Adaptive identification method and system for key factors of switching misoperation of transformer substation

Abstract

The invention discloses a transformer substation switching misoperation key factor adaptive identification method and system, and relates to the technical field of adaptive identification, and the method comprises the steps: employing a hierarchical structure system to process a system architecture, and constructing a hierarchical semantic feature extraction model for the text data of a transformer substation operation bill; for substation equipment state data, a time sequence feature extraction network based on improved LSTM is proposed, and a dynamic fusion method based on an attention mechanism is designed; a transformer substation operation knowledge distillation framework is provided, and transformer substation switching is recognized based on quantitative evaluation of gradient accumulation. Unified representation and analysis of various types of data are realized, a knowledge distillation technology is applied to the field of substation switching operation, a teacher-student network framework is designed, a complex expert experience model is compressed into a lightweight practical model, a soft label generation method based on temperature scaling is provided, fine-grained information of expert decision is effectively reserved, and the reliability of substation switching operation is improved. And real-time adaptive identification of risk factors is realized.

Description

Technical Field

[0001] The invention relates to the field of adaptive identification technology, and in particular to an adaptive identification method and system for key factors of switching misoperation of a transformer substation. Background Art

[0002] With the rapid development of power systems, the concept of smart grids has become increasingly popular. Smart grids not only require efficient data collection and processing capabilities, but also require powerful fault diagnosis and prediction functions to ensure the safe and stable operation of power systems. In this context, substations, as key nodes in power systems, have a direct impact on the efficiency and reliability of the entire power system through their intelligent management level. The operation of substations, especially switching operations, involves a large amount of data processing and complex operating procedures. The correctness and timeliness of these operations are crucial to ensuring the safety of power systems. However, with the continuous expansion of the scale of power systems and the increase in technical complexity, traditional manual operations and simple automation methods can no longer meet the needs of modern power systems. There is an urgent need to develop more intelligent and efficient substation operation and management technologies.

[0003] Most of the existing methods for identifying substation switching operations rely on single-mode data analysis, such as only based on historical operation records or equipment status monitoring data. This method has obvious limitations. First, single-mode data is difficult to fully reflect the actual situation of substation operations, which can easily lead to misjudgment. Secondly, traditional methods lack an effective knowledge transfer mechanism and cannot make full use of existing operating experience and knowledge to improve the processing effect of new tasks. Finally, existing methods usually do not consider the mutual influence between different operating factors and cannot provide accurate operational risk assessment. To this end, the present invention proposes a method for adaptively identifying key factors of substation switching misoperation. By constructing a hierarchical system architecture, using an improved LSTM model and attention mechanism to perform deep fusion of multi-modal data, and realizing effective migration of operational knowledge through a knowledge distillation framework, the accuracy and reliability of substation switching operation identification are improved, providing strong technical support for the safe and stable operation of the power system. Summary of the invention

[0004] In view of the problems existing in the existing adaptive identification method and system for key factors of switching misoperation of substations, the present invention is proposed.

[0005] Therefore, the purpose of the present invention is to provide a method and system for adaptively identifying key factors of substation switching erroneous operation. In order to solve the problem that erroneous operation is often caused by the interaction of multiple factors, the present invention proposes a method and system for adaptively identifying key factors of substation switching erroneous operation.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In the first aspect, an embodiment of the present invention provides an adaptive identification method for key factors of substation switching misoperation, which includes constructing a system architecture, using a hierarchical structural system to process the system architecture, and constructing a hierarchical semantic feature extraction model for the text data of substation operation tickets; for substation equipment status data, a time series feature extraction network based on an improved LSTM is proposed, and a dynamic fusion method based on an attention mechanism is designed; a substation operation knowledge distillation framework is proposed, and substation switching is identified based on quantitative evaluation of gradient accumulation.

[0008] As a preferred solution of the adaptive identification method of key factors of substation switching misoperation of the present invention, wherein: the construction of the system architecture includes adopting a hierarchical structure system;

[0009] The structural system includes a data acquisition layer, a feature extraction layer, a knowledge transfer layer and a decision output layer. The data acquisition layer is responsible for the acquisition and preprocessing of multi-source heterogeneous data of the substation, the feature extraction layer realizes the automatic learning of deep features, the knowledge transfer layer completes the intelligent reuse of substation operation experience knowledge, and the decision output layer gives the final recognition results and evaluation suggestions.

