GIS partial discharge defect identification method and related device

Abstract

The invention belongs to the field of fault detection, and discloses a GIS partial discharge defect identification method and a related device, and the method comprises the steps: obtaining a GIS partial discharge signal through a terminal collection unit, and transmitting the GIS partial discharge signal to an edge calculation node; extracting features of the GIS partial discharge signal through the edge computing node to obtain a feature vector, and calling a partial discharge defect recognition model issued by the cloud computing center to obtain a defect recognition result; wherein the partial discharge defect identification model is obtained through the following steps: constructing a GIS digital twinborn body through a cloud computing center according to inherent attribute data and real-time operation data of a GIS, constructing a classifier model, and training based on the GIS digital twinborn body to obtain the partial discharge defect identification model. The problems that in an existing mode, the requirements for network bandwidth, transmission rate and data center processing capacity are high, time delay and stability of data are poor, and recognition timeliness and accuracy are low are solved, and the requirements for real-time performance, timeliness and accuracy of GIS partial discharge defect recognition are guaranteed.

Description

Technical Field

[0001] The present invention belongs to the field of fault detection and relates to a GIS partial discharge defect identification method and related devices. Background Art

[0002] GIS (gas-insulated switchgear) plays a key role in power systems, ensuring stable operation, reducing failure rates, and improving the grid's ability to respond to emergencies. Identifying GIS partial discharge defects is crucial to ensuring power system safety. Partial discharge can indicate an internal equipment failure. If not detected promptly, it can accelerate equipment aging and even cause serious accidents. Accurately identifying GIS partial discharge defects allows proactive maintenance measures to prevent accidents and ensure stable grid operation. Furthermore, effective GIS partial discharge defect identification helps optimize maintenance plans, reduce operation and maintenance costs, and extend GIS service life, significantly contributing to improving the overall reliability and economic efficiency of the power system.

[0003] Currently, when diagnosing and identifying partial discharge defects in GIS, PD sensors first capture the raw PD signals within the system. These signals are then transmitted to a data processing center for extraction, dimensionality reduction, identification, and diagnosis. However, this processing approach places significant demands on network bandwidth, transmission rates, and data center processing power, making it inadequate for simultaneous monitoring and joint analysis of multiple switchgear units. Relying on existing on-site network and computing configurations makes it difficult to meet the timeliness and accuracy requirements. Furthermore, the centralized processing of large amounts of fault recordings and image data to a single node can lead to excessive node load, poor data latency, and poor stability. Summary of the Invention

[0004] The purpose of the present invention is to overcome the shortcomings of the above-mentioned prior art and provide a GIS partial discharge defect identification method and related devices.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] In a first aspect, the present invention provides a GIS partial discharge defect identification method, comprising: obtaining a GIS partial discharge signal through a terminal acquisition unit and sending it to an edge computing node; extracting features of the GIS partial discharge signal through the edge computing node to obtain a feature vector and calling a partial discharge defect identification model issued by a cloud computing center to obtain a defect identification result; wherein the partial discharge defect identification model is obtained in the following manner: constructing a GIS digital twin based on the inherent attribute data and real-time operation data of the GIS through the cloud computing center, and constructing a classifier model and training it based on the GIS digital twin to obtain the partial discharge defect identification model.

[0007] Optionally, obtaining the GIS local discharge signal through the terminal acquisition unit includes: collecting the discharge amount q, discharge pulse repetition rate n, discharge phase φ and discharge waveform of the GIS local discharge through the terminal acquisition unit, and generating the GIS local discharge RGB spectrum, n-φ histogram, qmax-φ histogram, qave-φ histogram and discharge waveform diagram as the GIS local discharge signal; wherein qmax is the maximum value of the discharge amount q of the local discharge; qave is the average value of the discharge amount q of the local discharge.

[0008] Optionally, the method of extracting the characteristics of the GIS local discharge signal through the edge computing node to obtain a feature vector and calling the local discharge defect recognition model issued by the cloud computing center to obtain the defect recognition result includes: extracting the characteristics of the GIS local discharge signal through the edge computing node to obtain a feature vector, and using a feature space dimensionality reduction method based on local linear embedding to reduce the dimension to obtain a reduced dimensionality feature vector, and inputting the reduced dimensionality feature vector into the local discharge defect recognition model issued by the cloud computing center to obtain the defect recognition result.

[0009] Optionally, after obtaining the defect identification result, the method further includes: generating a warning signal according to the defect identification result, and sending the warning signal to a cloud computing center for visual display, and sending the warning signal to a terminal collection unit or GIS for visual display.

