data, which makes it difficult to effectively integrate the professional prior knowledge of intelligent operation and maintenance into the model training, resulting in the insufficient accuracy of the trained model in abnormal state detection. Second, the abnormal state detection results given by the model trained based on the supervised learning model are difficult to be directly combined with the abnormal state detection of the terminal equipment, and still require certain manual intervention, resulting in reduced terminal operation and maintenance efficiency. Therefore, improving the accuracy of abnormal state detection and improving operation and maintenance efficiency have become urgent problems to be solved in the field of power grid technology. Summary of the invention

[0003] The present invention provides a method, device, storage medium and program product for intelligent operation and maintenance of a power Internet of Things terminal, which improves the accuracy of abnormal state detection of the power Internet of Things terminal and enhances the operation and maintenance efficiency of the power Internet of Things terminal.

[0004] According to one aspect of an embodiment of the present invention, a method for intelligent operation and maintenance of an electric power Internet of Things terminal is provided, which is applied to a terminal status detection device. The method includes: obtaining status detection data of the terminal device and a knowledge graph generation model issued by a central server; training the knowledge graph generation model based on the status detection data, and obtaining loss function gradient data of the knowledge graph generation model during the training process; uploading the loss function gradient data to the central server so that the central server trains a global knowledge graph generation model based on the loss function gradient data; obtaining real-time operation data of the terminal device, and uploading it to the central server, so as to determine the abnormal status detection results corresponding to the real-time operation data and the electric power Internet of Things terminal status operation and maintenance knowledge graph through the global knowledge graph generation model.

[0005] Another aspect of an embodiment of the present invention provides an intelligent operation and maintenance method for a power IoT terminal, which is applied to a central server. The method includes:

[0006] Obtain the loss function gradient data uploaded by the terminal status detection device; perform gradient aggregation on the loss function gradient data to form aggregated gradient data; train the global knowledge graph generation model based on the aggregated gradient data to obtain the trained global knowledge graph generation model; obtain the real-time operation data uploaded by the terminal device; determine the abnormal status detection result corresponding to the real-time operation data based on the real-time operation data and the trained global knowledge graph generation model; construct the power Internet of Things terminal status operation and maintenance knowledge graph corresponding to the terminal device based on the real-time operation data and the abnormal status detection result.

[0007] In one aspect of an embodiment of the present invention, there is provided an intelligent operation and maintenance device for a power IoT terminal, which is applied to a terminal status detection device. The device includes:

[0008] The data acquisition module is used to obtain the status detection data of the terminal device and the knowledge graph generation model issued by the central server;

[0009] A gradient data determination module, used to train the knowledge graph generation model based on the state detection data, and obtain the loss function gradient data of the knowledge graph generation model during the training process;

[0010] A gradient data uploading module, used to upload the loss function gradient data to the central server, so that the central server can train a global knowledge graph generation model based on the loss function gradient data;

[0011] An operation data acquisition module is used to acquire the real-time operation data of the terminal device and upload it to the central server, so as to determine the abnormal state detection results corresponding to the real-time operation data and the power Internet of Things terminal state operation and maintenance knowledge graph through the global knowledge graph generation model.

[0012] Another aspect of the embodiment of the present invention provides an intelligent operation and maintenance device for a power IoT terminal, which is applied to a central server and includes:

[0013] A gradient data acquisition module is used to obtain the loss function gradient data uploaded by the terminal state detection device;

[0014] A gradient data aggregation module, used for performing gradient aggregation on the loss function gradient data to form aggregated gradient data;

[0015] A model training module, used to train a global knowledge graph generation model based on the aggregated gradient data to obtain a trained global knowledge graph generation model;

[0016] An operation data acquisition module, used to acquire the real-time operation data uploaded by the terminal status detection device;

[0017] A detection result determination module, used to determine an abnormal state detection result corresponding to the real-time operation data based on the real-time operation data and the trained global knowledge graph generation model;

[0018] A knowledge graph construction module is used to construct a power Internet of Things terminal status operation and maintenance knowledge graph corresponding to the terminal device based on the real-time operation data and the abnormal state detection results.

[0019] According to another aspect of the present invention, a computer-readable storage medium is provided, which stores computer instructions, and the computer instructions are used to enable a processor to implement the intelligent operation and maintenance method of a power Internet of Things terminal according to any embodiment of the present invention when executed.

[0020] According to another aspect of the present invention, a computer program product is provided. The computer program product includes a computer program. When the computer program is executed by a processor, the computer program implements the intelligent operation and maintenance method of the power Internet of Things terminal of any embodiment of the present invention.

[0021] In an embodiment of the present invention, status detection data reflecting the operating status of the terminal device is obtained, and a knowledge graph generation model transmitted by the central server is obtained. The status detection data can be used as a training set of the knowledge graph generation model to train the knowledge graph generation model. During the training process, the gradient of the loss function with respect to the knowledge graph generation model parameters is calculated, and the calculated gradient is used as loss function gradient data. The loss function gradient data is uploaded to the central server so that the central server can train the global knowledge graph generation model for constructing the terminal device knowledge graph based on the obtained loss function gradient data, obtain real-time operation data reflecting the operating status of the terminal device, and transmit the obtained real-time operation data to the central server. After obtaining the real-time operation data, the central server can determine the abnormal state detection result of the terminal device and the power Internet of Things terminal state operation and maintenance knowledge graph based on the real-time operation data.

[0022] In an embodiment of the present invention, by uploading the loss function gradient data to the central server, so that the central server can directly optimize the training of the global knowledge graph generation model based on the loss function gradient data, the training pertinence of the global knowledge graph generation model can be improved, and then the detection accuracy of the abnormal state of the terminal device can be improved. By causally associating the abnormal state detection result with the actual operating state corresponding to the real-time operating data of the terminal device to construct the power Internet of Things terminal state operation and maintenance knowledge graph, the direct relationship between the abnormal state and the actual operating state of the terminal device can be directly revealed, the judgment accuracy of the abnormal state detection of the terminal device is improved, and the false alarm rate and the missed alarm rate are reduced. By obtaining the real-time operating data of the terminal device and uploading the real-time operating data to the central server, the central server can construct the power Internet of Things terminal state operation and maintenance knowledge graph based on the real-time operating data and the abnormal state detection result corresponding to the real-time operating data. The power Internet of Things terminal state operation and maintenance knowledge graph can directly provide users with intuitive and comprehensive equipment operation and maintenance knowledge, reduce the cost of manual operation and maintenance, and thus improve the operation and maintenance efficiency of the terminal equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 This is a flow chart of an intelligent operation and maintenance method for a power IoT terminal applied to a terminal status detection device provided according to the first embodiment of the present invention;

[0025] Figure 2 This is a flow chart of an intelligent operation and maintenance method of a power Internet of Things terminal applied to a central server according to Embodiment 2 of the present invention;

[0026] Figure 3 This is a flow chart of an integrated intelligent operation and maintenance method for a power IoT terminal provided according to the second embodiment of the present invention;

[0027] Figure 4 is a flow chart of another method for intelligent operation and maintenance of a power IoT terminal applied to a terminal status detection device according to Embodiment 3 of the present invention;

[0028] Figure 5 This is a flow chart of another intelligent operation and maintenance method of a power Internet of Things terminal applied to a central server provided in accordance with a fourth embodiment of the present invention;