[0010] As a preferred solution of the adaptive identification method of key factors of substation switching misoperation described in the present invention, wherein: the construction of a hierarchical semantic feature extraction model includes multimodal feature extraction for text data of substation operation tickets, and the formula for multimodal feature extraction is:

[0011] S=D(X_text)

[0012] D(X_text)=σ(W_t·E(X_text)+b_t)

[0013] Among them, X_text represents the input substation operation bill text data, E(·) represents the word embedding mapping function, which is used to convert the operation text into a dense vector representation, W_t represents the weight matrix of text feature extraction, b_t represents the text feature extraction bias vector, and σ represents the nonlinear activation function.

[0014] As a preferred solution of the adaptive identification method of key factors of substation switching misoperation described in the present invention, the proposed time series feature extraction network based on improved LSTM includes extracting equipment status features from substation equipment status data, and the formula for extracting equipment status features is:

[0015] H(t)=LSTM(X_device(t),H(t-1))

[0016] f(t)=σ(W_d·[H(t-1),X_device(t)]+b_d)

[0017] Among them, X_device(t) represents the substation equipment state input at time t, H(t) represents the equipment state hidden state vector, which is dynamically updated through the LSTM unit, and W_d and b_d represent the weight matrix and bias vector of equipment state feature extraction, respectively.

[0018] As a preferred solution of the adaptive identification method of key factors of substation switching misoperation described in the present invention, the dynamic fusion method based on the attention mechanism is designed, including calculating the feature fusion mechanism, and the formula for calculating the feature fusion mechanism is:

[0019]

[0020] Z=σ(W_f·[α_text; α_device; α_human]+b_f)

[0021] Among them, α_k represents the attention weights of various features of the substation, which is obtained through the interactive calculation of the query matrix Q_k, the key matrix K_k and the value matrix V_k. Z represents the fused feature representation, and the deep integration of features is achieved through the learnable parameter matrix W_f and the bias vector b_f.

[0022] As a preferred solution of the adaptive identification method of key factors of substation switching misoperation described in the present invention, the proposed substation operation knowledge distillation framework includes calculating the substation operation knowledge distillation framework according to the knowledge migration framework, and the formula for calculating the substation operation knowledge distillation framework is:

[0023] T(x) = softmax(g(f(x)) / τ)

[0024] L=λ_1·L_CE(y,S(x))+λ_2·L_KL(T(x),S(x))+λ_3·L_reg

[0025] Among them, T(x) represents the knowledge representation extracted by the substation expert model, τ is the temperature parameter, and the loss function includes the cross entropy loss L_CE, the knowledge distillation loss L_KL and the regularization term L_reg, which are balanced by the weight coefficients λ_1, λ_2, and λ_3.

[0026] As a preferred solution of the adaptive identification method of key factors of substation switching misoperation of the present invention, the identification of substation switching includes calculating a quantitative evaluation method based on gradient accumulation according to factor evaluation, and the formula for calculating the quantitative evaluation method based on gradient accumulation is:

[0027]

[0028] S(f_i)=I(f_i) / ∑I(f_j)

[0029] Among them, I(f_i) represents the influence of the substation misoperation factor f_i, which is obtained by calculating the gradient accumulation of the loss function on the factor; S(f_i) is the normalized importance score.

[0030] In the second aspect, an embodiment of the present invention provides an adaptive identification system for key factors of substation switching misoperation, which includes: a construction module, which constructs a system architecture, adopts a hierarchical structural system to process the system architecture, and constructs a hierarchical semantic feature extraction model for the text data of the substation operation ticket; a calculation module, which proposes a time series feature extraction network based on the improved LSTM for the substation equipment status data, and designs a dynamic fusion method based on the attention mechanism; an identification module, which proposes a substation operation knowledge distillation framework, and identifies the substation switching based on the quantitative evaluation of gradient accumulation.

[0031] In a third aspect, an embodiment of the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein: when the processor executes the computer program, any step of the above-mentioned method for adaptively identifying key factors of substation switching misoperation is implemented.

[0032] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the above-mentioned method for adaptively identifying key factors of substation switching misoperation is implemented.

[0033] The beneficial effects of the present invention are as follows: the present invention collaboratively analyzes multi-source data such as operation ticket information, equipment status, environmental parameters, etc., proposes a multimodal feature fusion processing mechanism based on deep learning, realizes unified representation and analysis of various types of data, applies knowledge distillation technology to the field of substation switching operations, designs a teacher-student network framework, compresses complex expert experience models into lightweight practical models, proposes a soft label generation method based on temperature scaling, effectively retains the fine-grained information of expert decisions, and realizes real-time adaptive identification of risk factors. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0035] Figure 1 A flowchart of a method and system for adaptively identifying key factors of substation switching misoperation provided by an embodiment of the present invention.

