Multi-dimensional power failure accurate studying and judging method

Abstract

The invention relates to the field of power failure fault judgment, in particular to a multi-dimensional power failure accurate studying and judging method. The method comprises the following steps: performing semantic association analysis on a fault repair work order, manually issued power failure information, distribution transformer terminal power failure event data, terminal offline data, line branch switch data, current section curve data and 10kV line fault signal data by adopting an artificial intelligence technology based on deep learning, and performing information cross query verification based on multi-source data; and the type of the power failure is intelligently judged. Therefore, the accuracy and efficiency of power failure fault judgment can be effectively improved, the workload of related personnel is reduced, and more powerful technical support is provided for marketing customer service.

Description

Technical Field

[0001] The present application relates to the field of power outage fault diagnosis, and specifically to a multi-dimensional power outage accurate analysis method. Background Art

[0002] Extreme disasters, equipment aging, overload and other factors often lead to power system failures, which in turn cause power outages. Power outages not only affect residents' daily lives, but also cause significant losses to commercial activities and industrial production. In the face of sudden power outages, it is urgent to find out the problem through in-depth analysis and comprehensive review of the incident, so as to carry out relevant technical research in a targeted manner, improve the emergency response capabilities and response efficiency of marketing customer service, and reduce the workload of grassroots units.

[0003] Traditional methods for judging power outage faults mainly rely on manual experience and simple data analysis. These methods are not only inefficient, but also easily affected by human factors, resulting in insufficient accuracy and reliability of the judgment results. As the complexity of the power system increases, traditional judgment methods are unable to cope with multi-dimensional and multi-source data.

[0004] Therefore, an optimized multi-dimensional method for accurate assessment of power outages is desired. Summary of the invention

[0005] This application is made in view of the above problems. One purpose of this application is to provide a multi-dimensional power outage accurate assessment method.

[0006] The embodiment of the present application provides a multi-dimensional power outage accurate assessment method, which includes: Obtain fault repair work orders, manually publish power outage information, power distribution terminal power outage event data, terminal offline data, line branch switch data, current cross-section curve data and 10kV line fault signal data; The fault repair work order, the manually released power outage information, the power distribution terminal power outage event data, the terminal offline data, the line branch switch data, the current cross-section curve data and the 10kV line fault signal data are embedded in coding and semantic association optimization to obtain an optimized fault repair work order embedded coding vector, an optimized manually released power outage information embedded coding vector, an optimized distribution terminal power outage event embedded coding vector, an optimized terminal offline data embedded coding vector, an optimized line branch switch data embedded coding vector, an optimized current cross-section curve data embedded coding vector and an optimized line fault signal data embedded coding vector; The optimized fault repair work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current cross-section curve data embedded coding vector and the optimized line fault signal data embedded coding vector are cross-domain query matching coded to obtain a fault cross-dimensional query matching feature vector; Based on the fault cross-dimensional query matching feature vector, a fault judgment result is generated.

[0007] For example, according to the multi-dimensional power outage accurate judgment method of the embodiment of the present application, the fault repair work order, the manually released power outage information, the distribution terminal power outage event data, the terminal offline data, the line branch switch data, the current cross-section curve data and the 10kV line fault signal data are embedded coded and semantically associated optimized to obtain an optimized fault repair work order embedded coding vector, an optimized manually released power outage information embedded coding vector, an optimized distribution terminal power outage event embedded coding vector, an optimized terminal offline data embedded coding vector, an optimized line branch switch data embedded coding vector, an optimized current cross-section curve data embedded coding vector and an optimized line fault signal data embedded coding vector, including: The fault repair work order, the manually released power outage information, the power distribution terminal power outage event data, the terminal offline data, the line branch switch data, the current cross-section curve data and the 10kV line fault signal data are embedded and encoded to obtain a fault repair work order embedded coding vector, a manually released power outage information embedded coding vector, a power distribution terminal power outage event embedded coding vector, a terminal offline data embedded coding vector, a line branch switch data embedded coding vector, a current cross-section curve data embedded coding vector, and a line fault signal data embedded coding vector; The fault repair work order embedded coding vector, the manually released power outage information embedded coding vector, the distribution terminal power outage event embedded coding vector, the terminal offline data embedded coding vector, the line branch switch data embedded coding vector, the current cross-section curve data embedded coding vector and the line fault signal data embedded coding vector are input into a feature association optimization module based on sequence endogenous correlation analysis to obtain the optimized fault repair work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current cross-section curve data embedded coding vector and the optimized line fault signal data embedded coding vector.

[0008] For example, according to the multi-dimensional power outage precision judgment method of the embodiment of the present application, the fault repair work order is embedded in a coding vector, the manually released power outage information is embedded in a coding vector, the distribution terminal power outage event is embedded in a coding vector, the terminal offline data is embedded in a coding vector, the line branch switch data is embedded in a coding vector, the current cross-section curve data is embedded in a coding vector, and the line fault signal data is embedded in a coding vector. Input into a feature association optimization module based on sequence endogenous correlation analysis to obtain the optimized fault repair work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current cross-section curve data embedded coding vector, and the optimized line fault signal data embedded coding vector, including: Calculate the semantic association score vector between any two embedded coding vectors in the feature set consisting of the embedded coding vector of the fault repair work order, the embedded coding vector of the manually released power outage information, the embedded coding vector of the power distribution terminal power outage event, the embedded coding vector of the terminal offline data, the embedded coding vector of the line branch switch data, the embedded coding vector of the current cross-section curve data and the embedded coding vector of the line fault signal data to obtain a sequence of semantic association score vectors; Calculating the mean vector of the sequence of semantic association score vectors to obtain a global representation vector of the sequence endogenous correlation; Based on the global representation vector of the endogenous correlation of the sequence, respectively calculating the associated optimization factors of the embedded coding vector of the fault repair work order, the embedded coding vector of the manually released power outage information, the embedded coding vector of the power distribution terminal power outage event, the embedded coding vector of the terminal offline data, the embedded coding vector of the line branch switch data, the embedded coding vector of the current cross-section curve data and the embedded coding vector of the line fault signal data to obtain a sequence of associated optimization factors; Inputting the sequence of association optimization factors into a Softmax activation function to obtain a sequence of association optimization weight factors; Taking each associated optimization weight factor in the sequence of associated optimization weight factors as a weight, the fault report work order embedded coding vector, the manually released power outage information embedded coding vector, the distribution terminal power outage event embedded coding vector, the terminal offline data embedded coding vector, the line branch switch data embedded coding vector, the current cross-section curve data embedded coding vector and the line fault signal data embedded coding vector are weighted respectively to obtain the optimized fault report work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current cross-section curve data embedded coding vector and the optimized line fault signal data embedded coding vector.