[0010] Optionally, constructing the classifier model includes: using a deep belief network to construct the classifier model; wherein the deep belief network is composed of a plurality of restricted Boltzmann machines and a classification network stacked in sequence.

[0011] Optionally, it also includes: obtaining historical feature vectors and defect identification results corresponding to the historical feature vectors and actual defect results through the cloud computing center as supplementary samples, and optimizing the partial discharge defect identification model based on the GIS digital twin and combined with the supplementary samples, obtaining the optimized partial discharge defect identification model and sending it to the edge computing node to update the current partial discharge defect identification model of the edge computing node.

[0012] In a second aspect, the present invention provides a GIS partial discharge defect identification device, comprising: a data acquisition module, used to obtain GIS partial discharge signals through a terminal acquisition unit and send them to an edge computing node; a defect identification module, used to extract features of the GIS partial discharge signals through the edge computing node, obtain feature vectors and call a partial discharge defect identification model issued by a cloud computing center to obtain defect identification results; wherein, the partial discharge defect identification model is obtained by: constructing a GIS digital twin based on the inherent attribute data and real-time operation data of the GIS through the cloud computing center, and constructing a classifier model and training it based on the GIS digital twin to obtain the partial discharge defect identification model.

[0013] Optionally, obtaining the GIS partial discharge signal through the terminal acquisition unit includes: collecting the discharge amount q, discharge pulse repetition rate n, discharge phase φ and discharge waveform of the GIS partial discharge through the terminal acquisition unit, and generating the GIS partial discharge RGB spectrum, n-φ histogram, qmax-φ histogram, qave-φ histogram and discharge waveform diagram as the GIS partial discharge signal.

[0014] Optionally, the method of extracting the characteristics of the GIS local discharge signal through the edge computing node to obtain a feature vector and calling the local discharge defect recognition model issued by the cloud computing center to obtain the defect recognition result includes: extracting the characteristics of the GIS local discharge signal through the edge computing node to obtain a feature vector, and using a feature space dimensionality reduction method based on local linear embedding to reduce the dimension to obtain a reduced dimensionality feature vector, and inputting the reduced dimensionality feature vector into the local discharge defect recognition model issued by the cloud computing center to obtain the defect recognition result.

[0015] Optionally, it also includes an early warning module, which is used to: generate an early warning signal based on the defect identification result, and send the early warning signal to the cloud computing center for visual display, and send the early warning signal to the terminal collection unit or GIS for visual display.

[0016] Optionally, constructing the classifier model includes: using a deep belief network to construct the classifier model; wherein the deep belief network is composed of a plurality of restricted Boltzmann machines and a classification network stacked in sequence.

[0017] Optionally, an optimization training module is also included, which is used to: obtain historical feature vectors and defect recognition results and actual defect results corresponding to the historical feature vectors through the cloud computing center and use them as supplementary samples, and optimize the partial discharge defect recognition model based on the GIS digital twin and combined with the supplementary samples to obtain the optimized partial discharge defect recognition model and send it to the edge computing node to update the current partial discharge defect recognition model of the edge computing node.

[0018] In a third aspect, the present invention provides a GIS partial discharge defect identification system, comprising a terminal acquisition unit, an edge computing node and a cloud computing center; the above-mentioned data acquisition module is set in the terminal acquisition unit; and the above-mentioned defect identification module is set in the edge computing node.

[0019] In a fourth aspect of the present invention, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned GIS partial discharge defect identification method when executing the computer program.

[0020] In a fifth aspect, the present invention provides a computer-readable storage medium storing a computer program, which implements the steps of the above-mentioned GIS partial discharge defect identification method when executed by a processor.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] The GIS partial discharge defect identification method of the present invention is based on a cloud-edge collaborative identification architecture of the terminal layer, edge layer, and cloud layer. At the terminal layer, a terminal acquisition unit acquires GIS partial discharge signals and sends them to an edge computing node. At the edge layer, the features of the GIS partial discharge signals are extracted to obtain feature vectors and call a partial discharge defect identification model issued by a cloud computing center to obtain defect identification results. The cloud layer then generates the partial discharge defect identification model. Based on the cloud-edge collaborative identification architecture, the excess data processing capacity of the edge layer is utilized to implement a distributed data processing technology that integrates network, computing, storage, and applications. The edge layer performs preliminary processing on the collected data at the edge of the network to achieve the diversion of massive data streams, effectively reducing network transmission pressure and data center processing pressure. This overcomes the existing problems of high requirements for network bandwidth, transmission rate, and data center processing capacity, poor data latency and stability, and low recognition timeliness. This alleviates the impact of large-scale multi-dimensional, heterogeneous data generated by massive devices on traditional equipment diagnosis models that rely on data centers, and ensures the real-time, timeliness, and accuracy requirements of GIS partial discharge defect identification. At the same time, a GIS digital twin is constructed in the cloud based on the inherent attribute data and real-time operation data of GIS, and a classifier model is trained based on the GIS digital twin to obtain a partial discharge defect recognition model, which accurately simulates the GIS operation status and realizes real-time monitoring and prediction of the GIS operation status. It overcomes the problems of difficult parameter determination and insufficient training of traditional shallow learning methods, and improves the accuracy and generalization ability of the recognition model. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is a flow chart of a GIS partial discharge defect identification method according to an embodiment of the present invention.

