A capacitive voltage transformer operation and maintenance decision method and system

Abstract

The present invention relates to a capacitor voltage transformer operation and maintenance decision method and system, the method comprising the following steps: collecting historical operation data of the capacitor voltage transformer and performing preprocessing; building a multi-task model, training the multi-task model through the preprocessed historical operation data of the capacitor voltage transformer in the whole life cycle, completing the remaining life prediction task and fault classification task of the capacitor voltage transformer through the trained multi-task model; based on the remaining life and fault category of the capacitor voltage transformer output by the multi-task model, performing operation and maintenance decision on the capacitor voltage transformer.

Description

Technical Field

[0001] The invention relates to a capacitor voltage transformer operation and maintenance decision method and system, belonging to the technical field of power transformer operation and maintenance. Background Art

[0002] Capacitor voltage transformers (CVTs) have been widely used in 110kV and above power projects due to their obvious advantages such as good insulation performance, low manufacturing cost, and anti-ferromagnetic resonance. Traditional transformer maintenance methods are usually based on preventive maintenance at fixed cycles or corrective maintenance after obvious faults occur, and there is a certain degree of subjectivity in manual decision-making on operation and maintenance plans. This method may either lead to excessive maintenance and increase unnecessary costs; or it may cause equipment failure due to failure to detect potential problems in a timely manner, affecting the stable operation of the power system.

[0003] The invention patent application with publication number "CN114626769A" discloses a capacitor voltage transformer operation and maintenance method and system. The patent integrates multiple indicators such as stability, importance, and fault degree of the capacitor voltage transformer, and calculates the operation and maintenance coefficient and determines the operation and maintenance strategy based on them, which effectively guides the operation and maintenance management of the capacitor voltage transformer, greatly reduces the blindness of operation and maintenance, and saves human resource costs.

[0004] However, the above solution does not take into account the life of the capacitive voltage transformer. For long-term operation and maintenance management, remaining life prediction is crucial, which can help operation and maintenance personnel make long-term plans and avoid unnecessary downtime and excessive maintenance. Summary of the invention

[0005] In order to solve the above problems existing in the prior art, the present invention proposes a capacitor voltage transformer operation and maintenance decision method and system.

[0006] The technical solution of the present invention is as follows:

[0007] On the one hand, the present invention provides a capacitor voltage transformer operation and maintenance decision method, comprising the following steps:

[0008] Collect historical operation data of capacitor voltage transformer and perform preprocessing;

[0009] Constructing a multi-task model, the multi-task model includes a coding layer, a shared feature extraction layer, a sub-task feature extraction layer and an output layer, training the multi-task model through preprocessed historical operation data of the capacitor voltage transformer in the whole life cycle, and completing the remaining life prediction task and fault classification task of the capacitor voltage transformer through the trained multi-task model;

[0010] Based on the remaining life and fault category of the capacitor voltage transformer output by the multi-task model, operation and maintenance decisions are made for the capacitor voltage transformer.

[0011] As a preferred embodiment of the present invention, the historical operation data includes: capacitor voltage transformer secondary voltage, environmental parameter data, capacitor voltage transformer remaining life, capacitor voltage transformer historical error evaluation data and its corresponding fault category.

[0012] As a preferred embodiment of the present invention, the preprocessing step of the historical operation data is: normalizing the continuous data and labeling the category data.

[0013] As a preferred embodiment of the present invention, the encoding layer is used to vector encode the preprocessed historical operation data, and concatenate the encoded data to form a comprehensive input vector , as shown in the following formula:

[0014] ;

[0015] in: Represents the historical operation data after preprocessing; Indicates encoding operation; is the first vector value, Pick ; Represents the total length of the synthesized input vector.