[0009] For example, according to the multi-dimensional power outage accurate judgment method of the embodiment of the present application, the semantic association score vector between any two embedded coding vectors in the feature set composed of the embedded coding vector of the fault repair work order, the embedded coding vector of the manually released power outage information, the embedded coding vector of the power distribution terminal power outage event, the embedded coding vector of the terminal offline data, the embedded coding vector of the line branch switch data, the embedded coding vector of the current cross-section curve data and the embedded coding vector of the line fault signal data is calculated to obtain a sequence of semantic association score vectors, including: The fault repair work order embedding coding vector and the manually released power outage information embedding coding vector are cascaded and multiplied by a weight coefficient matrix, and then dotted with a bias vector to obtain the semantic association score vector.

[0010] For example, according to the multi-dimensional power outage accurate judgment method of the embodiment of the present application, the correlation optimization factor of the fault repair work order embedded coding vector is calculated based on the global representation vector of the sequence endogenous correlation, including: The fault repair work order embedding coding vector and the sequence endogenous correlation global representation vector are multiplied by different weight coefficient vectors respectively and then added to obtain a semantic correlation coefficient; The semantic association coefficient is added with a bias parameter and then passed through a sigmoid activation function to obtain the association optimization factor.

[0011] For example, according to the multi-dimensional power outage accurate judgment method of the embodiment of the present application, the optimized fault repair work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current cross-section curve data embedded coding vector and the optimized line fault signal data embedded coding vector are cross-domain query matching coded to obtain a fault cross-dimensional query matching feature vector, including: The optimized fault report work order embedded coding vector is used as the query feature vector, and the feature set composed of the optimized fault report work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current section curve data embedded coding vector and the optimized line fault signal data embedded coding vector is used as the sequence of library feature vectors, and the sequence of the query feature vector and the library feature vector is input into the cross-domain attention gate query encoder based on the key matrix local association enhancement to obtain the fault cross-dimensional query matching feature vector.

[0012] For example, according to the multi-dimensional power outage accurate judgment method of the embodiment of the present application, the sequence of the query feature vector and the library feature vector is input into the cross-domain attention gate query encoder based on the key matrix local association enhancement to obtain the fault cross-dimensional query matching feature vector, including: Performing a linear transformation on each library feature vector in the sequence of library feature vectors to obtain an initial key matrix, wherein each row vector in the initial key matrix is ​​a library feature vector after each linear transformation; Calculating feature distribution difference energy coefficients between the query feature vector and each row vector in the initial key matrix to obtain a sequence of feature distribution difference energy coefficients; Comparing each feature distribution difference energy coefficient in the sequence of feature distribution difference energy coefficients with a predetermined threshold, and in response to the feature distribution difference energy coefficient being less than the predetermined threshold, taking a mean vector between a row vector corresponding to the feature distribution difference energy coefficient and the query feature vector as an updated row vector to obtain an updated key matrix; Processing the query feature vector using a value embedding matrix to obtain a value vector; Using the query feature vector as a query vector and each row vector in the update key matrix as a key vector, the query vector, the key vector and the value vector are input into a cross-domain attention gate query module based on a Transformer structure to obtain a sequence of fault feature cross-domain query attention vectors; The position-wise mean vector of the sequence of the fault feature cross-domain query focus vector is calculated to obtain the fault cross-dimensional query matching feature vector.

[0013] For example, according to the multi-dimensional power outage accurate judgment method of the embodiment of the present application, the characteristic distribution difference energy coefficient between the query feature vector and each row vector in the initial key matrix is ​​calculated to obtain a sequence of characteristic distribution difference energy coefficients, including: Calculate the position point addition, position point subtraction and position point multiplication between the query feature vector and the row vector to obtain a first matching fusion result, a second matching fusion result and a third matching fusion result; After concatenating the first matching fusion result, the second matching fusion result, and the third matching fusion result into a multi-scale semantic matching vector of a fault feature, one-dimensional convolution processing and maximum pooling processing are performed on the vector to obtain a fault feature association representation vector; The energy coefficient of the fault feature association representation vector is calculated to obtain the feature distribution difference energy coefficient.

[0014] For example, according to the multi-dimensional power outage accurate judgment method of the embodiment of the present application, the energy coefficient of the fault feature association representation vector is calculated to obtain the feature distribution difference energy coefficient, including: The square of the difference between the maximum eigenvalue of the fault feature association representation vector and its eigenmean, the feature variance of the fault feature association representation vector and the weighted sum of the bias terms are calculated, and the sum of the feature variance of the fault feature association representation vector and the bias term is divided by the weighted sum to obtain the feature distribution difference energy coefficient.

[0015] For example, according to the multi-dimensional power outage accurate judgment method of the embodiment of the present application, the fault judgment result is generated based on the cross-dimensional query matching feature vector of the fault, including: The fault cross-dimensional query matching feature vector is input into a fault analysis result generator based on a decoder to obtain the fault judgment result.

[0016] According to the multi-dimensional power outage accurate judgment method of the embodiment of the present application, it uses artificial intelligence technology based on deep learning to perform semantic association analysis on fault repair work orders, manually released power outage information, distribution transformer terminal power outage event data, terminal offline data, line branch switch data, current cross-section curve data and 10kV line fault signal data, and intelligently judges the type of power outage fault based on information cross-query verification of multi-source data. In this way, the accuracy and efficiency of power outage fault judgment can be effectively improved, the workload of relevant personnel can be reduced, and more powerful technical support can be provided for marketing customer service. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments of the present application are briefly introduced below. Obviously, the drawings described below only relate to some embodiments of the present application, and are not intended to limit the present application.