[0036] Figure 2 A schematic diagram of an adaptive identification method and system for key factors of substation switching misoperation provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0037] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0038] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0039] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0040] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0041] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0042] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0043] Example 1

[0044] Reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and provides a method for adaptively identifying key factors of substation switching misoperation, comprising:

[0045] S1: Build the system architecture, use a hierarchical structure system to process the system architecture, and build a hierarchical semantic feature extraction model for the text data of substation operation tickets.

[0046] Among them, building the system architecture includes adopting a hierarchical structural system;

[0047] The structural system includes data acquisition layer, feature extraction layer, knowledge transfer layer and decision output layer. The data acquisition layer is responsible for the acquisition and preprocessing of multi-source heterogeneous data of substations. The feature extraction layer realizes automatic learning of deep features. The knowledge transfer layer completes the intelligent reuse of substation operation experience knowledge. The decision output layer gives the final recognition results and evaluation suggestions.

[0048] S1.1: Constructing a hierarchical semantic feature extraction model includes extracting multimodal features from the text data of the substation operation ticket. The formula for multimodal feature extraction is:

[0049] S=D(X_text)

[0050] D(X_text)=σ(W_t·E(X_text)+b_t)

[0051] Among them, X_text represents the input substation operation bill text data, E(·) represents the word embedding mapping function, which is used to convert the operation text into a dense vector representation, W_t represents the weight matrix of text feature extraction, b_t represents the text feature extraction bias vector, and σ represents the nonlinear activation function.

[0052] S2: For the substation equipment status data, a time series feature extraction network based on improved LSTM is proposed, and a dynamic fusion method based on attention mechanism is designed.

[0053] Among them, the proposed time series feature extraction network based on improved LSTM includes equipment status feature extraction for substation equipment status data. The formula for equipment status feature extraction is:

[0054] H(t)=LSTM(X_device(t),H(t-1))

[0055] f(t)=σ(W_d·[H(t-1),X_device(t)]+b_d)

[0056] Among them, X_device(t) represents the substation equipment state input at time t, H(t) represents the equipment state hidden state vector, which is dynamically updated through the LSTM unit, and W_d and b_d represent the weight matrix and bias vector of equipment state feature extraction, respectively.

[0057] S2.1: Designing a dynamic fusion method based on the attention mechanism includes calculating the feature fusion mechanism. The formula for calculating the feature fusion mechanism is:

[0058]

[0059] Z=σ(W_f·[α_text; α_device; α_human]+b_f)

[0060] Among them, α_k represents the attention weights of various features of the substation, which is obtained through the interactive calculation of the query matrix Q_k, the key matrix K_k and the value matrix V_k. Z represents the fused feature representation, and the deep integration of features is achieved through the learnable parameter matrix W_f and the bias vector b_f.

[0061] S3: A knowledge distillation framework for substation operation is proposed to identify substation switching based on quantitative evaluation of gradient accumulation.

[0062] Among them, the proposed substation operation knowledge distillation framework includes calculating the substation operation knowledge distillation framework according to the knowledge migration framework. The formula for calculating the substation operation knowledge distillation framework is:

[0063] T(x) = softmax(g(f(x)) / τ)

[0064] L=λ_1·L_CE(y,S(x))+λ_2·L_KL(T(x),S(x))+λ_3·L_reg

[0065] Among them, T(x) represents the knowledge representation extracted by the substation expert model, τ is the temperature parameter, and the loss function includes the cross entropy loss L_CE, the knowledge distillation loss L_KL and the regularization term L_reg, which are balanced by the weight coefficients λ_1, λ_2, and λ_3.

[0066] S3.1: The identification of substation switching includes calculating a quantitative evaluation method based on gradient accumulation according to factor evaluation. The formula for calculating the quantitative evaluation method based on gradient accumulation is:

[0067]

[0068] S(f_i)=I(f_i) / ∑I(f_j)

[0069] Among them, I(f_i) represents the influence of the substation misoperation factor f_i, which is obtained by calculating the gradient accumulation of the loss function on the factor; S(f_i) is the normalized importance score.

[0070] In a preferred embodiment, a substation switching identification system based on multimodality and knowledge distillation includes a construction module, which constructs a system architecture, uses a hierarchical structural system to process the system architecture, and constructs a hierarchical semantic feature extraction model for the text data of the substation operation ticket; a calculation module, which proposes a time series feature extraction network based on the improved LSTM for the substation equipment status data, and designs a dynamic fusion method based on the attention mechanism; an identification module, which proposes a substation operation knowledge distillation framework, and identifies the substation switching based on the quantitative evaluation of gradient accumulation.