[0016] As a preferred embodiment of the present invention, the shared feature extraction layer is used to extract the comprehensive input vector For the shared features related to both the remaining life prediction task and the fault classification task, the specific steps are as follows:

[0017] Constructing the synthetic input vector The shared feature extraction unit mapped to its corresponding eigenvalue performs shared feature extraction, as shown in the following formula:

[0018] ;

[0019] in: For shared features; is a shared feature extraction unit; is the feature mapping parameter;

[0020] The vector-encoded capacitor voltage transformer secondary voltage and environmental parameter data in the preprocessed historical operation data are used as the input vector of the remaining life prediction task. , and Shared features The similarity calculation is performed to obtain the shared feature allocation weights for the remaining life prediction task, as shown in the following formula:

[0021] ;

[0022] ;

[0023] in: Indicates the secondary voltage of the capacitor voltage transformer; Indicates environmental parameter data; Represents the input vector of the remaining life prediction task No. vector value, Pick ; express Total length; Assign weights to the shared features for the remaining life prediction task;

[0024] The total input features of the remaining life prediction task are calculated based on the shared features of the remaining life prediction task, as shown in the following formula:

[0025] ;

[0026] in: represents the total input features for the remaining life prediction task;

[0027] The historical error evaluation data of capacitor voltage transformer after vector encoding in the preprocessed historical operation data is used as the input vector of the fault classification task , and Shared features Perform similarity calculation to obtain the shared feature allocation weights for the fault classification task, as shown in the following formula:

[0028] ;

[0029] ;

[0030] in: Represents the ratio difference data in the historical error evaluation data of the capacitor voltage transformer; Represents the angle difference data in the historical error evaluation data of the capacitor voltage transformer; Represents the input vector of the fault classification task No. vector value, Pick ; express Total length; Assign weights to shared features for the fault classification task;

[0031] The total input features of the fault classification task are obtained by calculating the weights of the shared features of the fault classification task, as shown in the following formula:

[0032] ;

[0033] in: represents the total input features for the fault classification task.

[0034] As a preferred embodiment of the present invention, the subtask feature extraction layer is constructed based on an improved LSTM-ResNet feature extraction network. The network structure is formed by stacking a multi-layer LSTM network and a ResNet network, wherein the LSTM part is used for feature extraction of the remaining life prediction task, and the ResNet part is used for feature extraction of the fault classification task. At the same time, the hidden state results output by each layer of the LSTM network during the feature extraction process will be sent in parallel to the identity mapping part of the corresponding ResNet network of the current level for fusion;

[0035] The specific steps of extracting features by the improved LSTM-ResNet feature extraction network are as follows:

[0036] The total input features that will be used for the remaining life prediction task Input the LSTM network in the first layer of LSTM-ResNet to extract features, as shown in the following formula:

[0037] ;

[0038] in: It is the feature output by the LSTM network in the first layer of LSTM-ResNet; Represents an LSTM network;

[0039] The total input features that will be used for the fault classification task Input the ResNet network in the first layer of LSTM-ResNet to extract features, as shown in the following formula:

[0040] ;

[0041] in: It is the feature output by the ResNet network in the first layer of LSTM-ResNet; Represents the ResNet network;

[0042] Then As the input of the LSTM network in the second layer LSTM-ResNet, As the input of the ResNet network in the second layer of LSTM-ResNet, the features of the second layer of LSTM-ResNet are extracted. And so on, the features extracted by the LSTM network and the ResNet network in the last layer of LSTM-ResNet are used as the final output.

[0043] As a preferred embodiment of the present invention, the output layer is composed of a fully connected layer, which is used to output the results of the remaining life prediction task and the fault classification task, and the process is shown in the following formula:

[0044] ;

[0045] ;

[0046] in: Output values ​​for the remaining life prediction task; The remaining life prediction task features output by LSTM-ResNet; The weight matrix for calculating the output results of the remaining life prediction task in the fully connected layer; It is the bias term for calculating the output result of the remaining life prediction task in the fully connected layer; When calculating the output result of the remaining life prediction task in the fully connected layer Activation function; Output values ​​for the fault classification task; It is the fault classification task feature output by LSTM-ResNet; It is the weight matrix for calculating the output results of the fault classification task in the fully connected layer; It is the bias term for calculating the output result of the fault classification task in the fully connected layer; Represents the normalized exponential activation function when calculating the output results of the fault classification task in the fully connected layer.