[0018] Figure 1 A flow chart of a multi-dimensional power outage accurate judgment method in an embodiment of the present application is shown; Figure 2 A flowchart of sub-step S520 of the multi-dimensional power outage accurate judgment method in an embodiment of the present application is shown; Figure 3 A schematic diagram of the structure of a multi-dimensional power outage accurate analysis system in an embodiment of the present application is shown; Figure 4 An application scenario diagram of the multi-dimensional power outage accurate judgment method in an embodiment of the present application is shown; and Figure 5 The hierarchical star-row topology structure of the distribution network in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.

[0020] The terms used in this specification are those common terms currently widely used in the art in consideration of the functions of the present application, but these terms may vary according to the intentions of those skilled in the art, precedents, or new technologies in the art. In addition, specific terms may be selected, and in this case, their detailed meanings will be described in the detailed description of the present application. Therefore, the terms used in the specification should not be understood as simple names, but rather as a general description based on the meaning of the terms and the present application.

[0021] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0022] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. At the same time, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0023] In addition, the application architecture diagram in the embodiment of the present application is intended to more clearly illustrate the technical solution in the embodiment of the present application, and does not constitute a limitation on the technical solution provided in the embodiment of the present application. Of course, for other application architectures and business applications, the technical solution provided in the embodiment of the present application is also applicable to similar problems.

[0024] The following is a non-restrictive explanation of the multi-dimensional power outage accurate assessment method provided according to at least one embodiment of the present application through several examples or embodiments. As described below, different features in these specific examples or embodiments can be combined with each other without conflicting with each other to obtain new examples or embodiments, and these new examples or embodiments also fall within the scope of protection of the present application.

[0025] In response to the above technical problems, the technical concept of this application is to use artificial intelligence technology based on deep learning to perform semantic association analysis on fault repair work orders, manually released power outage information, distribution transformer terminal power outage event data, terminal offline data, line branch switch data, current cross-section curve data and 10kV line fault signal data, and intelligently determine the type of power outage fault based on information cross-query verification of multi-source data. In this way, the accuracy and efficiency of power outage fault judgment can be effectively improved, the workload of relevant personnel can be reduced, and more powerful technical support can be provided for marketing customer service.

[0026] Based on this, Figure 1 The flowchart of the multi-dimensional power outage accurate judgment method in the embodiment of the present application is shown. Figure 1As shown, according to the multi-dimensional power outage accurate judgment method of the embodiment of the present application, the steps include: S510, obtaining a fault repair work order, manually released power outage information, power distribution terminal power outage event data, terminal offline data, line branch switch data, current cross-section curve data and 10kV line fault signal data; S520, embedding coding and semantic association optimization of the fault repair work order, the manually released power outage information, the power distribution terminal power outage event data, the terminal offline data, the line branch switch data, the current cross-section curve data and the 10kV line fault signal data to obtain an optimized fault repair work order embedded coding vector, an optimized manually released power outage information embedded coding vector, an optimized distribution terminal power outage event embedded coding vector, and an optimized terminal The offline data is embedded in the coding vector, the line branch switch data is embedded in the coding vector, the current cross-section curve data is embedded in the coding vector and the line fault signal data is embedded in the coding vector; S530, the optimized fault repair work order is embedded in the coding vector, the optimized manually released power outage information is embedded in the coding vector, the optimized distribution terminal power outage event is embedded in the coding vector, the optimized terminal offline data is embedded in the coding vector, the optimized line branch switch data is embedded in the coding vector, the optimized current cross-section curve data is embedded in the coding vector and the optimized line fault signal data is embedded in the coding vector for cross-domain query matching coding to obtain a fault cross-dimensional query matching feature vector; S540, based on the fault cross-dimensional query matching feature vector, a fault judgment result is generated.

[0027] Specifically, in the technical solution of the present application, first, the fault reporting work order, manually released power outage information, power distribution terminal power outage event data, terminal offline data, line branch switch data, current cross-sectional curve data and 10kV line fault signal data are obtained. It should be understood that the fault reporting work order provides the actual power outage reported by the user, the manually released power outage information provides the official power outage notice, the power distribution terminal power outage event data records the power outage event detected by the distribution terminal, the terminal offline data reflects the operating status of the terminal equipment, the line branch switch data reveals the operating status of the line switch, the current cross-sectional curve data provides detailed information on current changes, and the 10kV line fault signal data records the characteristic signals of the line fault. In the technical solution of the present application, by collecting data from multiple sources, different aspects and possible causes of power outage events can be fully covered, which helps to form a more complete and accurate power outage fault portrait.

[0028] Next, considering that different types of data usually have different formats, structures and meanings, in order to integrate these multi-source data into a unified analysis framework, in the technical solution of the present application, the above-mentioned acquired data are further embedded and encoded, and the data are mapped into a unified semantic feature space, and the content semantic features of the data are extracted, and the data of different formats and structures are converted into a unified numerical representation to obtain the embedded coding vector of the fault repair work order, the embedded coding vector of the manually released power outage information, the embedded coding vector of the power distribution terminal power outage event, the embedded coding vector of the terminal offline data, the embedded coding vector of the line branch switch data, the embedded coding vector of the current cross-section curve data, and the embedded coding vector of the line fault signal data. In the embodiment of the present application, the BERT model or the Word2Vec model can be used for embedded coding.

[0029] Then, considering that each data is independent of each other, there are often inherent correlations and mutual influences among them. For example, the fault repair work order usually contains the user's description of the power outage phenomenon, and these descriptions may have a certain correlation with the power outage event data detected by the distribution terminal. Therefore, in order to effectively mine the intrinsic connection between the data and optimize the data feature representation, this application introduces a feature association optimization module based on sequence endogenous correlation analysis to perform semantic association analysis and feature optimization processing on the fault repair work order embedded coding vector, the manually released power outage information embedded coding vector, the distribution terminal power outage event embedded coding vector, the terminal offline data embedded coding vector, the line branch switch data embedded coding vector, the current cross-section curve data embedded coding vector and the line fault signal data embedded coding vector. Specifically, the module first quantifies the semantic similarity or difference between any two data by calculating the semantic association between the two, so as to reveal the overall semantic structure of each data and generate a set of semantic association score vectors. Next, the aggregate information of the set of semantic association score vectors is used as the overall context information, and the association optimization factor of each data is calculated to quantify the consistency or deviation between the semantic features describing each data and the overall semantic structure of the data set, and this is used as a guide for feature association optimization. Furthermore, after the association optimization factor is normalized, it is used as a weight coefficient to perform weighted optimization on the embedded coding vector of the corresponding data, so that the overall context information is integrated into the semantic features of each data, so that it can more accurately reflect the characteristics of the power outage event, thereby obtaining the optimized fault repair work order embedded coding vector, the optimized manual power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current cross-section curve data embedded coding vector, and the optimized line fault signal data embedded coding vector.