[0024] Figure 2 This is a structural block diagram of a GIS partial discharge defect identification device according to an embodiment of the present invention.

[0025] Figure 3 This is a detailed structural block diagram of the GIS partial discharge defect identification device according to an embodiment of the present invention.

[0026] Figure 4 Schematic diagram of the GIS partial discharge defect identification system architecture according to an embodiment of the present invention. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions 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 embodiments described 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 making creative efforts should fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "first", "second", etc. in the description 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 numbers 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 clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0029] The present invention is described in further detail below with reference to the accompanying drawings:

[0030] See also Figure 1 In one embodiment of the present invention, a GIS partial discharge defect identification method is provided, specifically a GIS partial discharge defect identification method based on cloud-edge collaboration and digital twins, which can achieve the accuracy and generalization ability of GIS partial discharge defect identification.

[0031] Specifically, the GIS partial discharge defect identification method of the present invention includes the following steps:

[0032] S1: Obtain GIS partial discharge signals through the terminal acquisition unit and send them to the edge computing node.

[0033] S2: Extract the features of GIS partial discharge signals through edge computing nodes, obtain feature vectors, and call the partial discharge defect recognition model issued by the cloud computing center to obtain defect recognition results.

[0034] Among them, the partial discharge defect recognition model is obtained by the following method: a GIS digital twin is constructed based on the inherent attribute data and real-time operation data of GIS through the cloud computing center, and a classifier model is constructed and trained based on the GIS digital twin to obtain the partial discharge defect recognition model.

[0035] The GIS partial discharge defect identification method of the present invention is based on a cloud-edge collaborative identification architecture of the terminal layer, edge layer, and cloud layer. At the terminal layer, a terminal acquisition unit acquires GIS partial discharge signals and sends them to an edge computing node. At the edge layer, the features of the GIS partial discharge signals are extracted to obtain feature vectors and call a partial discharge defect identification model issued by a cloud computing center to obtain defect identification results. The cloud layer then generates the partial discharge defect identification model. Based on the cloud-edge collaborative identification architecture, the excess data processing capacity of the edge layer is utilized to implement a distributed data processing technology that integrates network, computing, storage, and applications. The edge layer performs preliminary processing on the collected data at the edge of the network to achieve the diversion of massive data streams, effectively reducing network transmission pressure and data center processing pressure. This overcomes the existing problems of high requirements for network bandwidth, transmission rate, and data center processing capacity, poor data latency and stability, and low recognition timeliness. This alleviates the impact of large-scale multi-dimensional, heterogeneous data generated by massive devices on traditional equipment diagnosis models that rely on data centers, and ensures the real-time, timeliness, and accuracy requirements of GIS partial discharge defect identification. At the same time, a GIS digital twin is constructed in the cloud based on the inherent attribute data and real-time operation data of GIS, and a classifier model is trained based on the GIS digital twin to obtain a partial discharge defect recognition model, which accurately simulates the GIS operation status and realizes real-time monitoring and prediction of the GIS operation status. It overcomes the problems of difficult parameter determination and insufficient training of traditional shallow learning methods, and improves the accuracy and generalization ability of the recognition model.

[0036] To explain, cloud-edge collaboration refers to a distributed model based on the "cloud-edge-end" architecture, in which the "cloud" is the central node of traditional cloud computing and the control end of edge computing; the "edge" is the edge side of cloud computing, which is divided into infrastructure edge and device edge; the "end" is the terminal device. In GIS partial discharge defect identification, the end is the ultra-high frequency sensor and broadband pulse current sensor that obtains partial discharge signals.

[0037] Compared to traditional center-based computing models, cloud-edge collaboration adds an edge computing link between distributed sensors and traditional cloud computing centers. This leverages the excess data processing capabilities of edge-side data acquisition and communication equipment, enabling distributed data processing technology that integrates networking, computing, storage, and applications. Edge computing performs preliminary processing on collected data at the edge of the network, diverting massive data streams and effectively reducing network transmission pressure and data center processing pressure.