[0047] As a preferred embodiment of the present invention, the specific steps of the capacitive voltage transformer for making operation and maintenance decisions are:

[0048] The fault categories of the capacitor voltage transformer are quantified to obtain the quantized value corresponding to each fault category, and then the operational health of the capacitor voltage transformer is calculated, as shown in the following formula:

[0049] ;

[0050] in: Indicates the current operational health of the capacitor voltage transformer; Indicates the normalized remaining life value of the current capacitor voltage transformer; Indicates the normalized quantized value of the fault category of the current capacitor voltage transformer; Indicates the historical failure rate of the current capacitor voltage transformer;

[0051] The standard score Z-Score of each capacitive voltage transformer in the calculation is as shown in the following formula:

[0052] ;

[0053] in: Indicates The standard score Z-Score of each capacitor voltage transformer; Indicates The operational health of the capacitor voltage transformer; Indicates the capacitive voltage transformer index; It represents the average operating health of all capacitor voltage transformers; represents the standard deviation of the operational health of all capacitor voltage transformers;

[0054] based on The size of the setting varies depending on the operation and maintenance decision.

[0055] On the other hand, the present invention also provides a capacitor voltage transformer operation and maintenance decision system, including a data acquisition module, a multi-task processing module and an operation and maintenance decision module;

[0056] The data acquisition module is used to collect historical operation data of the capacitor voltage transformer and perform preprocessing;

[0057] The multi-task processing module is used to construct a multi-task model, train the multi-task model through the pre-processed historical operation data of the capacitor voltage transformer in the whole life cycle, and complete the remaining life prediction task and fault classification task of the capacitor voltage transformer through the trained multi-task model;

[0058] The operation and maintenance decision module is used to make an operation and maintenance decision for the capacitor voltage transformer based on the remaining life and fault category of the capacitor voltage transformer output by the multi-task model.

[0059] The present invention has the following beneficial effects:

[0060] The present invention proposes a method for calculating the operating health of a transformer by combining the remaining life of the transformer with the degree of fault, designs a multi-level multi-task feature extraction structure, realizes accurate prediction of the remaining life of the transformer and accurate diagnosis of the fault category, and makes operation and maintenance decisions based on the accurate prediction of the remaining life of the transformer and the fault category, thereby realizing intelligent decision-making on the operation and maintenance mode of the capacitive voltage transformer. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 is a flow chart of the method of the present invention;

[0062] Figure 2 This is a structural diagram of the multi-task model of the present invention. DETAILED DESCRIPTION

[0063] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0064] It should be understood that the step numbers used in this document are only for convenience of description and are not intended to limit the order in which the steps are executed.

[0065] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0066] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0067] The term "and / or" means and includes any and all possible combinations of one or more of the associated listed items.

[0068] Embodiment 1:

[0069] See also Figure 1 , a capacitor voltage transformer operation and maintenance decision method, comprising the following steps:

[0070] Collect historical operation data of capacitor voltage transformer and perform preprocessing;

[0071] To build a multi-task model, see Figure 2 The multi-task model includes a coding layer, a shared feature extraction layer, a sub-task feature extraction layer and an output layer. The multi-task model is trained by preprocessing the historical operation data of the capacitor voltage transformer in the whole life cycle, and the remaining life prediction task and fault classification task of the capacitor voltage transformer are completed by the trained multi-task model;

[0072] Based on the remaining life and fault category of the capacitor voltage transformer output by the multi-task model, operation and maintenance decisions are made for the capacitor voltage transformer.

[0073] As a preferred implementation of this embodiment, the historical operation data includes: capacitor voltage transformer secondary voltage, environmental parameter data, capacitor voltage transformer remaining life, capacitor voltage transformer historical error evaluation data and its corresponding fault category;

[0074] The environmental parameter data further includes secondary load and temperature data;

[0075] The remaining life of the transformer is calculated by subtracting the used years from the total used years collected from the historical records of the capacitor voltage transformer;

[0076] The historical error evaluation data includes historical ratio difference data and angle difference data of the mutual inductor;

[0077] The corresponding fault categories include normal, damping device degradation, high-voltage capacitor damp, low-voltage capacitor damp, low-voltage capacitor breakdown, and high-voltage capacitor breakdown.