[0071] The above-mentioned unit modules may be embedded in or independent of a processor in a computer device in the form of hardware, or may be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to the above-mentioned modules.

[0072] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0073] In summary, the present invention collaboratively analyzes multi-source data such as operation ticket information, equipment status, and environmental parameters, proposes a multimodal feature fusion processing mechanism based on deep learning, realizes the unified representation and analysis of various types of data, applies knowledge distillation technology to the field of substation switching operation, designs a teacher-student network framework, compresses complex expert experience models into lightweight practical models, and proposes a soft label generation method based on temperature scaling to effectively retain the fine-grained information of expert decision-making and realize real-time adaptive identification of risk factors.

[0074] Example 2

[0075] Reference Figure 1 and Figure 2 , which is the second embodiment of the present invention, and this embodiment provides a method for adaptively identifying key factors of substation switching misoperation. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through simulation experiments.

[0076] In order to verify the effectiveness of the proposed substation switching identification method based on multimodality and knowledge distillation, a detailed experimental study was carried out. In the experiment, an actual operation dataset from a large substation was used. The dataset contained a total of 12,000 switching operation records in the past year. Each record contained multimodal information such as operation ticket text data and equipment status data. First, the data was preprocessed and preliminarily screened through the constructed hierarchical system architecture to ensure the quality of the data. Then, in the feature extraction stage, the improved LSTM model was used to process the time series data of the equipment status. At the same time, a deep neural network based on word embedding was used to extract semantic features of the operation ticket text. Subsequently, the two types of features, text and equipment status, were dynamically fused through the designed attention mechanism to achieve deep fusion of multimodal data. On this basis, through the knowledge distillation framework, the operation knowledge is successfully extracted from the pre-trained expert model and applied to the target model, which significantly improves the generalization ability and recognition accuracy of the model. Finally, the experimental results show that compared with the traditional method, the method of the present invention improves the recognition accuracy by about 15%, reaching more than 95%, and the false alarm rate is reduced by about 10%, which proves the effectiveness and superiority of the method of the present invention in practical applications. The experimental data of the present invention are shown in Table 1 below:

[0077] Table 1 Experimental data table of the present invention

[0078] Data Types describe Specific values Total number of operation records Number of switching operation records used in the experiment 12,000 times Time Range Recording time span Past year Feature extraction Equipment status data processing model Improved LSTM model Feature extraction Operation ticket text data processing model Deep Neural Networks Based on Word Embedding Fusion Method Multimodal data fusion mechanism Attention Mechanism Knowledge transfer Operational Knowledge Transfer Framework Knowledge Distillation Framework Experimental Results Compared with traditional methods, the recognition accuracy is improved Improved by about 15% Experimental Results Final recognition accuracy Reach more than 95% Experimental Results Reduced false positive rate compared to traditional methods Reduced by about 10%

[0079] The experiment is based on 12,000 substation switching operation records in the past year. The improved LSTM model is used to process the equipment status data, and the deep neural network based on word embedding is used to extract the text features of the operation ticket. The deep fusion of multimodal data is realized through the attention mechanism. After combining the knowledge distillation framework for operation knowledge migration, the experimental results show that compared with the traditional method, the recognition accuracy of the method of the present invention is improved by about 15%, reaching more than 95%, and the false alarm rate is reduced by about 10%. The comparison between the present invention and the prior art is shown in Table 2 below:

[0080] Table 2 Comparison between the present invention and the prior art

[0081]

[0082] The technical solution of the present invention performs feature extraction through multimodal data fusion, an improved LSTM model and a deep neural network based on word embedding, and adopts a dynamic attention mechanism and a knowledge distillation framework, which significantly improves the recognition accuracy of substation switching operations to more than 95%, and reduces the false alarm rate to less than 10%. Compared with the existing technology, the present invention shows stronger comprehensive advantages in data processing, feature extraction, feature fusion and knowledge transfer, and effectively improves the safety and reliability of substation operations.

[0083] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An adaptive identification method for key factors of substation switching misoperation, characterized by: include, Construct the system architecture, use a hierarchical structure system to process the system architecture, and build a hierarchical semantic feature extraction model for the text data of substation operation tickets; For substation equipment status data, a time series feature extraction network based on improved LSTM is proposed, and a dynamic fusion method based on attention mechanism is designed; A substation operation knowledge distillation framework is proposed to identify substation switching based on quantitative evaluation of gradient accumulation.