[0029] Figure 6It is a schematic diagram of the device structure of a method for intelligent operation and maintenance of a power Internet of Things terminal applied to a terminal status detection device provided in an embodiment of the present invention;

[0030] Figure 7 It is a schematic diagram of the device structure of an intelligent operation and maintenance method of a power Internet of Things terminal applied to a central server according to an embodiment of the present invention;

[0031] Figure 8 It is a structural schematic diagram of an electronic device for implementing the intelligent operation and maintenance method of a power Internet of Things terminal according to an embodiment of the present invention. DETAILED DESCRIPTION

[0032] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0033] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0034] Embodiment 1

[0035] Figure 1 This is a flow chart of a method for intelligent operation and maintenance of a power Internet of Things terminal applied to a terminal status detection device provided in an embodiment of the present invention. This embodiment can be applied to the case where a knowledge graph is used to detect abnormalities in a terminal device. The method can be executed by an intelligent operation and maintenance device for a power Internet of Things terminal. The intelligent operation and maintenance device for a power Internet of Things terminal can be implemented in the form of hardware and / or software. The intelligent operation and maintenance device for a power Internet of Things terminal can be configured in an electronic device. Figure 1 As shown, the method includes:

[0036] S101. Obtain the status detection data of the terminal device and the knowledge graph generation model issued by the central server.

[0037] In an embodiment of the present invention, the status detection data can be understood as information that reflects the operating status of the terminal device. For example, the status detection data may include: voltage information of the terminal device or temperature information value of the terminal device. The status detection data can be used as a training set to train a knowledge graph generation model. The status detection data can be abnormal status data of the terminal status historically collected by the terminal status detection device, or it can be public data currently available in the power network for abnormal status detection.

[0038] The knowledge graph generation model can be understood as an algorithm framework that can be used to generate abnormal state detection results of terminal devices based on input data. For example, the knowledge graph generation model includes but is not limited to a decision tree model, a support vector machine model, a random forest model, or a convolutional neural network model.

[0039] Specifically, status detection data reflecting the operating status of the terminal device is obtained, and a knowledge graph generation model transmitted by the central server is obtained. It can be understood that the knowledge graph generation model can be an untrained knowledge graph generation model or a knowledge graph generation model that has completed training and is waiting for updating and optimization.

[0040] S102. Train the knowledge graph generation model based on the state detection data, and obtain the loss function gradient data of the knowledge graph generation model during the training process.

[0041] In an embodiment of the present invention, the loss function gradient data can be understood as data used to guide the optimization of the knowledge graph generation model parameters. It can be understood that the loss function gradient data can be the gradient of the loss function, which is obtained by training the knowledge graph generation model based on the state detection data.

[0042] Specifically, the state detection data of the terminal device is obtained, and the state detection data can be used as a training set of the knowledge graph generation model to train the knowledge graph generation model. During the training process, the gradient of the loss function with respect to the knowledge graph generation model parameters is calculated, and the calculated gradient is used as the loss function gradient data, where the loss function can be understood as a type of data used to measure the difference between the model prediction and the actual result, and the gradient refers to the derivative of the loss function with respect to the model parameters.

[0043] For example, a method for obtaining loss function gradient data based on calculating the loss function may include: during the training process of the knowledge graph generation model, when the loss function meets the convergence condition, using the gradient of the corresponding loss function as the loss function gradient data.

[0044] S103. Upload the loss function gradient data to the central server so that the central server trains the global knowledge graph generation model based on the loss function gradient data.

[0045] In an embodiment of the present invention, the global knowledge graph generation model can be understood as another algorithm framework, which can be used to learn the relationship between entities based on input data and generate or update the operation and maintenance knowledge graph of the terminal device.

[0046] Specifically, the loss function gradient data is obtained and uploaded to the central server, so that the central server can train the global knowledge graph generation model used to construct the terminal device knowledge graph based on the obtained loss function gradient data. It can be understood that uploading the loss function gradient data to the central server and using the central server to construct the knowledge graph can reduce the computing and communication costs of the terminal state detection device.

[0047] For example, methods of uploading loss function gradient data to a central server may include: a terminal state detection device establishes a communication connection with the central server through a network communication protocol, and the terminal state detection device transmits the loss function gradient data to the central server through the communication connection, or the terminal state detection device and the central server are connected through a physical interface, and the terminal state detection device transmits the loss function gradient data to the central server through the physical interface.

[0048] S104. Obtain the real-time operation data of the terminal device and upload it to the central server, so as to determine the abnormal state detection results corresponding to the real-time operation data and the power Internet of Things terminal state operation and maintenance knowledge graph through the global knowledge graph generation model.

[0049] In an embodiment of the present invention, real-time operation data can be understood as information collected in real time that reflects the operating status of the terminal device. For example, the real-time operation data may include: current information of the terminal device or voltage information of the terminal device, etc., which is the data basis for constructing the knowledge graph of the terminal status operation and maintenance of the power Internet of Things.

[0050] The abnormal state detection result can be understood as result data used to reflect whether there is an abnormality in the terminal device. For example, the abnormal state detection result can include: an abnormal value of the terminal device or an abnormal state corresponding to the abnormal value of the terminal device.

[0051] The power IoT terminal status operation and maintenance knowledge graph can be understood as a knowledge base, which covers various entities, concepts and relationships between entities and concepts corresponding to the terminal devices. For example, entities may include: terminal device type or terminal device location, etc., and concepts may include: abnormal status of terminal equipment or abnormal cause of terminal equipment, etc.

[0052] Specifically, real-time operation data reflecting the operation status of the terminal device is acquired and transmitted to the central server, so that the central server can determine the abnormal state detection result of the terminal device and the power IoT terminal state operation and maintenance knowledge graph based on the real-time operation data after acquiring the real-time operation data. It can be understood that based on the knowledge graph, abnormal detection and analysis of the terminal device can be realized.

[0053] For example, the steps of determining abnormal state detection results of terminal equipment and the power Internet of Things terminal state operation and maintenance knowledge graph based on real-time operation data may include: the real-time operation data may be input into a trained global knowledge graph generation model to obtain abnormal state detection results, entities corresponding to the real-time operation data may be extracted from the real-time operation data, abnormal state relationships may be determined based on the real-time operation data, state attribution relationships between real-time operation data and entities may be determined based on the abnormal state detection results, and a power Internet of Things terminal state operation and maintenance knowledge graph may be constructed based on entities, state attribution relationships and abnormal state relationships.