[0078] As a preferred implementation of this embodiment, the preprocessing step of the historical operation data is: normalizing the continuous data and labeling the category data.

[0079] As a preferred implementation of this embodiment, the encoding layer is used to vector encode the pre-processed historical operation data, and splice the encoded data to form a comprehensive input vector , as shown in the following formula:

[0080] ;

[0081] in: Represents the historical operation data after preprocessing; Indicates encoding operation; is the first vector value, Pick ; Represents the total length of the synthesized input vector.

[0082] As a preferred implementation of this embodiment, the shared feature extraction layer is used to extract the comprehensive input vector For the shared features related to both the remaining life prediction task and the fault classification task, the specific steps are as follows:

[0083] Constructing the synthetic input vector The shared feature extraction unit mapped to its corresponding eigenvalue performs shared feature extraction, as shown in the following formula:

[0084] ;

[0085] in: For shared features; is a shared feature extraction unit; is the feature mapping parameter;

[0086] The vector-encoded capacitor voltage transformer secondary voltage and environmental parameter data in the preprocessed historical operation data are used as the input vector of the remaining life prediction task. , and Shared features The similarity calculation is performed to obtain the shared feature allocation weights for the remaining life prediction task, as shown in the following formula:

[0087] ;

[0088] ;

[0089] in: Indicates the secondary voltage of the capacitor voltage transformer; Indicates environmental parameter data; Represents the input vector of the remaining life prediction task No. vector value, Pick ; express Total length; Assign weights to the shared features for the remaining life prediction task;

[0090] The total input features of the remaining life prediction task are calculated based on the shared features of the remaining life prediction task, as shown in the following formula:

[0091] ;

[0092] in: represents the total input features for the remaining life prediction task;

[0093] The historical error evaluation data of capacitor voltage transformer after vector encoding in the preprocessed historical operation data is used as the input vector of the fault classification task , and Shared features Perform similarity calculation to obtain the shared feature allocation weights for the fault classification task, as shown in the following formula:

[0094] ;

[0095] ;

[0096] in: Represents the ratio difference data in the historical error evaluation data of the capacitor voltage transformer; Represents the angle difference data in the historical error evaluation data of the capacitor voltage transformer; Represents the input vector of the fault classification task No. vector value, Pick ; express Total length; Assign weights to shared features for the fault classification task;

[0097] The total input features of the fault classification task are obtained by calculating the weights of the shared features of the fault classification task, as shown in the following formula:

[0098] ;

[0099] in: represents the total input features for the fault classification task.

[0100] As a preferred implementation of this embodiment, the subtask feature extraction layer is constructed based on an improved LSTM-ResNet feature extraction network. The network structure is formed by stacking a multi-layer LSTM network and a ResNet network, wherein the LSTM part is used for feature extraction of the remaining life prediction task, and the ResNet part is used for feature extraction of the fault classification task. At the same time, the hidden state results output by each layer of the LSTM network during the feature extraction process will be sent in parallel to the identity mapping part of the corresponding ResNet network at the current level for fusion, so that the extracted remaining life prediction related features can also be used to support the fault classification task, so that the feature information can be more fully utilized, and the output result is more accurate;

[0101] The specific steps of extracting features by the improved LSTM-ResNet feature extraction network are as follows:

[0102] The total input features that will be used for the remaining life prediction task Input the LSTM network in the first layer of LSTM-ResNet to extract features, as shown in the following formula:

[0103] ;

[0104] in: It is the feature output by the LSTM network in the first layer of LSTM-ResNet; Represents an LSTM network;

[0105] The total input features that will be used for the fault classification task Input the ResNet network in the first layer of LSTM-ResNet to extract features, as shown in the following formula:

[0106] ;

[0107] in: It is the feature output by the ResNet network in the first layer of LSTM-ResNet; Represents the ResNet network;

[0108] Then As the input of the LSTM network in the second layer LSTM-ResNet, As the input of the ResNet network in the second layer of LSTM-ResNet, the feature extraction of the second layer of LSTM-ResNet is performed. And so on, the high semantic features extracted by the LSTM network and the ResNet network in the last layer of LSTM-ResNet are used as the final output.