2. The adaptive identification method for key factors of substation switching misoperation according to claim 1 is characterized by: The construction system architecture includes adopting a hierarchical structure system; The structural system includes a data acquisition layer, a feature extraction layer, a knowledge transfer layer and a decision output layer. The data acquisition layer is responsible for the acquisition and preprocessing of multi-source heterogeneous data of the substation, the feature extraction layer realizes the automatic learning of deep features, the knowledge transfer layer completes the intelligent reuse of substation operation experience knowledge, and the decision output layer gives the final recognition results and evaluation suggestions.

3. The adaptive identification method for key factors of substation switching misoperation according to claim 2 is characterized by: The construction of the hierarchical semantic feature extraction model includes performing multimodal feature extraction on the text data of the substation operation ticket. The formula for the multimodal feature extraction is: S=D(X_text) D(X_text)=σ(W_t·E(X_text)+b_t) Among them, X_text represents the input substation operation bill text data, E(·) represents the word embedding mapping function, which is used to convert the operation text into a dense vector representation, W_t represents the weight matrix of text feature extraction, b_t represents the text feature extraction bias vector, and σ represents the nonlinear activation function.

4. The adaptive identification method for key factors of substation switching misoperation according to claim 3 is characterized by: The proposed time series feature extraction network based on improved LSTM includes extracting equipment status features from substation equipment status data. The formula for extracting equipment status features is: H(t)=LSTM(X_device(t),H(t-1)) f(t)=σ(W_d·[H(t-1),X_device(t)]+b_d) Among them, X_device(t) represents the substation equipment state input at time t, H(t) represents the equipment state hidden state vector, which is dynamically updated through the LSTM unit, and W_d and b_d represent the weight matrix and bias vector of equipment state feature extraction, respectively.

5. The method for adaptively identifying key factors of substation switching misoperation according to claim 4, characterized in that: The design of the dynamic fusion method based on the attention mechanism includes calculating the feature fusion mechanism, and the formula for calculating the feature fusion mechanism is: α_k=softmax(W_k·Q_k(K_k) T )V_k Z=σ(W_f·[α_text; α_device; α_human]+b_f) Among them, α_k represents the attention weights of various features of the substation, which is obtained through the interactive calculation of the query matrix Q_k, the key matrix K_k and the value matrix V_k. Z represents the fused feature representation, and the deep integration of features is achieved through the learnable parameter matrix W_f and the bias vector b_f.

6. The adaptive identification method for key factors of substation switching misoperation according to claim 5 is characterized by: The proposed substation operation knowledge distillation framework includes calculating the substation operation knowledge distillation framework according to the knowledge migration framework. The formula for calculating the substation operation knowledge distillation framework is: T(x) = softmax(g(f(x)) / τ) L=λ_1·L_CE(y,S(x))+λ_2·L_KL(T(x),S(x))+λ_3·L_reg Among them, T(x) represents the knowledge representation extracted by the substation expert model, τ is the temperature parameter, and the loss function includes the cross entropy loss L_CE, the knowledge distillation loss L_KL and the regularization term L_reg, which are balanced by the weight coefficients λ_1, λ_2, and λ_3.

7. The adaptive identification method for key factors of substation switching misoperation according to claim 6 is characterized by: The identification of substation switching includes calculating a quantitative evaluation method based on gradient accumulation according to factor evaluation, and the formula for calculating the quantitative evaluation method based on gradient accumulation is: S(f_i)=I(f_i) / ∑I(f_j) Among them, I(f_i) represents the influence of the substation misoperation factor f_i, which is obtained by calculating the gradient accumulation of the loss function on the factor; S(f_i) is the normalized importance score.

8. A system for adaptively identifying key factors of substation switching erroneous operation, based on the method for adaptively identifying key factors of substation switching erroneous operation according to any one of claims 1 to 7, characterized in that: include, The construction module constructs the system architecture, uses a hierarchical structure system to process the system architecture, and constructs a hierarchical semantic feature extraction model for the text data of the substation operation ticket; The computing module proposes a time series feature extraction network based on improved LSTM for substation equipment status data and designs a dynamic fusion method based on the attention mechanism; The identification module proposes a substation operation knowledge distillation framework and identifies substation switching based on quantitative evaluation of gradient accumulation.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for adaptively identifying key factors of substation switching erroneous operation according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for adaptively identifying key factors of substation switching erroneous operation according to any one of claims 1 to 7 are implemented.

Images