[0030] Accordingly, in step S520, if Figure 2 As shown, the fault repair work order, the manually released power outage information, the power distribution terminal power outage event data, the terminal offline data, the line branch switch data, the current cross-section curve data and the 10kV line fault signal data are embedded in coding and semantic association optimization to obtain an optimized fault repair work order embedded coding vector, an optimized manually released power outage information embedded coding vector, an optimized power distribution terminal power outage event embedded coding vector, an optimized terminal offline data embedded coding vector, an optimized line branch switch data embedded coding vector, an optimized current cross-section curve data embedded coding vector and an optimized line fault signal data embedded coding vector, including: S521, the fault repair work order, the manually released power outage information, the power distribution terminal power outage event data, the terminal offline data, the line branch switch data, the current cross-section curve data and the 10kV line fault signal data are embedded in coding to obtain an optimized fault repair work order embedded coding vector, an optimized manually released power outage information embedded coding vector, a power distribution terminal power outage event data, the terminal offline data, the line branch switch data, the current cross-section curve data and the 10kV line fault signal data The electrical event embedding coding vector, the terminal offline data embedding coding vector, the line branch switch data embedding coding vector, the current cross-section curve data embedding coding vector, and the line fault signal data embedding coding vector; S522, the fault repair work order embedding coding vector, the manual power outage information embedding coding vector, the distribution terminal power outage event embedding coding vector, the terminal offline data embedding coding vector, the line branch switch data embedding coding vector, the current cross-section curve data embedding coding vector and the line fault signal data embedding coding vector are input into the feature association optimization module based on sequence endogenous correlation analysis to obtain the optimized fault repair work order embedding coding vector, the optimized manual power outage information embedding coding vector, the optimized distribution terminal power outage event embedding coding vector, the optimized terminal offline data embedding coding vector, the optimized line branch switch data embedding coding vector, the optimized current cross-section curve data embedding coding vector and the optimized line fault signal data embedding coding vector.

[0031] In step S522, the fault repair work order embedded coding vector, the manually released power outage information embedded coding vector, the distribution terminal power outage event embedded coding vector, the terminal offline data embedded coding vector, the line branch switch data embedded coding vector, the current cross-section curve data embedded coding vector and the line fault signal data embedded coding vector are input into a feature association optimization module based on sequence endogenous correlation analysis to obtain the optimized fault repair work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, and the optimized line branch switch data embedded coding vector. , the optimized current cross-section curve data embedding coding vector and the optimized line fault signal data embedding coding vector, including: calculating the semantic association score vector between any two embedded coding vectors in the feature set composed of the fault repair work order embedding coding vector, the manually released power outage information embedding coding vector, the distribution terminal power outage event embedding coding vector, the terminal offline data embedding coding vector, the line branch switch data embedding coding vector, the current cross-section curve data embedding coding vector and the line fault signal data embedding coding vector to obtain a sequence of semantic association score vectors; calculating the mean vector of the sequence of semantic association score vectors to obtain a global representation vector of the sequence endogenous correlation; Based on the global representation vector of the endogenous correlation of the sequence, the associated optimization factors of the fault repair work order embedding coding vector, the manually released power outage information embedding coding vector, the distribution terminal power outage event embedding coding vector, the terminal offline data embedding coding vector, the line branch switch data embedding coding vector, the current cross-section curve data embedding coding vector and the line fault signal data embedding coding vector are respectively calculated to obtain a sequence of associated optimization factors; the sequence of associated optimization factors is input into the Softmax activation function to obtain a sequence of associated optimization weight factors; each associated optimization weight factor in the sequence of associated optimization weight factors is used as a weight to embed the fault repair work order. The coding vector, the manually released power outage information embedded coding vector, the distribution terminal power outage event embedded coding vector, the terminal offline data embedded coding vector, the line branch switch data embedded coding vector, the current cross-section curve data embedded coding vector and the line fault signal data embedded coding vector are weighted to obtain the optimized fault report work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current cross-section curve data embedded coding vector and the optimized line fault signal data embedded coding vector.

[0032] Specifically, the semantic association score vector between any two embedded coding vectors in the feature set composed of the embedded coding vector of the fault report work order, the embedded coding vector of the manually released power outage information, the embedded coding vector of the distribution terminal power outage event, the embedded coding vector of the terminal offline data, the embedded coding vector of the line branch switch data, the embedded coding vector of the current cross-section curve data and the embedded coding vector of the line fault signal data is calculated to obtain a sequence of semantic association score vectors, including: cascading the embedded coding vector of the fault report work order and the embedded coding vector of the manually released power outage information, multiplying them by a weight coefficient matrix, and then dot-adding them with a bias vector to obtain the semantic association score vector.

[0033] Among them, based on the global representation vector of the sequence intrinsic correlation, the association optimization factor of the fault repair work order embedded coding vector is calculated, including: multiplying the fault repair work order embedded coding vector and the global representation vector of the sequence intrinsic correlation by different weight coefficient vectors and then performing addition operations to obtain semantic association coefficients; adding a bias parameter to the semantic association coefficient and then passing it through a sigmoid activation function to obtain the association optimization factor.