[0054] In an embodiment of the present invention, status detection data reflecting the operating status of a terminal device is obtained, and a knowledge graph generation model transmitted by a central server is obtained. The status detection data can be used as a training set of the knowledge graph generation model to train the knowledge graph generation model. During the training process, the gradient of the loss function with respect to the parameters of the knowledge graph generation model is calculated, and the calculated gradient is used as loss function gradient data. The loss function gradient data is uploaded to the central server so that the central server can train a global knowledge graph generation model for constructing a knowledge graph of the terminal device based on the obtained loss function gradient data, obtain real-time operating data reflecting the operating status of the terminal device, and transmit the obtained real-time operating data to the central server. After obtaining the real-time operating data, the central server can determine the abnormal status detection result of the terminal device and the power Internet of Things terminal status operation and maintenance knowledge graph based on the real-time operating data. In an embodiment of the present invention, by uploading the loss function gradient data to the central server, so that the central server can directly optimize the training of the global knowledge graph generation model based on the loss function gradient data, the training pertinence of the global knowledge graph generation model can be improved, and then the detection accuracy of the abnormal state of the terminal device can be improved. By causally associating the abnormal state detection result with the actual operating state corresponding to the real-time operating data of the terminal device to construct the power Internet of Things terminal state operation and maintenance knowledge graph, the direct relationship between the abnormal state and the actual operating state of the terminal device can be directly revealed, the judgment accuracy of the abnormal state detection of the terminal device is improved, and the false alarm rate and the missed alarm rate are reduced. By obtaining the real-time operating data of the terminal device and uploading the real-time operating data to the central server, the central server can construct the power Internet of Things terminal state operation and maintenance knowledge graph based on the real-time operating data and the abnormal state detection result corresponding to the real-time operating data. The power Internet of Things terminal state operation and maintenance knowledge graph can directly provide users with intuitive and comprehensive equipment operation and maintenance knowledge, reduce the cost of manual operation and maintenance, and thus improve the operation and maintenance efficiency of the terminal equipment.

[0055] Embodiment 2

[0056] Figure 2 The present invention provides a flow chart of an intelligent operation and maintenance method for a power IoT terminal applied to a central server. Based on the above-mentioned embodiment, the present invention provides an intelligent operation and maintenance method for a power IoT terminal applied to a central server. Figure 2 As shown, the method includes:

[0057] S201. Obtain loss function gradient data uploaded by the terminal state detection device.

[0058] Specifically, the central server can obtain the loss function gradient data uploaded by the terminal state detection device. After the central server obtains the loss function gradient data, it can process the obtained loss function gradient data to perform subsequent operations.

[0059] S202, performing gradient aggregation on the loss function gradient data to form aggregated gradient data.

[0060] In the embodiment of the present invention, gradient aggregation can be understood as a data processing method for merging data. For example, the operations included in gradient aggregation may include: weighted average operation or gradient compression operation, etc.

[0061] Aggregated gradient data can be understood as the result data after gradient aggregation of loss function gradient data, which is used to guide the training of the global knowledge graph generation model.

[0062] Specifically, the loss function gradient data is obtained, and gradient aggregation processing is performed on the obtained loss function gradient data to obtain aggregated gradient data for guiding the training of the global knowledge graph generation model.

[0063] It is understandable that the source of the loss function gradient data acquired by the central server may include: loss function gradient data from different training batches in the terminal state detection device or loss function gradient data from different computing nodes in the terminal state detection device.

[0064] S203. Train the global knowledge graph generation model based on the aggregated gradient data to obtain a trained global knowledge graph generation model.

[0065] Specifically, aggregated gradient data can be obtained and used to directly guide the updating of model parameters of the global knowledge graph generation model, or during the training process of the global knowledge graph generation model, the aggregated gradient data can be used to assist in adjusting the model parameters of the global knowledge graph generation model. When the model parameters of the global knowledge graph generation model are adjusted so that the global knowledge graph generation model reaches the preset goals, the training of the global knowledge graph generation model can be stopped to obtain a global knowledge graph generation model that has completed training, and subsequent operations can be performed based on the global knowledge graph generation model that has completed training.

[0066] For example, the preset goal may include: a preset performance indicator or reaching a preset time, etc.

[0067] S204: Acquire real-time operation data uploaded by the terminal status detection device.

[0068] Specifically, the central server may obtain the real-time operation data collected in real time and uploaded by the terminal status detection device, and may process the obtained real-time operation data to perform subsequent operations.

[0069] S205. Determine the abnormal state detection result corresponding to the real-time operation data based on the real-time operation data and the trained global knowledge graph generation model.

[0070] Specifically, the real-time operation data uploaded by the terminal device is obtained, and the real-time operation data uploaded by the terminal device is input into the global knowledge graph generation model that has been trained, and the potential abnormal data contained in the real-time operation data is intelligently detected using the global knowledge graph generation model that has been trained, and the detection result is output, and the detection result is used as the abnormal state detection result of the real-time operation data this time.

[0071] For example, the abnormal state detection result may include: an abnormal value of a terminal device, an abnormal state to which the abnormal value of the terminal device belongs, a terminal device corresponding to the abnormal state, or a terminal state detection device corresponding to the abnormal state, etc.

[0072] S206. Construct a power IoT terminal status operation and maintenance knowledge graph corresponding to the terminal device based on real-time operation data and abnormal status detection results.

[0073] Specifically, real-time operation data and abnormal state detection results are obtained. Based on the obtained real-time operation data and abnormal state detection results, the trained global knowledge graph generation model can be used to construct a power Internet of Things terminal state operation and maintenance knowledge graph, which can be used to guide the operation and maintenance of terminal equipment.

[0074] For example, the steps of constructing a power Internet of Things terminal status operation and maintenance knowledge graph using a trained global knowledge graph generation model may include: extracting entities from real-time operation data, determining the attribution relationship between the real-time operation data and each entity based on the abnormal state detection results, determining the abnormal state relationship of the real-time operation data, and constructing a power Internet of Things terminal status operation and maintenance knowledge graph based on entities, attribution relationships and abnormal state relationships using the trained global knowledge graph generation model.

[0075] In an embodiment of the present invention, the central server can obtain the loss function gradient data uploaded by the terminal state detection device, perform gradient aggregation processing on the obtained loss function gradient data, obtain aggregated gradient data for guiding the training of the global knowledge graph generation model, input the obtained aggregated gradient data into the global knowledge graph generation model, and the global knowledge graph generation model can perform iterative learning based on the aggregated gradient data. When the global knowledge graph generation model reaches a preset goal, the training of the global knowledge graph generation model can be stopped to obtain a global knowledge graph generation model that has completed training. The central server can obtain the loss function gradient data uploaded by the terminal state detection device, perform gradient aggregation processing on the obtained loss function gradient data, obtain aggregated gradient data for guiding the training of the global knowledge graph generation model, and input the acquired aggregated gradient data into the global knowledge graph generation model. The global knowledge graph generation model can perform iterative learning based on the aggregated gradient data. When the global knowledge graph generation model reaches a preset goal, the training of the global knowledge graph generation model can be stopped to obtain a global knowledge graph generation model that has completed training. The real-time operation data collected in real time is transmitted, and the real-time operation data uploaded by the terminal device is input into the global knowledge graph generation model that has been trained. The potential abnormal data contained in the real-time operation data is intelligently detected by using the global knowledge graph generation model that has been trained, and the detection result is output. The detection result is used as the abnormal state detection result of the real-time operation data this time. Based on the acquired real-time operation data and the abnormal state detection result, the trained global knowledge graph generation model can be used to construct a power Internet of Things terminal state operation and maintenance knowledge graph, and the operation and maintenance of the terminal equipment can be guided based on the power Internet of Things terminal state operation and maintenance knowledge graph. In an embodiment of the present invention, the loss function gradient data is aggregated by a central server, and the aggregated gradient data obtained after aggregation is used to train the global knowledge graph generation model, which can improve the construction efficiency and pertinence of the global knowledge graph generation model, improve the efficiency of data processing and the accuracy of detection, and based on real-time operation data and abnormal state detection results, the central server uses the trained global knowledge graph generation model to construct a power Internet of Things terminal state operation and maintenance knowledge graph. The power Internet of Things terminal state operation and maintenance knowledge graph contains the causal relationship between the abnormal state detection results and the actual operation state of the terminal device. Based on the causal relationship, the direct relationship between the abnormal state and the actual operation state of the terminal device can be directly revealed, thereby improving the judgment accuracy of abnormal state detection of the terminal device and reducing the false alarm rate and missed alarm rate of the detection. The power Internet of Things terminal state operation and maintenance knowledge graph provides a scientific guiding basis for the operation and maintenance of the terminal device, thereby improving the operation and maintenance efficiency of the terminal device.