[0038] In a possible embodiment, obtaining a GIS partial discharge signal through a terminal acquisition unit includes: collecting the discharge amount q, discharge pulse repetition rate n, discharge phase φ, and discharge waveform of the GIS partial discharge through the terminal acquisition unit, and generating an RGB spectrum, an n-φ histogram, a qmax-φ histogram, a qave-φ histogram, and a discharge waveform diagram of the GIS partial discharge as the GIS partial discharge signal; wherein qmax is the maximum value of the discharge amount q of the partial discharge; and qave is the average value of the discharge amount q of the partial discharge.

[0039] Explanatory purposes, the terminal layer includes terminal equipment, namely the GIS, and terminal acquisition units. Terminal acquisition units are various types of sensors used to collect local discharge signals and real-time operating status in the GIS, such as optical sensors, photomultipliers, ultrasonic sensors, ultra-high frequency sensors, and broadband pulse current sensors. Acquisition units, including ultra-high frequency sensors and broadband pulse current sensors, are deployed on the on-site GIS to obtain signals such as the GIS's local discharge discharge quantity q, discharge pulse repetition rate n, discharge phase φ, and discharge waveform. These signals then form data such as the GIS's local discharge RGB spectrum, n-φ histogram, qmax-φ histogram, qave-φ histogram, and discharge waveform diagram, serving as the GIS local discharge signal. qmax is the maximum value of the local discharge discharge quantity q, and qave is the average value of the local discharge discharge quantity q. For example, one GIS corresponds to one terminal acquisition unit.

[0040] For example, after acquiring GIS partial discharge signals, the terminal acquisition unit connects to the edge computing node. Furthermore, to ensure data security and account for the impact of on-site electromagnetic interference, the GIS partial discharge defect identification architecture based on cloud-edge collaborative computing uses optical fiber for data transmission. Furthermore, signal transmission is unidirectional from the GIS to the terminal acquisition unit, meaning the terminal acquisition unit only acquires raw signal data from the GIS. Bidirectional data transmission is used between the terminal acquisition unit and the edge computing node.

[0041] In one possible implementation, the method of extracting the features of the GIS local discharge signal through the edge computing node to obtain a feature vector and calling the local discharge defect recognition model issued by the cloud computing center to obtain a defect recognition result includes: extracting the features of the GIS local discharge signal through the edge computing node to obtain a feature vector, and using a feature space dimensionality reduction method based on local linear embedding to reduce the dimension to obtain a reduced dimensionality feature vector, and inputting the reduced dimensionality feature vector into the local discharge defect recognition model issued by the cloud computing center to obtain a defect recognition result.

[0042] Explanatory note: The edge layer, located above the terminal layer, analyzes and processes collected GIS partial discharge signals. It also handles computing tasks assigned by the cloud computing center, offloading the massive computational workload of the data processing center. A single edge computing node can correspond to multiple terminal acquisition units, enabling feature extraction, dimensionality reduction, and in-situ recognition.

[0043] For example, in terms of signal transmission mode, there is bidirectional data transmission between the terminal layer and the edge layer. On the one hand, the edge computing node obtains the signal from the terminal acquisition unit for processing, and on the other hand, the edge computing node sends the analysis results or decision instructions to the terminal acquisition unit.

[0044] At the same time, there is bidirectional data transmission between the edge layer and the cloud layer. The edge computing nodes transmit the pre-processing results to the cloud computing center for centralized computing, and the cloud computing center sends the results of centralized computing or tasks that require edge computing nodes to be processed to the edge layer for processing.

[0045] Explanatory, the edge computing node performs feature extraction based on the GIS local discharge signal to obtain a multi-dimensional feature vector of the GIS local discharge signal. At the same time, the multi-dimensional feature vector of the GIS local discharge signal does not necessarily improve the accuracy of GIS local discharge defect recognition. On the contrary, the existence of redundant feature quantities will not only increase the amount of calculation and affect the recognition efficiency, but also the correlation between the feature quantities will affect the recognition process. Therefore, in this embodiment, a feature space dimensionality reduction method based on local linear embedding is used to reduce the dimensionality of the multi-dimensional feature vector of the GIS local discharge signal. Among them, local linear embedding dimensionality reduction is a nonlinear dimensionality reduction algorithm, which can well preserve the original manifold structure after dimensionality reduction, and the nonlinear dimensionality reduction method is more in line with the characteristics of the high-dimensional feature space of local discharge. The calculation process of local linear embedding dimensionality reduction is as follows:

[0046] Original sample space X={x1,x2,…,x N}, where x i is a sample vector with t-dimensional features, N represents the number of samples. Based on the KNN algorithm, each point x in X is obtainedi The k nearest neighbor points N(x i ) = {x i1 , x i2 , …, x ik}}. Assume that the sample x i can be linearly represented by each sample in N(x i ). Calculate the linear reconstruction error of the sample point x i , and use the mean square error as the error loss function:

[0047]

[0048] where, w ij is the coefficient of x j in the nearest neighbor points, and it needs to satisfy the normalization restriction, that is:

[0049]

[0050] Furthermore, when expanding w ij to the entire dataset, for samples that are not nearest neighbor points, set w ij to 0. Solve for the weight coefficient w ij when J(w) is minimized based on Lagrange multipliers. During the solution process, let: Z ij = (x i - x j ) T (x i - x k ), then the weight coefficient w ij is:

[0051]

[0052] In the low-dimensional space after dimensionality reduction, w​​​​​​​​​​​​​​​​​​​​​​​​​i ′)=tr(X′MX′ T )

[0056] Where tr(A) represents the trace of matrix A.

[0057] The above equation is calculated using the eigenvalue decomposition method, and the eigenvector corresponding to the smallest d non-zero eigenvalues ​​of M is taken as X′. The local linear embedding dimensionality reduction method has relatively low computational complexity and is easy to implement, making it suitable for dimensionality reduction in high-dimensional feature spaces in edge computing scenarios.

[0058] The edge computing node uses the reduced dimensionality feature vector as the input of the partial discharge defect recognition model for identification, and can feed back the defect recognition results to the cloud computing center for processing.

[0059] In a possible embodiment, after obtaining the defect identification result, the method further includes: generating a warning signal according to the defect identification result, sending the warning signal to a cloud computing center for visual display, and sending the warning signal to a terminal collection unit or GIS for visual display.

[0060] Explanation: The edge computing node can generate an early warning signal based on the defect identification results, and send the early warning signal to the cloud computing center for visual display, and send the early warning signal to the terminal collection unit or GIS for visual display. Alternatively, the defect identification results can be sent directly to the cloud computing center, which will generate the early warning signal and send it to the edge computing node, terminal collection unit, and GIS for visual display.

[0061] In a possible implementation, constructing the classifier model includes: constructing the classifier model using a deep belief network; wherein the deep belief network is formed by stacking a plurality of restricted Boltzmann machines and a classification network in sequence.

[0062] For example, the cloud computing center obtains the inherent attribute data and real-time operation data of GIS through edge computing nodes and terminal collection units, and then uses digital twin technology to build a GIS digital twin. By relying on the inherent attribute data and real-time operation data of GIS to build a GIS digital twin, the operating status of GIS can be accurately simulated, and real-time monitoring and prediction of the GIS operating status can be achieved, thereby improving the accuracy of fault diagnosis.

[0063] For illustrative purposes, this implementation utilizes a deep belief network-based classifier model trained on the GIS digital twin to generate a partial discharge defect recognition model. Typical types of GIS partial discharge defects include corona discharge, air gap discharge, creeping discharge, and suspended discharge. Therefore, the classifier model outputs five types: no defect, corona discharge, air gap discharge, creeping discharge, and suspended discharge.

[0064] Explanation: The deep belief network is composed of several restricted Boltzmann machines (RBM) and a classification network stacked in sequence. There is no connection between neurons in the same layer of RBM. The activation of neurons in the visible layer V and hidden layer H are independent of each other. Assume that the input of the visible layer unit is v = {v1,v2,…,v m}, the hidden layer unit output is h={h1,h2,…,h n}, network parameters θ={ω,a,b}.

[0065] The RBM energy function is defined as:

[0066]

[0067] Among them, ω ij Indicates a i with b j The connection weight between i represents the bias of the i-th node in the visible layer, b j represents the bias of the jth node in the hidden layer.

[0068] According to the definition of the energy function, the joint probability distribution of each node in the visible layer V and the hidden layer H is obtained as:

[0069]

[0070] When the Sigmoid function is used as the activation function, when the samples in the visible layer V (input samples) are known, calculate h j The probability of being 1 is:

[0071]

[0072] Similarly, when the hidden layer H (output result) is known, calculate v i The probability of being 1 is:

[0073]

[0074] During the training process, the objective function of RBM is:

[0075]

[0076] Among them, <.> P(h|v,θ) is the mathematical expectation of the probability distribution of the hidden layer when the input sample data is known, <.> P(v,h|θ) is the mathematical expectation of the joint probability distribution of the visible layer and the hidden layer.

[0077] The training process of RBM is the process of maximizing the objective function of RBM. In this embodiment, the training is performed by a sampling method based on the contrastive divergence algorithm. After the RBM layer is trained, a classification network (such as a classification network based on the softmax function) is used to implement the classification function.