[0109] As a preferred implementation of this embodiment, the output layer is composed of a fully connected layer, which is used to output the results of the remaining life prediction task and the fault classification task. The process is shown in the following formula:

[0110] ;

[0111] ;

[0112] in: Output values ​​for the remaining life prediction task; The remaining life prediction task features output by LSTM-ResNet; The weight matrix for calculating the output results of the remaining life prediction task in the fully connected layer; It is the bias term for calculating the output result of the remaining life prediction task in the fully connected layer; When calculating the output result of the remaining life prediction task in the fully connected layer Activation function; Output values ​​for the fault classification task; It is the fault classification task feature output by LSTM-ResNet; It is the weight matrix for calculating the output results of the fault classification task in the fully connected layer; It is the bias term for calculating the output result of the fault classification task in the fully connected layer; Represents the normalized exponential activation function when calculating the output results of the fault classification task in the fully connected layer.

[0113] By setting the mean square error loss function and the cross entropy loss function for the remaining life prediction task and the fault category diagnosis task respectively, and training the model with a large amount of sample transformer historical operation data, a well-trained multi-task model is obtained for the remaining life prediction and fault category diagnosis of the transformer to be tested.

[0114] As a preferred implementation of this embodiment, the specific steps of the capacitive voltage transformer making an operation and maintenance decision are:

[0115] According to expert experience, the fault categories of the capacitor voltage transformer are quantified to obtain the quantized values ​​corresponding to each fault category. In this embodiment, the initial quantized values ​​of the transformer fault categories are set to: [normal, damping device degradation, low voltage capacitor damped, high voltage capacitor damped, low voltage capacitor breakdown, high voltage capacitor breakdown] = [0, 1, 1, 2, 3, 4], and then the operation health of the capacitor voltage transformer is calculated, as shown in the following formula:

[0116] ;

[0117] in: Indicates the current operational health of the capacitor voltage transformer; Indicates the normalized remaining life value of the current capacitor voltage transformer; Indicates the normalized quantized value of the fault category of the current capacitor voltage transformer; Indicates the historical failure rate of the current capacitor voltage transformer, which can be obtained through the archive information of the capacitor voltage transformer;

[0118] The standard score Z-Score of each capacitive voltage transformer in the calculation is as shown in the following formula:

[0119] ;

[0120] in: Indicates The standard score Z-Score of a capacitor voltage transformer is expressed as The degree of deviation from the mean value of the operating health of each transformer; Indicates The operational health of the capacitor voltage transformer; Indicates the capacitive voltage transformer index; It represents the average operating health of all capacitor voltage transformers; represents the standard deviation of the operational health of all capacitor voltage transformers;

[0121] based on The size of the setting is different operation and maintenance decisions, specifically:

[0122] like , then the operation and maintenance strategy of the capacitor voltage transformer is to postpone one verification cycle;

[0123] like , then the operation and maintenance strategy of the capacitor voltage transformer is determined to be normal operation and maintenance according to the capacitor voltage transformer verification cycle;

[0124] like , the operation and maintenance strategy is to immediately arrange a power outage for maintenance.

[0125] Embodiment 2:

[0126] A capacitive voltage transformer operation and maintenance decision system, comprising a data acquisition module, a multi-task processing module and an operation and maintenance decision module;

[0127] The data acquisition module is used to collect historical operation data of the capacitor voltage transformer and perform preprocessing;

[0128] The multi-task processing module is used to construct a multi-task model, train the multi-task model through the pre-processed historical operation data of the capacitor voltage transformer in the whole life cycle, and complete the remaining life prediction task and fault classification task of the capacitor voltage transformer through the trained multi-task model;

[0129] The operation and maintenance decision module is used to make an operation and maintenance decision for the capacitor voltage transformer based on the remaining life and fault category of the capacitor voltage transformer output by the multi-task model.

[0130] The system is used to implement the method in Example 1, which will not be described in detail here.