[0076] It can be understood that based on the above steps S101, S102, S103, S104, S201, S202, S203, S204, S205 and S206, an integrated intelligent operation and maintenance method of power Internet of Things terminals based on federated knowledge graph can also be constructed, which covers the complete operation and maintenance process from the central server to the terminal status detection device, such as Figure 3As shown, the method includes: the central server can establish a connection with multiple terminal state detection devices, wherein the terminal state detection device and the central server can respectively serve as the client of federated learning and the server end of the federated learning center aggregation, wherein the central server can be an independent server or a server cluster composed of multiple servers. It can be understood that under the federated learning framework, there can be a central server and several terminal state detection devices; each terminal state detection device can obtain the state detection data of multiple terminal devices separately, and each terminal state detection device can obtain the knowledge graph generation model sent by the central server through the connection established with the central server, and use the knowledge graph generation model sent by the central server as the local knowledge graph generation model. It can be understood that the local knowledge graph generation model used by each terminal state detection device is the same, and each terminal state detection device can train the local knowledge graph generation model based on local private data. At the terminal state detection device end, the local knowledge graph generation model is trained using the state detection data of the terminal device obtained to obtain the loss function gradient data. The calculation formula of the loss function gradient data can be: Among them, in the formula, x t,i and t,i denote the local data and true label of the i-th terminal state detection device in the t-th iteration, F(·) denotes the output of the knowledge graph generation model, and W tGenerate the weight of the model for the local knowledge graph, l represents the loss function between the label prediction value output by the model and the true label, the loss function l is calculated by the distance between the label prediction value output by the model and the true label, and the loss function gradient data is obtained through derivation operation. The whole process of calculating the loss function gradient data will be continuously iterated until the learning stops when the loss function meets the convergence condition. The obtained loss function gradient data is uploaded to the central server. After receiving the loss function gradient data, the central server performs gradient aggregation on the loss function gradient data to obtain aggregated gradient data. The central server trains the global knowledge graph generation model based on the aggregated gradient data. After completing the training, the central server can use the trained global knowledge graph generation model to process new state detection data. At the terminal state detection device end, the state detection data of the terminal device is collected in real time to obtain new state detection data. The new state detection data is used as the real-time operation data of the terminal device, and the obtained real-time operation data is uploaded to the central server. After obtaining the real-time operation data, the central server inputs the real-time operation data into the trained global knowledge graph generation model. The trained global knowledge graph generation model processes and analyzes the real-time operation data to obtain the abnormal state detection result corresponding to the real-time operation data. At the central server end, Entities corresponding to the terminal devices are extracted from the real-time operation data, and the state attribution relationship between the real-time operation data of the terminal devices and each entity is determined based on the abnormal state detection result. The abnormal state relationship corresponding to the real-time operation data of the terminal devices is determined according to the spatiotemporal characteristics, background characteristics and handshake characteristics of the real-time operation data of the terminal devices. The determined entities, state attribution relationships and abnormal state relationships are used to construct a triple V = (ε, R, τ), where ε represents an entity set; R represents a relationship set between entities, and τ represents a triple set consisting of entities and relationships. The triple V constructed above is prone to conflict in the description of the same entity due to the uncertainty and incompleteness of various information in the power network system. Therefore, alignment processing and fusion processing are required to integrate the data in the triple V, and the triple V after alignment and fusion processing is used to construct a knowledge graph of power Internet of Things terminal state operation and maintenance, which can be stored in the Neptune graph database. In an embodiment of the present invention, the terminal status detection device can also receive the aggregated gradient data sent by the central server, and then perform the next round of update training on the model parameters of the local knowledge graph generation model according to the aggregated gradient data until the global knowledge graph generation model training of the central server is completed.

[0077] Embodiment 3

[0078] Figure 4This is another flow chart of an intelligent operation and maintenance method for a power IoT terminal applied to a terminal status detection device provided by an embodiment of the present invention. Based on the above embodiment, an intelligent operation and maintenance method for a power IoT terminal applied to a terminal status detection device is provided by an embodiment of the present invention. Figure 4 As shown, the method includes:

[0079] S301. Obtain the knowledge graph generation model sent by the central server.

[0080] S302. Collect status detection data of terminal devices based on edge intelligence technology, and use the status detection data as a training set for the knowledge graph generation model.

[0081] In an embodiment of the present invention, edge intelligence technology can be understood as a data acquisition technology, which is a technology that combines artificial intelligence with edge computing and is used to collect status detection data of terminal devices.

[0082] The training set can be understood as a set of data sets used to train the model. The state detection data can be used as the training set for training the knowledge graph generation model.

[0083] Specifically, the collected state detection data can be used as a training set for training the knowledge graph generation model to train the knowledge graph generation model.

[0084] For example, in the process of training the knowledge graph generation model by using the state detection data as the training set of the knowledge graph generation model, the knowledge graph generation model can optimize its own model parameters by learning the effective information in the training set.

[0085] S303: Preprocess the training set to obtain a preprocessed data set, wherein the preprocessing at least includes: format conversion.

[0086] In an embodiment of the present invention, preprocessing refers to a series of processing operations performed on the state detection data before inputting the state detection data into the knowledge graph generation model. For example, the preprocessing operations may include: data cleaning operations, format conversion operations, or feature selection operations, etc.

[0087] The preprocessed dataset refers to the data set obtained after preprocessing the state detection data. The preprocessed dataset is used as the input of the knowledge graph generation model and can be used for model training.

[0088] Format conversion refers to the process of converting data from one format to another to ensure the compatibility of data in model training.

[0089] Specifically, the state detection data is used as a training set for training the knowledge graph generation model. Before the state detection data is input into the knowledge graph generation model and the knowledge graph generation model is trained, the state detection data can be preprocessed to obtain a preprocessed data set, wherein the operation of preprocessing the state detection data at least includes: converting the data format of the state detection data into the format of the knowledge graph generation model input data.

[0090] For example, the step of converting the data format of the state detection data into the format of the input data of the knowledge graph generation model may include: using the Pandas library in Python to convert the image into the DataFrame format, and then converting the data in the training set into a black-and-white image of 28*28 pixels, and packaging the above black-and-white image into an IDX format data set. The IDX format data set can be used as the input of the knowledge graph generation model to train the knowledge graph generation model.

[0091] S304: Train the knowledge graph generation model based on the gradient descent training rules and the preprocessed data set to obtain the loss function gradient data.

[0092] In an embodiment of the present invention, the gradient descent training rule can be understood as an algorithm for optimizing the knowledge graph generation model. By way of example, the gradient descent training rule may include: a federated stochastic gradient descent algorithm or a batch gradient descent algorithm.

[0093] Specifically, a preprocessed data set of state detection data is obtained, and the preprocessed data set is input into the knowledge graph generation model. The gradient descent training rule used to optimize the knowledge graph generation model can be used to train the knowledge graph generation model issued by the central server. When the loss function corresponding to the graph generation model meets the preset conditions, the corresponding loss function gradient can be used as the loss function gradient data.