[0078] In one possible implementation, the GIS partial discharge defect identification method further includes: obtaining historical feature vectors and defect identification results and actual defect results corresponding to the historical feature vectors through a cloud computing center as supplementary samples, and optimizing a partial discharge defect identification model based on a GIS digital twin and in combination with the supplementary samples, obtaining an optimized partial discharge defect identification model and sending it to an edge computing node to update the current partial discharge defect identification model of the edge computing node.

[0079] Explanatory note: On the one hand, the cloud computing center transmits the trained partial discharge defect recognition model to the edge computing node for on-site identification. On the other hand, it simultaneously receives the defect recognition results and corresponding warning signals from the on-site identification, and obtains the corresponding actual defect results and combines them into supplementary samples to expand the sample library used to train the partial discharge defect recognition model. Then, based on the GIS digital twin and combined with the supplementary samples, the partial discharge defect recognition model is regularly optimized. The optimized partial discharge defect recognition model is sent to the edge computing node to update the current partial discharge defect recognition model of the edge computing node, thereby ensuring the continued accuracy of the edge computing node in GIS partial discharge defect recognition.

[0080] The GIS partial discharge defect identification method of the present invention is based on a GIS partial discharge defect identification cloud-edge collaborative computing architecture that includes a terminal layer, an edge layer, and a cloud layer. Ultra-high frequency sensors and broadband pulse current sensors are used at the terminal layer to obtain GIS partial discharge signals. Feature extraction and feature dimensionality reduction are performed at the edge layer to construct feature spaces for different defect types. Furthermore, a digital twin is constructed by integrating GIS inherent attribute data and real-time operation data, and a partial discharge defect identification model based on a deep belief network is proposed. The digital twin construction and deep belief network training process are deployed at the cloud layer, and the deep belief network identification process is deployed at the edge layer. Efficient training and accurate reasoning are achieved based on the cloud-edge collaborative computing architecture.

[0081] The following are device embodiments of the present invention, which can be used to perform the method embodiments of the present invention. For details not disclosed in the device embodiments, please refer to the method embodiments of the present invention.

[0082] See also Figure 2In another embodiment of the present invention, a GIS partial discharge defect identification device is provided, which can be used to implement the above-mentioned GIS partial discharge defect identification method. Specifically, the GIS partial discharge defect identification device includes a data acquisition module and a defect identification module.

[0083] Among them, the data acquisition module is used to obtain GIS partial discharge signals through the terminal acquisition unit and send them to the edge computing node; the defect identification module is used to extract the characteristics of GIS partial discharge signals through the edge computing node, obtain the feature vector and call the partial discharge defect identification model issued by the cloud computing center to obtain the defect identification result; among them, the partial discharge defect identification model is obtained in the following way: the cloud computing center constructs a GIS digital twin based on the inherent attribute data and real-time operation data of the GIS, and constructs a classifier model and trains it based on the GIS digital twin to obtain the partial discharge defect identification model.

[0084] In a possible embodiment, obtaining the GIS partial discharge signal through the terminal acquisition unit includes: collecting the discharge amount q, discharge pulse repetition rate n, discharge phase φ and discharge waveform of the GIS partial discharge through the terminal acquisition unit, and generating an RGB spectrum, an n-φ histogram, a qmax-φ histogram, a qave-φ histogram and a discharge waveform diagram of the GIS partial discharge as the GIS partial discharge signal.

[0085] In one possible implementation, the method of extracting the features of the GIS local discharge signal through the edge computing node to obtain a feature vector and calling the local discharge defect recognition model issued by the cloud computing center to obtain a defect recognition result includes: extracting the features of the GIS local discharge signal through the edge computing node to obtain a feature vector, and using a feature space dimensionality reduction method based on local linear embedding to reduce the dimension to obtain a reduced dimensionality feature vector, and inputting the reduced dimensionality feature vector into the local discharge defect recognition model issued by the cloud computing center to obtain a defect recognition result.

[0086] In a possible embodiment, an early warning module is further included, which is used to: generate an early warning signal according to the defect identification result, and send the early warning signal to the cloud computing center for visual display, and send the early warning signal to the terminal collection unit or GIS for visual display.

[0087] In a possible implementation, constructing the classifier model includes: constructing the classifier model using a deep belief network; wherein the deep belief network is formed by stacking a plurality of restricted Boltzmann machines and a classification network in sequence.

[0088] In one possible implementation, an optimization training module is further included, which is used to: obtain historical feature vectors and defect identification results and actual defect results corresponding to the historical feature vectors through a cloud computing center and use them as supplementary samples; and optimize the partial discharge defect identification model based on the GIS digital twin and in combination with the supplementary samples to obtain an optimized partial discharge defect identification model and send it to the edge computing node to update the current partial discharge defect identification model of the edge computing node.