[0034] In one example, the fault repair work order embedded coding vector, the manually released power outage information embedded coding vector, the distribution terminal power outage event embedded coding vector, the terminal offline data embedded coding vector, the line branch switch data embedded coding vector, the current cross-section curve data embedded coding vector and the line fault signal data embedded coding vector are input into a feature association optimization module based on sequence endogenous correlation analysis to obtain the optimized fault repair work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current cross-section curve data embedded coding vector and the optimized line fault signal data embedded coding vector, including: The following semantic feature association optimization formula processes the fault repair work order embedded coding vector, the manually released power outage information embedded coding vector, the distribution terminal power outage event embedded coding vector, the terminal offline data embedded coding vector, the line branch switch data embedded coding vector, the current cross-section curve data embedded coding vector and the line fault signal data embedded coding vector to obtain the optimized fault repair work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current cross-section curve data embedded coding vector and the optimized line fault signal data embedded coding vector, wherein the semantic feature association optimization formula is: in, represents a feature set consisting of the fault repair work order embedded coding vector, the manually released power outage information embedded coding vector, the distribution terminal power outage event embedded coding vector, the terminal offline data embedded coding vector, the line branch switch data embedded coding vector, the current cross-section curve data embedded coding vector and the line fault signal data embedded coding vector, , , , and Respectively represent the first, second, and , and embedding code vector, Indicates cascade operation, is the weight coefficient matrix, is the bias vector, is the first in the sequence of the semantic association score vector semantic association score vectors, The value of is the number of vectors in the sequence of the semantic association score vectors, is the global representation vector of the endogenous correlation of the sequence, and Represent different weight coefficient vectors, is the bias parameter, is the sigmoid activation function, is the normalized exponential function, For the The associated optimization factor, The first of the feature sets consisting of the optimized fault repair work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current cross-section curve data embedded coding vector and the optimized line fault signal data embedded coding vector is represented. Embedding code vector.

[0035] Secondly, the multi-source data is further cross-queried and verified to further explore the potential causes of the power outage event and improve the accuracy and reliability of the power outage fault judgment. In the technical solution of the present application, it is considered that the fault repair work order is usually initiated directly by the user, which directly reflects the user's power outage experience and fault situation. Therefore, the present application adopts a fault information query and verification method based on user feedback, using the optimized fault repair work order embedded coding vector as the query feature vector, and the optimized fault repair work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current cross-section curve data embedded coding vector and the optimized line fault signal data embedded coding vector as the sequence of library feature vectors, and the query feature vector and the sequence of the library feature vector are input into the cross-domain attention gate query encoder based on the key matrix local association enhancement for information cross-query verification to improve the pertinence and accuracy of the judgment. Specifically, the encoder first performs a linear transformation on each library feature vector to adjust its feature representation dimension so that it is aligned with the feature dimension of the query feature vector to construct an initial key matrix. Then, the query feature vector is used as a common soft anchor reference point, and the local correlation strength contained between each key vector is revealed by calculating the feature distribution difference energy coefficient between the query feature vector and each key vector in the initial key matrix. Secondly, for the key vector corresponding to the feature distribution difference energy coefficient less than a predetermined threshold, its feature representation is updated using its mean vector with the query feature vector, thereby further strengthening the library feature representation that is more similar to the query feature, and generating an updated key matrix to improve the pertinence and matching quality of the key matrix. Furthermore, a query vector and a value vector are constructed based on the query feature vector, and each row vector in the updated key matrix is ​​used as a key vector. Through the self-attention calculation of the Transformer structure, cross-domain information query and integration are realized to obtain a cross-domain query attention vector between the fault repair work order and each data. Finally, the position-based mean of each cross-domain query attention vector is calculated to integrate the matching information of all data and generate a fault cross-dimensional query matching feature vector.

[0036] Correspondingly, in step S530, the optimized fault repair work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current section curve data embedded coding vector and the optimized line fault signal data embedded coding vector are cross-domain query matching coding to obtain a fault cross-dimensional query matching feature vector, including: using the optimized fault repair work order embedded coding vector as a query feature vector, and using the feature set composed of the optimized fault repair work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current section curve data embedded coding vector and the optimized line fault signal data embedded coding vector as a sequence of library feature vectors, and inputting the query feature vector and the sequence of library feature vectors into a cross-domain attention gate query encoder based on key matrix local association enhancement to obtain the fault cross-dimensional query matching feature vector.

[0037] The sequence of the query feature vector and the library feature vector is input into a cross-domain attention gate query encoder based on the local association enhancement of the key matrix to obtain the fault cross-dimensional query matching feature vector, including: performing a linear transformation on each library feature vector in the sequence of the library feature vector to obtain an initial key matrix, wherein each row vector in the initial key matrix is ​​a library feature vector after each linear transformation; calculating a feature distribution difference energy coefficient between the query feature vector and each row vector in the initial key matrix to obtain a sequence of feature distribution difference energy coefficients; comparing each feature distribution difference energy coefficient in the sequence of feature distribution difference energy coefficients with a predetermined threshold, and in response to the feature distribution difference energy coefficients, performing a linear transformation on each library feature vector in the sequence of the library feature vectors to obtain a sequence of feature distribution difference energy coefficients; comparing each feature distribution difference energy coefficient in the sequence of the feature distribution difference energy coefficients with a predetermined threshold, and The distribution difference energy coefficient is less than a predetermined threshold, and the row vector corresponding to the feature distribution difference energy coefficient and the mean vector between the query feature vector are used as the updated row vector to obtain an updated key matrix; the query feature vector is processed using a value embedding matrix to obtain a value vector; the query feature vector is used as the query vector and each row vector in the updated key matrix is ​​used as a key vector, and the query vector, the key vector and the value vector are input into a cross-domain attention gate query module based on a Transformer structure to obtain a sequence of fault feature cross-domain query attention vectors; the position-wise mean vector of the sequence of the fault feature cross-domain query attention vectors is calculated to obtain the fault cross-dimensional query matching feature vector.

[0038] Specifically, the feature distribution difference energy coefficients between the query feature vector and each row vector in the initial key matrix are calculated to obtain a sequence of feature distribution difference energy coefficients, including: calculating the position point addition, position point subtraction and position point multiplication between the query feature vector and the row vector to obtain a first matching fusion result, a second matching fusion result and a third matching fusion result; after cascading the first matching fusion result, the second matching fusion result and the third matching fusion result into a fault feature multi-scale semantic matching vector, one-dimensional convolution processing and maximum pooling processing are performed on them to obtain a fault feature association representation vector; and the energy coefficient of the fault feature association representation vector is calculated to obtain the feature distribution difference energy coefficient.