[0094] For example, the steps of using the federated stochastic gradient descent algorithm to train the knowledge graph generation model sent by the central server may include: the server may distribute the knowledge graph generation model to all terminal state detection devices participating in the training; after each terminal state detection device receives the knowledge graph generation model, it uses the federated stochastic gradient descent algorithm to train the knowledge graph generation model based on its own training set; during the training process, the loss function is calculated, and the gradient of the loss function is calculated; after the terminal state detection device completes the local training, the loss function gradient can be obtained.

[0095] By way of example, the preset conditions may include: the loss function satisfies a convergence condition or the loss function satisfies a specified number of iterations condition.

[0096] S305. Upload the loss function gradient data to the central server so that the central server trains the global knowledge graph generation model based on the loss function gradient data.

[0097] S306. Obtain the real-time operation data of the terminal device and upload it to the central server to determine the abnormal state detection results corresponding to the real-time operation data and the power Internet of Things terminal state operation and maintenance knowledge graph through the global knowledge graph generation model.

[0098] In an embodiment of the present invention, a knowledge graph generation model transmitted by a central server is obtained. The state detection data of the terminal device can be collected based on edge intelligence technology to obtain the state detection data of the terminal device, and the collected state detection data is used as a training set for training the knowledge graph generation model. Before the training set is input into the knowledge graph generation model to train the knowledge graph generation model, the state detection data needs to be preprocessed to obtain a preprocessed data set, wherein the operation of preprocessing the state detection data at least includes: converting the data format of the state detection data into the format of the input data of the knowledge graph generation model, and inputting the preprocessed data set into the knowledge graph generation model. In the model, the gradient descent training rule for optimizing the knowledge graph generation model can be used to train the knowledge graph generation model sent by the central server. When the loss function corresponding to the graph generation model meets the preset conditions, the corresponding loss function gradient can be used as the loss function gradient data, and the loss function gradient data can be uploaded to the central server, so that the central server can train the global knowledge graph generation model based on the loss function gradient data, obtain the real-time operation data of the terminal device, and upload it to the central server, so that the central server can determine the abnormal state detection results corresponding to the real-time operation data and the power Internet of Things terminal state operation and maintenance knowledge graph through the global knowledge graph generation model. In an embodiment of the present invention, through edge intelligence technology, the status detection data of the terminal device can be efficiently collected, the data transmission delay can be reduced, the efficiency of data processing can be improved, and the status detection data of the terminal device can be used as a training set for training the knowledge graph generation model, thereby enhancing the practicality and accuracy of the knowledge graph generation model. By preprocessing the training set, the data format input into the knowledge graph generation model is ensured to be unified, the quality of the training set is improved, and the efficiency and accuracy of the model training are thereby improved. The model parameters are continuously iterated and updated through the gradient descent training rule, gradually approaching the optimal solution, thereby improving the detection accuracy of the knowledge graph generation model.

[0099] Embodiment 4

[0100] Figure 5 This is another flow chart of an intelligent operation and maintenance method for a power IoT terminal applied to a central server provided by an embodiment of the present invention. Based on the above embodiment, an intelligent operation and maintenance method for a power IoT terminal applied to a central server is provided by an embodiment of the present invention. Figure 5 As shown, the method includes:

[0101] S401. Obtain loss function gradient data uploaded by the terminal state detection device.

[0102] S402, performing gradient aggregation on the loss function gradient data to form aggregated gradient data.

[0103] S403. Train the global knowledge graph generation model based on the aggregated gradient data to obtain a trained global knowledge graph generation model.

[0104] S404: Acquire real-time operation data uploaded by the terminal status detection device.

[0105] S405. Determine the abnormal state detection result corresponding to the real-time operation data based on the real-time operation data and the trained global knowledge graph generation model, wherein the abnormal state detection result includes at least: the abnormal value, the abnormal state to which the abnormal value belongs, the terminal device corresponding to the abnormal state, and the terminal state detection device corresponding to the abnormal state.

[0106] In the embodiment of the present invention, an abnormal value may be understood as data beyond a normal range. For example, the causes of abnormal values ​​in real-time operation data may include: equipment failure or environmental changes.

[0107] The abnormal state refers to an abnormal operating state of the terminal device. For example, the abnormal state may include: an overheated state of the terminal device or an overloaded state of the terminal device.

[0108] Specifically, the real-time operation data is input into the trained global knowledge graph generation model, and the trained global knowledge graph generation model is used to perform intelligent detection on the potential abnormal data contained in the real-time operation data, at least detecting and determining abnormal values ​​that exceed the normal range and the abnormal states corresponding to the abnormal values, the terminal devices corresponding to the abnormal states, and the terminal state detection devices corresponding to the abnormal states in the real-time operation data, and taking the detection result containing at least the abnormal values ​​and the abnormal states corresponding to the abnormal values, the terminal devices corresponding to the abnormal states, and the terminal state detection devices corresponding to the abnormal states as the abnormal state detection result.

[0109] S406: Extract entities corresponding to the real-time operation data from the real-time operation data, where the entities at least include: a terminal status detection device and a terminal device.

[0110] In the embodiment of the present invention, an entity may be understood as an actually existing identifiable object. For example, the entity extracted from the real-time operation data at least includes: a terminal status detection device and a terminal device.

[0111] Specifically, the real-time operation data uploaded by the terminal status detection device is obtained, the actually existing identifiable objects are extracted from the obtained real-time operation data, and the actually existing identifiable objects are used as the entities corresponding to the real-time operation data, wherein the entity includes at least: the terminal status detection device and the terminal device.

[0112] S407: Determine the state attribution relationship between the real-time operation data and the entity based on the abnormal state detection result.

[0113] In the embodiment of the present invention, the state attribution relationship can be understood as an attribution relationship between real-time operation data and an entity, that is, the real-time operation data specifically belongs to which operation and maintenance terminal state detection device, or specifically belongs to which terminal device.

[0114] Specifically, the status attribution relationship between the real-time operation data of the terminal device and each entity can be determined according to the abnormal status detection result of the real-time operation data of the terminal device, that is, based on the abnormal status detection result, it can be determined which terminal status detection device and terminal device the real-time operation data uploaded by the terminal status detection device specifically belongs to.

[0115] S408: Determine the abnormal state relationship corresponding to the real-time operation data based on the spatiotemporal characteristics, background characteristics, and handshake characteristics of the real-time operation data.

[0116] In the embodiment of the present invention, the spatiotemporal feature refers to the attribute information of the real-time operation data in the time dimension and the space dimension. For example, the spatiotemporal feature may include information such as the timestamp of the real-time operation data or the position of the timestamp in the network topology.

[0117] Background features refer to additional information related to real-time operation data. For example, background features may include: historical maintenance records of terminal devices corresponding to the real-time operation data or the operation environment of the real-time operation data.

[0118] The handshake feature refers to the interactive information used to confirm and establish a connection during the transmission of real-time operation data. For example, the handshake feature may include: handshake request information, handshake response information or handshake confirmation information.

[0119] The abnormal state relationship can be understood as the association relationship between the real-time operation data and the feature in the real-time operation data. For example, the abnormal state relationship can include: normal state or abnormal state, etc.