[0089] All relevant contents of each step involved in the embodiment of the aforementioned GIS partial discharge defect identification method can be referred to the functional description of the corresponding functional modules of the GIS partial discharge defect identification device in the embodiment of the present invention, and will not be repeated here.

[0090] In one possible implementation, see Figure 3 The defect recognition module is provided with a feature extraction module, a feature dimension reduction module, and an on-site recognition module. Among them, the feature extraction module is used to extract the features of the GIS partial discharge signal to obtain a feature vector, the feature dimension reduction module is used to reduce the dimension of the feature vector to obtain a reduced dimension feature vector, and the on-site recognition module is used to input the reduced dimension feature vector into the partial discharge defect recognition model issued by the cloud computing center to obtain a defect recognition result. The cloud computing center is provided with a digital twin construction module and a classifier construction module. The digital twin construction module is used to construct a GIS digital twin based on the inherent attribute data and real-time operation data of the GIS, and the classifier construction module is used to construct a classifier model and obtain a partial discharge defect recognition model based on the GIS digital twin training.

[0091] In another embodiment of the present invention, a GIS partial discharge defect identification system is provided, including a terminal acquisition unit, an edge computing node and a cloud computing center; the above-mentioned data acquisition module is set in the terminal acquisition unit; and the above-mentioned defect identification module is set in the edge computing node.

[0092] Explanatory, see Figure 4 The GIS partial discharge defect identification system is based on the cloud-edge collaborative identification architecture of the terminal layer, edge layer and cloud layer. The terminal layer is implemented based on GIS and terminal acquisition units, the edge layer is implemented based on edge computing nodes, and one edge computing node corresponds to several terminal acquisition units. The cloud layer is implemented based on the cloud computing center, and one cloud computing center corresponds to several edge computing nodes.

[0093] The module division in the embodiments of the present invention is illustrative and represents only one logical functional division. In actual implementation, other division methods may be used. Furthermore, the functional modules in various embodiments of the present invention may be integrated into a single processor, exist physically as separate modules, or two or more modules may be integrated into a single module. The integrated modules may be implemented in either hardware or software functional modules.

[0094] In another embodiment of the present invention, a computer device is provided, comprising a processor and a memory, wherein the memory is configured to store a computer program, the computer program including program instructions, and the processor is configured to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or may be another general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic device, discrete hardware component, etc. It is the computing core and control core of the terminal and is suitable for implementing one or more instructions, specifically loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding function. The processor described in this embodiment of the present invention can be used to operate the GIS partial discharge defect identification method.

[0095] In another embodiment of the present invention, a storage medium is provided, specifically a computer-readable storage medium (Memory). The computer-readable storage medium is a memory device within a computer device, used to store programs and data. It is understood that the computer-readable storage medium herein may include both built-in storage media within the computer device and, of course, extended storage media supported by the computer device. The computer-readable storage medium provides storage space, which stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for being loaded and executed by a processor. These instructions may be one or more computer programs (including program code). It should be noted that the computer-readable storage medium herein may be high-speed RAM memory or non-volatile memory, such as at least one disk drive. The processor may load and execute the one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the GIS partial discharge defect identification method described in the above-mentioned embodiment.

[0096] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0097] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0098] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0099] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A GIS partial discharge defect identification method, characterized in that: include: The GIS partial discharge signal is acquired through the terminal acquisition unit and sent to the edge computing node; The edge computing node extracts the characteristics of the GIS partial discharge signal, obtains the feature vector, and calls the partial discharge defect recognition model issued by the cloud computing center to obtain the defect recognition result. Among them, the partial discharge defect recognition model is obtained by the following method: a GIS digital twin is constructed based on the inherent attribute data and real-time operation data of GIS through the cloud computing center, and a classifier model is constructed and trained based on the GIS digital twin to obtain the partial discharge defect recognition model.

2. The GIS partial discharge defect identification method according to claim 1, characterized in that: The obtaining of GIS partial discharge signals by the terminal acquisition unit includes: The terminal acquisition unit collects the discharge amount q, discharge pulse repetition rate n, discharge phase φ and discharge waveform of the GIS partial discharge, and generates the GIS partial discharge RGB spectrum, n-φ histogram, qmax-φ histogram, qave-φ histogram and discharge waveform as the GIS partial discharge signal; among them, qmax is the maximum value of the partial discharge discharge amount q; qave is the average value of the partial discharge discharge amount q.