[0039] More specifically, calculating the energy coefficient of the fault feature association representation vector to obtain the feature distribution difference energy coefficient includes: calculating the square of the difference between the maximum eigenvalue of the fault feature association representation vector and its eigenmean, the feature variance of the fault feature association representation vector, and the weighted sum of the bias terms, and dividing the sum of the feature variance of the fault feature association representation vector and the bias term by the weighted sum to obtain the feature distribution difference energy coefficient.

[0040] In one example, the sequence of the query feature vector and the library feature vector is input into a cross-domain attention gate query encoder based on key matrix local association enhancement to obtain the fault cross-dimensional query matching feature vector, including: processing the sequence of the query feature vector and the library feature vector with the following cross-domain query matching formula to obtain the fault cross-dimensional query matching feature vector, wherein the cross-domain query matching formula is: in, represents a sequence of library feature vectors, , , and are the first, second, and and library feature vectors, The value of is the number of feature vectors in the sequence of the library feature vectors, is the linear transformation matrix, , and are different bias terms, represents the initial bond matrix, , , and are the first, second, and and row vectors, i.e., the library feature vectors after linear transformation, represents the query feature vector, represents dot product, Indicates point addition, Indicates point reduction, Indicates cascade operation, represents the maximum pooling operation, represents a one-dimensional convolution operation, Indicates Fault feature association representation vector, , and Respectively represent the The characteristic variance, maximum eigenvalue and characteristic mean of the fault feature association vector are is the first in the sequence of the characteristic distribution difference energy coefficients The characteristic distribution difference energy coefficient, is the predetermined threshold, represents the updated key matrix, represents the first row vectors, is the value embedding matrix, is a value vector, is the normalized exponential function, The value of is the length of the row vector in the update key matrix, represents the transpose of a vector, It is Fault feature cross-domain query attention vector, Represents the fault cross-dimensional query matching feature vector.

[0041] Then, the fault cross-dimensional query matching feature vector is input into a fault analysis result generator based on a decoder to obtain the fault judgment result. In the technical solution of the present application, the decoder performs deep learning and feature extraction on the fault cross-dimensional query matching feature vector through a series of neural network layers to fully understand the multi-dimensional power outage event information contained in the fault cross-dimensional query matching feature vector, and then gradually decodes and generates corresponding fault judgment results, including fault type (such as distribution transformer fault, user single-household fault), fault impact range and possible fault causes, etc.

[0042] Accordingly, in step S540, a fault judgment result is generated based on the fault cross-dimensional query matching feature vector, including: inputting the fault cross-dimensional query matching feature vector into a fault analysis result generator based on a decoder to obtain the fault judgment result.

[0043] In a preferred example, considering the optimization of fault reporting work order embedded coding vector, optimization of manually released power outage information embedded coding vector, optimization of distribution terminal power outage event embedded coding vector, optimization of terminal offline data embedded coding vector, optimization of line branch switch data embedded coding vector, optimization of current cross-section curve data embedded coding vector and optimization of line fault signal data embedded coding vector, they respectively represent the contextual semantic association enhancement features of fault reporting work order, manually released power outage information, distribution terminal power outage event data, terminal offline data, line branch switch data, current cross-section curve data and 10kV line fault signal data. Therefore, when the optimized fault repair work order embedded coding vector is used as the query feature vector and the feature set composed of the optimized fault repair work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current section curve data embedded coding vector and the optimized line fault signal data embedded coding vector is used as a sequence of library feature vectors, and a cross-domain attention gate query encoding based on key matrix local association enhancement is performed, due to the difference in the local association enhancement of the semantic feature key matrix of heterogeneous data and the superposition of the cross-domain attention gate query encoding difference, the fault cross-dimensional query matching feature vector will have a diversified set expression distribution of the encoded semantic features. Therefore, it is expected to improve the balance between the regression mapping accuracy and completeness when the fault cross-dimensional query matching feature vector is input into the fault analysis result generator based on the decoder for decoding regression, thereby improving the accuracy of the obtained fault judgment result.

[0044] Based on this, in the preferred example, the fault cross-dimensional query matching feature vector is input into the fault analysis result generator based on the decoder to obtain the fault judgment result, which specifically includes: determining the fault cross-dimensional query matching maximum eigenvalue and the fault cross-dimensional query matching minimum eigenvalue of the fault cross-dimensional query matching feature vector, and calculating the fault cross-dimensional query matching mean and the fault cross-dimensional query matching standard deviation of the feature set of the fault cross-dimensional query matching feature vector; calculating the quotient of the fault cross-dimensional query matching mean and the fault cross-dimensional query matching standard deviation to obtain the fault cross-dimensional query matching statistical normalization value; calculating the inverse of each eigenvalue of the fault cross-dimensional query matching feature vector, and comparing it with the fault cross-dimensional query matching maximum eigenvalue and the fault cross-dimensional query matching standard deviation. After multiplying the difference of the minimum eigenvalues ​​of the fault cross-dimensional query matching, subtract the point-wise difference from the fault cross-dimensional query matching statistical normalized value to obtain the fault cross-dimensional query matching distribution approximation vector; calculate an exponential function with a natural constant as the base, with each eigenvalue of the fault cross-dimensional query matching distribution approximation vector as the exponent, and add the point-wise difference with the fault cross-dimensional query matching statistical normalized value to obtain the fault cross-dimensional query matching probability approximation vector; calculate the base-two logarithm of the absolute value of each eigenvalue of the fault cross-dimensional query matching probability approximation vector to obtain the optimized fault cross-dimensional query matching feature vector; input the optimized fault cross-dimensional query matching feature vector into the decoder-based fault analysis result generator to obtain the fault judgment result.

[0045] Among them, the fault cross-dimensional query matching feature vector is recorded as The optimization process is expressed as follows: in, is the fault cross-dimensional query matching feature vector, , , and are the feature sets of the fault cross-dimensional query matching feature vectors The mean, standard deviation, minimum and maximum values ​​of Indicates point reduction, represents dot product, Indicates point addition, To calculate the inverse of each eigenvalue of the fault cross-dimensional query matching eigenvector, is an exponential operation of a value, wherein the exponential operation of the value represents the calculation of the value of a natural exponential function with the value as a power, and is the logarithmic function operation with base 2, is the optimized fault cross-dimensional query matching feature vector.