[0120] Specifically, relevant feature information is extracted from the real-time operation data, wherein the feature information at least includes: spatiotemporal features, background features, handshake features and other information. Based on the spatiotemporal features, background features, handshake features and other information extracted from the real-time operation data, an algorithm can be applied to determine the abnormal state relationship corresponding to the real-time operation data, that is, to determine whether the real-time state detection data of the terminal device has an abnormal state tendency, wherein the algorithm may include: a machine learning algorithm or a deep learning algorithm, etc.

[0121] S409: construct a triple data set using entities, state attribution relationships, and abnormal state relationships, and align and fuse the triple data sets to obtain a comprehensive data set.

[0122] In the embodiment of the present invention, the triple data set refers to a data set constructed by entities, state attribution relationships, and abnormal state relationships, and is used to represent the relationships between different entities.

[0123] Alignment and fusion processing can be understood as a data processing method for aligning and integrating entity and attribute information. For example, alignment and fusion processing technologies may include: alignment processing technology and fusion processing technology. Alignment processing technology may include: string matching technology or word vector matching technology, etc. Fusion processing technology may include: attribute value matching technology or rule matching technology, etc.

[0124] The comprehensive dataset refers to the triple dataset that has been aligned and fused, and is used to construct the knowledge graph of power IoT terminal status operation and maintenance.

[0125] Specifically, based on the entities, state attribution relationships and abnormal state relationships determined based on real-time operation data and abnormal state detection results, a triple data set used to represent the relationship between different entities can be constructed, and the triple data set is further aligned and fused, and the triple data set after the alignment and fusion processing is used as a comprehensive data set.

[0126] For example, the reasons for continuing to perform alignment and fusion processing on the triple data set may include: there may be conflicting descriptions of the same entity in the constructed triple data set. To avoid the occurrence of conflicting descriptions, alignment and fusion processing can be performed on the triple data set, and the triple data set after alignment and fusion processing can be used as a comprehensive data set.

[0127] S410. Construct a knowledge graph of the power IoT terminal status operation and maintenance of the terminal device based on the comprehensive data set.

[0128] Specifically, the entities in the comprehensive data set can be converted into nodes, and the relationships can be converted into edges, and the corresponding power Internet of Things terminal status operation and maintenance knowledge graph can be constructed in the form of point-edge.

[0129] For example, the real-time operation data based on the terminal status, terminal equipment and abnormal status detection results are visualized in the form of points and edges, which can improve the convenience of subsequent query on the power Internet of Things terminal status operation and maintenance knowledge graph.

[0130] For example, the Neptune graph database can be used to store the power IoT terminal status operation and maintenance knowledge graph.

[0131] In an embodiment of the present invention, loss function gradient data uploaded by a terminal state detection device is obtained, the loss function gradient data is gradient aggregated to form aggregated gradient data, a global knowledge graph generation model is trained based on the aggregated gradient data to obtain a trained global knowledge graph generation model, real-time operation data uploaded by the terminal state detection device is obtained, the real-time operation data is input into the trained global knowledge graph generation model, and the trained global knowledge graph generation model is used to perform intelligent detection on potential abnormal data contained in the real-time operation data, at least abnormal values ​​that exceed the normal range and abnormal states corresponding to the abnormal values, terminal devices corresponding to the abnormal states, and terminal state detection devices corresponding to the abnormal states are detected and determined in the real-time operation data, and a detection result containing at least the abnormal values ​​and the abnormal states corresponding to the abnormal values, the terminal devices corresponding to the abnormal states, and the terminal state detection devices corresponding to the abnormal states are used as abnormal state detection results, and the actually existing identifiable pairs are extracted from the acquired real-time operation data. The above-mentioned actually existing identifiable objects are used as entities corresponding to the real-time operation data. The state attribution relationship between the real-time operation data of the terminal device and each entity can be determined according to the abnormal state detection result of the real-time operation data of the terminal device, and relevant feature information is extracted from the real-time operation data, wherein the feature information at least includes: spatiotemporal features, background features, handshake features and other information. According to the spatiotemporal features, background features, handshake features and other information extracted from the real-time operation data, an algorithm can be applied to determine the abnormal state relationship corresponding to the real-time operation data. Based on the entities, state attribution relationships and abnormal state relationships determined by the real-time operation data and the abnormal state detection results, a triple data set for representing the relationship between different entities can be constructed, and the triple data set is further aligned and fused. The triple data set after the alignment and fusion processing is used as a comprehensive data set. The entities in the comprehensive data set can be converted into nodes, and the relationships can be converted into edges, and the corresponding power Internet of Things terminal state operation and maintenance knowledge graph can be constructed in the form of points and edges. In an embodiment of the present invention, by constructing a power Internet of Things terminal status operation and maintenance knowledge graph based on entities, status attribution relationships and abnormal status relationships, the power Internet of Things terminal status operation and maintenance knowledge graph can provide operation and maintenance personnel with intuitive and comprehensive equipment status information, which can effectively improve the operation and maintenance efficiency of terminal equipment.

[0132] Based on the above embodiments, embodiments of the present invention also include: receiving an abnormal state detection query request triggered by a user; calling a preset entity recognition model to extract a target entity in the abnormal state detection query request; querying the target entity in the knowledge graph based on the target entity for an abnormal state relationship, and feeding back the abnormal state relationship to the user.

[0133] In an embodiment of the present invention, an abnormal state detection query request can be understood as a signal for requesting an abnormal state detection query. By way of example, the source of the abnormal state detection query request may include: a user manually issuing an abnormal state detection query request through an interactive interface or a user triggering an abnormal state detection query request through a voice command, etc.

[0134] The preset entity recognition model can be understood as an algorithm framework for automatically identifying and extracting target entities from abnormal state detection query requests triggered by users.

[0135] The target entity refers to a specific object that the user wishes to query through an abnormal state detection query request. For example, the target entity may include: a terminal device or a terminal state detection device, etc.

[0136] Specifically, an abnormal state detection query request triggered by a target user such as a power operation and maintenance personnel is received, and then a pre-configured preset entity recognition model is called to extract the target entity from the abnormal state detection query request, and then based on the target entity, the edge associated with the target entity is retrieved in the power Internet of Things terminal state operation and maintenance knowledge graph, and the edge represents the target entity corresponding to the abnormal state relationship, and finally the association relationship associated with the target entity obtained from the power Internet of Things terminal state operation and maintenance knowledge graph is returned to the corresponding target user.

[0137] For example, in an embodiment of the present invention, a question-and-answer system can be established to use the power Internet of Things terminal status operation and maintenance knowledge graph for natural language question-and-answer tasks, that is, the structured knowledge stored in the power Internet of Things terminal status operation and maintenance knowledge graph is utilized, combined with natural language understanding technology and reasoning technology, to provide users with knowledge graph-based question-and-answer services. The above-mentioned question-and-answer system can be responsible for receiving abnormal status detection query requests and performing subsequent operations based on the abnormal status detection query requests.

[0138] Embodiment 5

[0139] Figure 6 The present invention provides a schematic diagram of a structure of a power IoT terminal intelligent operation and maintenance device applied to a terminal status detection device. Figure 6 As shown, the device comprises:

[0140] The data acquisition module 501 is used to acquire the status detection data of the terminal device and the knowledge graph generation model issued by the central server;

[0141] A gradient data determination module 502 is used to train the knowledge graph generation model based on the state detection data and obtain the loss function gradient data of the knowledge graph generation model during the training process;

[0142] A gradient data uploading module 503 is used to upload the loss function gradient data to the central server so that the central server can train the global knowledge graph generation model based on the loss function gradient data;

[0143] The operation data acquisition module 504 is used to obtain the real-time operation data of the terminal device and upload it to the central server to determine the abnormal state detection results corresponding to the real-time operation data and the power Internet of Things terminal state operation and maintenance knowledge graph through the global knowledge graph generation model.