[0131] In the embodiments of the present application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Among them, A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c can be represented by: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, c can be single or multiple.

[0132] Those of ordinary skill in the art will appreciate that the various units and algorithm steps described in the embodiments disclosed herein can be implemented in a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0133] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0134] In several embodiments provided in the present application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory; hereinafter referred to as: ROM), random access memory (Random Access Memory; hereinafter referred to as: RAM), disk or optical disk, and other media that can store program codes.

[0135] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A capacitor voltage transformer operation and maintenance decision method, characterized in that: The following steps are involved: Collect historical operation data of capacitor voltage transformer and perform preprocessing; Constructing a multi-task model, the multi-task model includes a coding layer, a shared feature extraction layer, a sub-task feature extraction layer and an output layer, training the multi-task model through preprocessed historical operation data of the capacitor voltage transformer in the whole life cycle, and completing the remaining life prediction task and fault classification task of the capacitor voltage transformer through the trained multi-task model; The encoding layer is used to perform vector encoding on the pre-processed historical operation data; The shared feature extraction layer is used to extract shared features in vector coding that are relevant to both the remaining life prediction task and the fault classification task; The subtask feature extraction layer is constructed based on an improved LSTM-ResNet feature extraction network. The network structure is formed by stacking multiple layers of LSTM networks and ResNet networks, wherein the LSTM part is used for feature extraction of the remaining life prediction task, and the ResNet part is used for feature extraction of the fault classification task. At the same time, the hidden state results output by each layer of LSTM network during the feature extraction process will be sent in parallel to the identity mapping part of the corresponding ResNet network of the current layer for fusion; The output layer is composed of a fully connected layer and is used to output the results of the remaining life prediction task and the fault classification task; Based on the remaining life and fault category of the capacitor voltage transformer output by the multi-task model, operation and maintenance decisions are made for the capacitor voltage transformer.

2. A capacitor voltage transformer operation and maintenance decision method according to claim 1, characterized in that: The historical operation data includes: capacitor voltage transformer secondary voltage, environmental parameter data, capacitor voltage transformer remaining life, capacitor voltage transformer historical error evaluation data and corresponding fault categories.

3. A capacitor voltage transformer operation and maintenance decision method according to claim 1, characterized in that: The preprocessing steps of the historical operation data are: normalizing the continuous data and labeling the category data.

4. A capacitor voltage transformer operation and maintenance decision method according to claim 1, characterized in that: The specific steps of vector encoding the pre-processed historical operation data are as follows: splicing the pre-processed historical operation data to form a comprehensive input vector , as shown in the following formula: ； in: Represents the historical operation data after preprocessing; Indicates encoding operation; is the first vector value, Pick ; Represents the total length of the synthesized input vector.

5. A capacitor voltage transformer operation and maintenance decision method according to claim 4, characterized in that: The specific steps for extracting the shared features in the vector encoding that are relevant to both the remaining life prediction task and the fault classification task are: Constructing the synthetic input vector The shared feature extraction unit mapped to its corresponding eigenvalue performs shared feature extraction, as shown in the following formula: ； in: For shared features; is a shared feature extraction unit; is the feature mapping parameter; The vector-encoded capacitor voltage transformer secondary voltage and environmental parameter data in the preprocessed historical operation data are used as the input vector of the remaining life prediction task. , and Shared features The similarity calculation is performed to obtain the shared feature allocation weights for the remaining life prediction task, as shown in the following formula: ； ； in: Indicates the secondary voltage of the capacitor voltage transformer; Indicates environmental parameter data; Represents the input vector of the remaining life prediction task No. vector value, Pick ; express Total length; Assign weights to the shared features for the remaining life prediction task; The total input features of the remaining life prediction task are calculated based on the shared features of the remaining life prediction task, as shown in the following formula: ； in: represents the total input features for the remaining life prediction task; The historical error evaluation data of capacitor voltage transformer after vector encoding in the preprocessed historical operation data is used as the input vector of the fault classification task , and Shared features Perform similarity calculation to obtain the shared feature allocation weights for the fault classification task, as shown in the following formula: ； ； in: Represents the ratio difference data in the historical error evaluation data of the capacitor voltage transformer; Represents the angle difference data in the historical error evaluation data of the capacitor voltage transformer; Represents the input vector of the fault classification task No. vector value, Pick ; express Total length; Assign weights to shared features for the fault classification task; The total input features of the fault classification task are obtained by calculating the weights of the shared features of the fault classification task, as shown in the following formula: ； in: represents the total input features for the fault classification task.