[0046] That is, by performing random statistical normalization on the diversified feature set of the fault cross-dimensional query matching feature vector, a standardized continuous probability density distribution is approximated for the response hypothesis test of the confidence space constructed based on the overall eigenvalue of the fault cross-dimensional query matching feature vector relative to each eigenvalue of the fault cross-dimensional query matching feature vector, thereby establishing the target reachability of the discrete feature distribution of the fault cross-dimensional query matching feature vector to a unified regression target, so as to achieve a balanced executable performance between mapping accuracy and mapping completeness in the decoding regression process based on the discrete feature distribution of the fault cross-dimensional query matching feature vector, and improve the accuracy of the fault judgment result obtained by the fault analysis result generator based on the decoder input of the fault cross-dimensional query matching feature vector.

[0047] Based on the above embodiments, see Figure 3 As shown, it is a structural schematic diagram of a multi-dimensional power outage precision analysis system 800 in an embodiment of the present application. The multi-dimensional power outage precision analysis system 800 includes: a data acquisition module 810, which is used to obtain fault repair work orders, manually released power outage information, power distribution terminal power outage event data, terminal offline data, line branch switch data, current cross-section curve data and 10kV line fault signal data; a coding optimization module 820, which is used to embed coding and semantically optimize the fault repair work order, the manually released power outage information, the power distribution terminal power outage event data, the terminal offline data, the line branch switch data, the current cross-section curve data and the 10kV line fault signal data to obtain an optimized fault repair work order embedded coding vector, an optimized manually released power outage information embedded coding vector, an optimized distribution terminal power outage event embedded coding vector, and an optimized terminal offline data embedded coding vector. Vector, optimized line branch switch data embedded coding vector, optimized current cross-section curve data embedded coding vector and optimized line fault signal data embedded coding vector; a cross-domain query matching coding module 830, used to perform cross-domain query matching coding on the optimized fault repair work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current cross-section curve data embedded coding vector and the optimized line fault signal data embedded coding vector to obtain a fault cross-dimensional query matching feature vector; a fault judgment result generation module 840, used to generate a fault judgment result based on the fault cross-dimensional query matching feature vector.

[0048] Here, those skilled in the art can understand that the specific functions and operations of each module in the multi-dimensional power outage accurate judgment system 800 have been described in the above reference. Figure 1 to Figure 2The multi-dimensional power outage accurate judgment method is introduced in detail in the description of the multi-dimensional power outage accurate judgment method, and therefore, its repeated description will be omitted.

[0049] Figure 4 FIG. 1 is an application scenario diagram of the multi-dimensional power outage accurate judgment method according to an embodiment of the present application. Figure 4 As shown, in this application scenario, first, the fault repair work order, manual power outage information, power distribution terminal power outage event data, terminal offline data, line branch switch data, current cross-section curve data and 10kV line fault signal data are obtained (for example, Figure 4 D), then, the fault repair work order, the manually released power outage information, the power distribution terminal power outage event data, the terminal offline data, the line branch switch data, the current cross-section curve data and the 10kV line fault signal data are input to a server (for example, Figure 4 In S) shown in , the server can use the multi-dimensional power outage precise analysis algorithm to process the fault reporting work order, the manually released power outage information, the distribution terminal power outage event data, the terminal offline data, the line branch switch data, the current cross-section curve data and the 10kV line fault signal data to obtain a fault judgment result.

[0050] Furthermore, the present application constructs a multi-dimensional power outage precision assessment model. It should be understood that the construction of a multi-dimensional power outage precision assessment model is to improve the prediction and evaluation accuracy of power outage events by integrating information from multiple data sources. The construction of the model specifically includes the following steps: Multi-source data verification: Verify data from different data sources, including fault repair work orders, manually released power outage information, distribution transformer terminal power outage event data, terminal offline data, line branch switch data, current cross-section curve data and 10kV line fault signal data, etc., to ensure the accuracy and consistency of the data; Cross-validation: By comparing and analyzing the same or related information provided by different data sources, potential data inconsistencies or omissions are discovered and processed accordingly; Remove duplicate data: In multiple data sources, there may be duplicate records or duplicate data that affect the accuracy and performance of the model. Therefore, before building the power outage model, the data needs to be deduplicated to ensure the uniqueness of each record; Collect equipment online status information: Collect equipment is a key component for monitoring and controlling power networks. By obtaining the online status information of the collection equipment, it can be judged whether the equipment is operating normally and potential faults or problems can be discovered in time; Active call information: The active call of low-voltage meters means that the meter equipment actively sends data to the system to provide information about power consumption, voltage, current, etc. By obtaining the active call information of low-voltage meters, more accurate power consumption data and power load information can be obtained, thereby improving the accuracy of the power outage model; Intelligent outbound calling: Intelligent outbound calling is a technology that uses automated robots to make telephone calls. Through RPA intelligent outbound calling, users are actively contacted to obtain real-time information and feedback about power outages, thereby supplementing and verifying the information provided by other data sources; Operator equipment online data: Operators cooperating with power companies may provide online data of related equipment to verify and supplement the data collected by the power supply service command platform, thereby improving the reliability of the model.

[0051] By integrating and validating these different data sources, the power outage model can more accurately predict and evaluate power outage events. The verification and cross-validation of these data ensures the accuracy and consistency of the data, while the use of online status information of equipment, active call information, intelligent outbound calls, and online data of operator equipment can increase the reliability and comprehensiveness of the data. Ultimately, the combination of these steps will help build a more accurate and reliable power outage precision judgment model.

[0052] Hierarchical star topology is a commonly used mathematical model for studying radial network systems. The hierarchical star structure network is divided into multiple levels, with the first level being the central node and the nodes radiating from the parent node. For radial distribution networks, the hierarchical star topology can be used to establish its structural model and study its fault diagnosis strategy.