[0144] On the basis of the above embodiments, in an embodiment of the present invention, the abnormal state detection result at least includes: an abnormal value, an abnormal state to which the abnormal value belongs, a terminal device corresponding to the abnormal state, and a terminal state detection device corresponding to the abnormal state.

[0145] Based on the above embodiments, embodiments of the present invention also include: receiving an abnormal state detection query request triggered by a user; calling a preset entity recognition model to extract a target entity in the abnormal state detection query request; querying the target entity in the knowledge graph based on the target entity for an abnormal state relationship, and feeding back the abnormal state relationship to the user.

[0146] Based on the above embodiment, the data acquisition module 501 is specifically used to collect status detection data of terminal devices based on edge intelligence technology.

[0147] The gradient data determination module 502 is specifically used to use the state detection data as a training set for the knowledge graph generation model; preprocess the training set to obtain a preprocessed data set, wherein the preprocessing at least includes: format conversion; training the knowledge graph generation model to be trained based on the gradient descent training rules and the preprocessed data set to obtain the loss function gradient data.

[0148] In an embodiment of the present invention, status detection data reflecting the operating status of a terminal device is obtained, and a knowledge graph generation model transmitted by a central server is obtained. The status detection data can be used as a training set of the knowledge graph generation model to train the knowledge graph generation model. During the training process, the gradient of the loss function with respect to the parameters of the knowledge graph generation model is calculated, and the calculated gradient is used as loss function gradient data. The loss function gradient data is uploaded to the central server so that the central server can train a global knowledge graph generation model for constructing a knowledge graph of the terminal device based on the obtained loss function gradient data, obtain real-time operating data reflecting the operating status of the terminal device, and transmit the obtained real-time operating data to the central server. After obtaining the real-time operating data, the central server can determine the abnormal status detection result of the terminal device and the power Internet of Things terminal status operation and maintenance knowledge graph based on the real-time operating data. In an embodiment of the present invention, by uploading the loss function gradient data to the central server, so that the central server can directly optimize the training of the global knowledge graph generation model based on the loss function gradient data, the training pertinence of the global knowledge graph generation model can be improved, and then the detection accuracy of the abnormal state of the terminal device can be improved. By causally associating the abnormal state detection result with the actual operating state corresponding to the real-time operating data of the terminal device to construct the power Internet of Things terminal state operation and maintenance knowledge graph, the direct relationship between the abnormal state and the actual operating state of the terminal device can be directly revealed, the judgment accuracy of the abnormal state detection of the terminal device is improved, and the false alarm rate and the missed alarm rate are reduced. By obtaining the real-time operating data of the terminal device and uploading the real-time operating data to the central server, the central server can construct the power Internet of Things terminal state operation and maintenance knowledge graph based on the real-time operating data and the abnormal state detection result corresponding to the real-time operating data. The power Internet of Things terminal state operation and maintenance knowledge graph can directly provide users with intuitive and comprehensive equipment operation and maintenance knowledge, reduce the cost of manual operation and maintenance, and thus improve the operation and maintenance efficiency of the terminal equipment.

[0149] Embodiment 6

[0150] Figure 7 The present invention provides a schematic diagram of a structure of an intelligent operation and maintenance device for a power IoT terminal applied to a central server. Figure 7 As shown, the device comprises:

[0151] A gradient data acquisition module 601 is used to acquire the loss function gradient data uploaded by the terminal state detection device;

[0152] A gradient data aggregation module 602 is used to perform gradient aggregation on the loss function gradient data to form aggregated gradient data;

[0153] A model training module 603 is used to train the global knowledge graph generation model based on the aggregated gradient data to obtain a trained global knowledge graph generation model;

[0154] The operation data acquisition module 604 is used to acquire the real-time operation data uploaded by the terminal device;

[0155] A detection result determination module 605 is used to determine an abnormal state detection result corresponding to the real-time operation data based on the real-time operation data and the trained global knowledge graph generation model;

[0156] The knowledge graph construction module 606 is used to construct the knowledge graph corresponding to the terminal device based on the real-time operation data and abnormal state detection results.

[0157] In an embodiment of the present invention, the central server can obtain the loss function gradient data uploaded by the terminal state detection device, perform gradient aggregation processing on the obtained loss function gradient data, obtain aggregated gradient data for guiding the training of the global knowledge graph generation model, input the obtained aggregated gradient data into the global knowledge graph generation model, and the global knowledge graph generation model can perform iterative learning based on the aggregated gradient data. When the global knowledge graph generation model reaches a preset goal, the training of the global knowledge graph generation model can be stopped to obtain a global knowledge graph generation model that has completed training. The central server can obtain the loss function gradient data uploaded by the terminal state detection device, perform gradient aggregation processing on the obtained loss function gradient data, obtain aggregated gradient data for guiding the training of the global knowledge graph generation model, and input the acquired aggregated gradient data into the global knowledge graph generation model. The global knowledge graph generation model can perform iterative learning based on the aggregated gradient data. When the global knowledge graph generation model reaches a preset goal, the training of the global knowledge graph generation model can be stopped to obtain a global knowledge graph generation model that has completed training. The real-time operation data collected in real time is transmitted, and the real-time operation data uploaded by the terminal device is input into the global knowledge graph generation model that has been trained. The potential abnormal data contained in the real-time operation data is intelligently detected by using the global knowledge graph generation model that has been trained, and the detection result is output. The detection result is used as the abnormal state detection result of the real-time operation data this time. Based on the acquired real-time operation data and the abnormal state detection result, the trained global knowledge graph generation model can be used to construct a power Internet of Things terminal state operation and maintenance knowledge graph, and the operation and maintenance of the terminal equipment can be guided based on the power Internet of Things terminal state operation and maintenance knowledge graph. In an embodiment of the present invention, the loss function gradient data is aggregated by a central server, and the aggregated gradient data obtained after aggregation is used to train the global knowledge graph generation model, which can improve the construction efficiency and pertinence of the global knowledge graph generation model, improve the efficiency of data processing and the accuracy of detection, and based on real-time operation data and abnormal state detection results, the central server uses the trained global knowledge graph generation model to construct a power Internet of Things terminal state operation and maintenance knowledge graph. The power Internet of Things terminal state operation and maintenance knowledge graph contains the causal relationship between the abnormal state detection results and the actual operation state of the terminal device. Based on the causal relationship, the direct relationship between the abnormal state and the actual operation state of the terminal device can be directly revealed, thereby improving the judgment accuracy of abnormal state detection of the terminal device and reducing the false alarm rate and missed alarm rate of the detection. The power Internet of Things terminal state operation and maintenance knowledge graph provides a scientific guiding basis for the operation and maintenance of the terminal device, thereby improving the operation and maintenance efficiency of the terminal device.