6. A capacitor voltage transformer operation and maintenance decision method according to claim 5, characterized in that: The specific steps of extracting features by the improved LSTM-ResNet feature extraction network are: The total input features that will be used for the remaining life prediction task Input the LSTM network in the first layer of LSTM-ResNet to extract features, as shown in the following formula: ； in: It is the feature output by the LSTM network in the first layer of LSTM-ResNet; Represents an LSTM network; The total input features that will be used for the fault classification task Input the ResNet network in the first layer of LSTM-ResNet to extract features, as shown in the following formula: ； in: It is the feature output by the ResNet network in the first layer of LSTM-ResNet; Represents the ResNet network; Then As the input of the LSTM network in the second layer LSTM-ResNet, As the input of the ResNet network in the second layer of LSTM-ResNet, the features of the second layer of LSTM-ResNet are extracted. And so on, the features extracted by the LSTM network and the ResNet network in the last layer of LSTM-ResNet are used as the final output.

7. A capacitor voltage transformer operation and maintenance decision method according to claim 6, characterized in that: The process of the output layer outputting the results of the remaining life prediction task and the fault classification task is shown in the following formula: ； ； in: Output values ​​for the remaining life prediction task; The remaining life prediction task features output by LSTM-ResNet; The weight matrix for calculating the output results of the remaining life prediction task in the fully connected layer; It is the bias term for calculating the output result of the remaining life prediction task in the fully connected layer; When calculating the output result of the remaining life prediction task in the fully connected layer Activation function; Output values ​​for the fault classification task; It is the fault classification task feature output by LSTM-ResNet; It is the weight matrix for calculating the output results of the fault classification task in the fully connected layer; It is the bias term for calculating the output result of the fault classification task in the fully connected layer; Represents the normalized exponential activation function when calculating the output results of the fault classification task in the fully connected layer.

8. A capacitor voltage transformer operation and maintenance decision method according to claim 7, characterized in that: The specific steps of the capacitive voltage transformer for making operation and maintenance decisions are as follows: The fault categories of the capacitor voltage transformer are quantified to obtain the quantized value corresponding to each fault category, and then the operational health of the capacitor voltage transformer is calculated, as shown in the following formula: ； in: Indicates the current operational health of the capacitor voltage transformer; Indicates the normalized remaining life value of the current capacitor voltage transformer; Indicates the normalized fault category quantization value of the current capacitor voltage transformer; Indicates the historical failure rate of the current capacitor voltage transformer; represents a natural constant; Calculate the standard score Z-Score of each capacitor voltage transformer as shown in the following formula: ； in: Indicates The standard score Z-Score of each capacitor voltage transformer; Indicates The operational health of the capacitor voltage transformer; Indicates the capacitive voltage transformer index; It represents the average operating health of all capacitor voltage transformers; represents the standard deviation of the operational health of all capacitor voltage transformers; based on The size of the settings varies depending on the operation and maintenance decisions.

9. A capacitor voltage transformer operation and maintenance decision system, used in the method according to any one of claims 1 to 8, characterized in that: Including data acquisition module, multi-task processing module and operation and maintenance decision module; The data acquisition module is used to collect historical operation data of the capacitor voltage transformer and perform preprocessing; The multi-task processing module is used to construct a multi-task model, train the multi-task model through the pre-processed historical operation data of the capacitor voltage transformer in the whole life cycle, and complete the remaining life prediction task and fault classification task of the capacitor voltage transformer through the trained multi-task model; The operation and maintenance decision module is used to make an operation and maintenance decision for the capacitor voltage transformer based on the remaining life and fault category of the capacitor voltage transformer output by the multi-task model.

Images