3. The GIS partial discharge defect identification method according to claim 1, characterized in that: The feature extraction of the GIS partial discharge signal through the edge computing node, the obtained feature vector and the call of the partial discharge defect recognition model issued by the cloud computing center to obtain the defect recognition results include: The features of GIS partial discharge signals are extracted through edge computing nodes to obtain feature vectors, and the feature space dimensionality reduction method based on local linear embedding is used to reduce the dimension to obtain reduced dimensionality feature vectors. The reduced dimensionality feature vectors are input into the partial discharge defect recognition model issued by the cloud computing center to obtain the defect recognition results.

4. The GIS partial discharge defect identification method according to claim 1, characterized in that: After obtaining the defect identification result, the following steps are further included: Generate early warning signals based on defect identification results, and send the early warning signals to the cloud computing center for visual display, and send the early warning signals to the terminal collection unit or GIS for visual display.

5. The GIS partial discharge defect identification method according to claim 1, characterized in that: Describing the construction of a classifier model includes: A deep belief network is used to construct a classifier model; the deep belief network is composed of several restricted Boltzmann machines and a classification network stacked in sequence.

6. The GIS partial discharge defect identification method according to claim 1, characterized in that: Also includes: Historical feature vectors, defect recognition results corresponding to the historical feature vectors, and actual defect results are obtained through the cloud computing center as supplementary samples. The partial discharge defect recognition model is optimized based on the GIS digital twin and combined with the supplementary samples to obtain the optimized partial discharge defect recognition model and send it to the edge computing node to update the current partial discharge defect recognition model of the edge computing node.

7. A GIS partial discharge defect identification device, characterized in that: include: The data acquisition module is used to obtain GIS partial discharge signals through the terminal acquisition unit and send them to the edge computing node; The defect recognition module is used to extract the characteristics of GIS partial discharge signals through edge computing nodes, obtain feature vectors, and call the partial discharge defect recognition model issued by the cloud computing center to obtain defect recognition results; Among them, the partial discharge defect recognition model is obtained by the following method: a GIS digital twin is constructed based on the inherent attribute data and real-time operation data of GIS through the cloud computing center, and a classifier model is constructed and trained based on the GIS digital twin to obtain the partial discharge defect recognition model.

8. The GIS partial discharge defect identification device according to claim 7, characterized in that: The obtaining of GIS partial discharge signals by the terminal acquisition unit includes: The terminal acquisition unit collects the discharge amount q, discharge pulse repetition rate n, discharge phase φ and discharge waveform of the GIS partial discharge, and generates the GIS partial discharge RGB spectrum, n-φ histogram, qmax-φ histogram, qave-φ histogram and discharge waveform as the GIS partial discharge signal.

9. The GIS partial discharge defect identification device according to claim 7, characterized in that: The feature extraction of the GIS partial discharge signal through the edge computing node, the obtained feature vector and the call of the partial discharge defect recognition model issued by the cloud computing center to obtain the defect recognition results include: The features of GIS partial discharge signals are extracted through edge computing nodes to obtain feature vectors, and the feature space dimensionality reduction method based on local linear embedding is used to reduce the dimension to obtain reduced dimensionality feature vectors. The reduced dimensionality feature vectors are input into the partial discharge defect recognition model issued by the cloud computing center to obtain the defect recognition results.

10. The GIS partial discharge defect identification device according to claim 7, characterized in that: It also includes an early warning module, which is used to: An early warning signal is generated based on the defect identification result, and the early warning signal is sent to the cloud computing center for visual display, and the early warning signal is sent to the terminal collection unit or GIS for visual display.

11. The GIS partial discharge defect identification device according to claim 7, characterized in that: The constructing of the classifier model comprises: A deep belief network is used to construct a classifier model; the deep belief network is composed of several restricted Boltzmann machines and a classification network stacked in sequence.

12. The GIS partial discharge defect identification device according to claim 7, characterized in that: It also includes an optimization training module, which is used to: Historical feature vectors, defect recognition results corresponding to the historical feature vectors, and actual defect results are obtained through the cloud computing center as supplementary samples. The partial discharge defect recognition model is optimized based on the GIS digital twin and combined with the supplementary samples to obtain the optimized partial discharge defect recognition model and send it to the edge computing node to update the current partial discharge defect recognition model of the edge computing node.

13. A GIS partial discharge defect identification system, characterized in that: It includes a terminal acquisition unit, an edge computing node and a cloud computing center; the data acquisition module described in claim 7 is set in the terminal acquisition unit; the defect identification module described in claim 7 is set in the edge computing node.

14. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the GIS partial discharge defect identification method according to any one of claims 1 to 6 are implemented.

15. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the GIS partial discharge defect identification method according to any one of claims 1 to 6 are implemented.

Images