[0158] The knowledge graph construction module 606 is specifically used to extract entities corresponding to the real-time operation data from the real-time operation data, wherein the entities include at least: terminal status detection equipment and terminal equipment; determine the status attribution relationship between the real-time operation data and the entities based on the abnormal status detection results; determine the abnormal status relationship corresponding to the real-time operation measurement data based on the spatiotemporal characteristics, background characteristics and handshake characteristics of the real-time operation data; construct a triple data set using entities, status attribution relationships and abnormal status relationships, and align and fuse the entities and attribute information of the entities in the triple data set to obtain a comprehensive data set; and construct a power Internet of Things terminal status operation and maintenance knowledge graph for terminal equipment based on the comprehensive data set.

[0159] Embodiment 7

[0160] Figure 8 A schematic diagram of an electronic device that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown in the embodiments of the present invention, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments of the present invention described and / or required herein.

[0161] like Figure 8 As shown, the electronic device includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the ROM 12 or the computer program loaded from the storage unit 18 to the RAM 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. The input / output (I / O) interface 15 is also connected to the bus 14.

[0162] A number of components in the electronic device are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0163] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit, a graphics processing unit, various special artificial intelligence computing chips, various processors running machine learning model algorithms, digital signal processors, and any appropriate processors, controllers, microcontrollers, etc. The processor 11 executes the various methods and processes described above, for example, the intelligent operation and maintenance method of the power Internet of Things terminal.

[0164] In some embodiments, the intelligent operation and maintenance method of the power Internet of Things terminal may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the intelligent operation and maintenance method of the power Internet of Things terminal may be performed. Alternatively, in other embodiments, the processor 11 may be configured as the intelligent operation and maintenance method of the power Internet of Things terminal in any other appropriate manner (for example, by means of firmware).

[0165] Various implementations of the systems and techniques described above in the embodiments of the present invention may be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays, application specific integrated circuits, application specific standard products, systems on a chip, load programmable logic devices, computer hardware, firmware, software, and / or combinations thereof. These various implementations may include: being implemented in one or more computer programs, which may be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general programmable processor, which may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0166] The computer programs for implementing the methods of the embodiments of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer programs are executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer programs may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0167] In the context of an embodiment of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a RAM, a ROM, an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0168] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a cathode ray tube or a liquid crystal display monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including: acoustic input, voice input, or tactile input).

[0169] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local area networks, wide area networks, blockchain networks, and the Internet.

[0170] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and virtual private server services.

[0171] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0172] The above specific implementations do not constitute a limitation on the protection scope of the embodiments of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent operation and maintenance method for a power IoT terminal, applied to a terminal status detection device, characterized in that: The method comprises: Obtain the status detection data of the terminal device and the knowledge graph generation model issued by the central server; Training the knowledge graph generation model based on the state detection data, and obtaining loss function gradient data of the knowledge graph generation model during the training process; Uploading the loss function gradient data to the central server so that the central server trains a global knowledge graph generation model based on the loss function gradient data; The real-time operation data of the terminal device is obtained and uploaded to the central server, so as to determine the abnormal state detection result corresponding to the real-time operation data and the power Internet of Things terminal state operation and maintenance knowledge graph through the global knowledge graph generation model.

2. The method according to claim 1, characterized in that The step of training the knowledge graph generation model based on the state detection data and obtaining loss function gradient data of the knowledge graph generation model during the training process includes: Collecting the status detection data of the terminal device based on edge intelligence technology, and using the status detection data as a training set for the knowledge graph generation model; Preprocessing the training set to obtain a preprocessed data set, wherein the preprocessing at least includes: format conversion; The knowledge graph generation model is trained based on the gradient descent training rule and the preprocessed data set to obtain the loss function gradient data.

3. An intelligent operation and maintenance method for power IoT terminals, applied to a central server, characterized in that: The method comprises: Obtain the loss function gradient data uploaded by the terminal status detection device; Gradient aggregate the loss function gradient data to form aggregated gradient data; Training a global knowledge graph generation model based on the aggregated gradient data to obtain a trained global knowledge graph generation model; Acquire the real-time operation data uploaded by the terminal status detection device; Determine an abnormal state detection result corresponding to the real-time operation data based on the real-time operation data and the trained global knowledge graph generation model; Based on the real-time operation data and the abnormal state detection results, a power Internet of Things terminal state operation and maintenance knowledge graph corresponding to the terminal device is constructed.

4. The method according to claim 3, characterized in that The constructing of the power IoT terminal state operation and maintenance knowledge graph corresponding to the terminal device based on the real-time operation data and the abnormal state detection result includes: Extracting an entity corresponding to the real-time operation data from the real-time operation data, wherein the entity at least includes: the terminal state detection device and the terminal device; Determine a state attribution relationship between the real-time operation data and the entity based on the abnormal state detection result; Determine the abnormal state relationship corresponding to the real-time operation data based on the spatiotemporal characteristics, background characteristics and handshake characteristics of the real-time operation data; Constructing a triple data set using the entity, the state attribution relationship and the abnormal state relationship, and aligning and fusing the triple data set to obtain a comprehensive data set; The power Internet of Things terminal status operation and maintenance knowledge graph of the terminal device is constructed based on the comprehensive data set.

5. The method according to claim 3, characterized in that: The abnormal state detection result includes at least: an abnormal value, an abnormal state to which the abnormal value belongs, a terminal device corresponding to the abnormal state, and a terminal state detection device corresponding to the abnormal state.

6. The method according to claims 3-5, characterized in that: The method further comprises: Receive abnormal status detection query requests triggered by users; Calling a preset entity recognition model to extract the target entity in the abnormal state detection query request; Based on the target entity, the abnormal state relationship corresponding to the target entity is queried in the knowledge graph, and the abnormal state relationship is fed back to the user.

7. An intelligent operation and maintenance device for a power IoT terminal, applied to a terminal status detection device, characterized in that: The device comprises: The data acquisition module is used to obtain the status detection data of the terminal device and the knowledge graph generation model issued by the central server; A gradient data determination module, used to train the knowledge graph generation model based on the state detection data, and obtain the loss function gradient data of the knowledge graph generation model during the training process; A gradient data uploading module, used to upload the loss function gradient data to the central server, so that the central server can train a global knowledge graph generation model based on the loss function gradient data; An operation data acquisition module is used to acquire the real-time operation data of the terminal device and upload it to the central server, so as to determine the abnormal state detection results corresponding to the real-time operation data and the power Internet of Things terminal state operation and maintenance knowledge graph through the global knowledge graph generation model.

8. An intelligent operation and maintenance device for power IoT terminals, applied to a central server, characterized in that: The device comprises: A gradient data acquisition module is used to obtain the loss function gradient data uploaded by the terminal state detection device; A gradient data aggregation module, used for performing gradient aggregation on the loss function gradient data to form aggregated gradient data; A model training module, used to train a global knowledge graph generation model based on the aggregated gradient data to obtain a trained global knowledge graph generation model; An operation data acquisition module, used to acquire the real-time operation data uploaded by the terminal status detection device; A detection result determination module, used to determine an abnormal state detection result corresponding to the real-time operation data based on the real-time operation data and the trained global knowledge graph generation model; A knowledge graph construction module is used to construct a power Internet of Things terminal status operation and maintenance knowledge graph corresponding to the terminal device based on the real-time operation data and the abnormal state detection results.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the intelligent operation and maintenance method for a power Internet of Things terminal according to any one of claims 1 to 6 when executed.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the computer program implements the intelligent operation and maintenance method of the power Internet of Things terminal according to any one of claims 1 to 6.

Images