[0053] Furthermore, in the distribution network emergency repair command fault analysis system, the distribution network fault is divided into five levels for analysis: manual release of power outage information, power outage events at distribution transformer terminals, dispatching automation system monitoring, distribution self-system monitoring, and manual investigation (intelligent outbound call) data. Establish a hierarchical star-line topology structure for the distribution network. Figure 5 shown. Figure 5 In the example, the central node represents the feeder trunk line, and the trunk line is composed of As a part of the trunk line, the branch line is a secondary node of the corresponding trunk line, where the element is the branch line corresponding to the trunk line, and the element is a set of The following three layers are the distribution transformers, low-voltage lines and low-voltage customers corresponding to the upper-level distribution equipment, which are combined , Indicates that each parent node corresponds to a set of child nodes in the next layer; similarly, each parent node corresponds to a set of child nodes in the next layer; similarly, each set of child nodes corresponds to a parent node in the previous layer.

[0054] If the device operation information of the corresponding node is power failure, then On the contrary, if the device corresponding to the node being tested is not powered off, then Through the connection relationship between upper and lower nodes, the fault information of the node can be verified, and the parent node The fault information is , and its corresponding set of all child nodes is , Fault information vector , can be judged or verified according to the formula .

[0055] like , then it is judged to be a distribution transformer fault. Node The corresponding user single household failure.

[0056] Based on the above embodiments, another exemplary embodiment of an electronic device is also provided in the embodiments of the present application. In some possible implementations, the electronic device in the embodiments of the present application may include a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor can implement the steps of the multi-dimensional power outage accurate judgment method in the above embodiments when executing the program.

[0057] The embodiment of the present application also provides a computer-readable storage medium, on which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, the multi-dimensional power outage precision judgment method according to the embodiment of the present application described with reference to the above figures can be executed. The computer-readable storage medium includes, but is not limited to, for example, volatile memory and / or non-volatile memory. The volatile memory may, for example, include a random access memory (RAM) and / or a cache memory (cache), etc. The non-volatile memory may, for example, include a read-only memory (ROM), a hard disk, a flash memory, etc.

[0058] The embodiment of the present application also provides a computer program product or a computer program, which includes computer executable instructions, and the computer executable instructions are stored in a computer readable storage medium. The processor of the computer device reads the computer executable instructions from the computer readable storage medium, and the processor executes the computer executable instructions, so that the computer device executes the multi-dimensional power outage accurate judgment method according to the embodiment of the present application.

[0059] Those skilled in the art will appreciate that the contents disclosed in this application may be subject to various variations and improvements. For example, the various devices or components described above may be implemented by hardware, or by software, firmware, or a combination of some or all of the three.

[0060] In addition, although the present application makes various references to certain units in the system according to embodiments of the present application, any number of different units can be used and run on the client and / or server. The units are only illustrative, and different aspects of the system and method can use different units.

[0061] Those skilled in the art will appreciate that all or part of the steps in the above method can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, such as a read-only memory, a disk or an optical disk. Optionally, all or part of the steps in the above embodiment can also be implemented using one or more integrated circuits. Accordingly, each module / unit in the above embodiment can be implemented in the form of hardware or in the form of software function modules. The present application is not limited to any particular form of combination of hardware and software.

[0062] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this application belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an idealized or extremely formal sense, unless explicitly defined as such herein.

[0063] The above is an explanation of the present application and should not be considered as limiting thereof. Although several exemplary embodiments of the present application have been described, those skilled in the art will readily appreciate that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present application.

Claims

1. A multi-dimensional power outage accurate judgment method, characterized in that: include: Obtain fault repair work orders, manually publish power outage information, power distribution terminal power outage event data, terminal offline data, line branch switch data, current cross-section curve data and 10kV line fault signal data; The fault repair work order, the manually released power outage information, the power distribution terminal power outage event data, the terminal offline data, the line branch switch data, the current cross-section curve data and the 10kV line fault signal data are embedded in coding and semantic association optimization to obtain an optimized fault repair work order embedded coding vector, an optimized manually released power outage information embedded coding vector, an optimized distribution terminal power outage event embedded coding vector, an optimized terminal offline data embedded coding vector, an optimized line branch switch data embedded coding vector, an optimized current cross-section curve data embedded coding vector and an optimized line fault signal data embedded coding vector; The optimized fault repair work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current cross-section curve data embedded coding vector and the optimized line fault signal data embedded coding vector are cross-domain query matching coded to obtain a fault cross-dimensional query matching feature vector; Based on the fault cross-dimensional query matching feature vector, a fault judgment result is generated.

2. The multi-dimensional power outage accurate judgment method according to claim 1 is characterized in that: The fault repair work order, the manually released power outage information, the power distribution terminal power outage event data, the terminal offline data, the line branch switch data, the current cross-section curve data and the 10kV line fault signal data are embedded and optimized for semantic association to obtain an optimized fault repair work order embedded coding vector, an optimized manually released power outage information embedded coding vector, an optimized distribution terminal power outage event embedded coding vector, an optimized terminal offline data embedded coding vector, an optimized line branch switch data embedded coding vector, an optimized current cross-section curve data embedded coding vector and an optimized line fault signal data embedded coding vector, including: The fault repair work order, the manually released power outage information, the power distribution terminal power outage event data, the terminal offline data, the line branch switch data, the current cross-section curve data and the 10kV line fault signal data are embedded and encoded to obtain a fault repair work order embedded coding vector, a manually released power outage information embedded coding vector, a power distribution terminal power outage event embedded coding vector, a terminal offline data embedded coding vector, a line branch switch data embedded coding vector, a current cross-section curve data embedded coding vector, and a line fault signal data embedded coding vector; The fault repair work order embedded coding vector, the manually released power outage information embedded coding vector, the distribution terminal power outage event embedded coding vector, the terminal offline data embedded coding vector, the line branch switch data embedded coding vector, the current cross-section curve data embedded coding vector and the line fault signal data embedded coding vector are input into a feature association optimization module based on sequence endogenous correlation analysis to obtain the optimized fault repair work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current cross-section curve data embedded coding vector and the optimized line fault signal data embedded coding vector.

3. The multi-dimensional power outage accurate judgment method according to claim 2 is characterized in that: The fault repair work order embedding coding vector, the manually released power outage information embedding coding vector, the distribution terminal power outage event embedding coding vector, the terminal offline data embedding coding vector, the line branch switch data embedding coding vector, the current cross-section curve data embedding coding vector and the line fault signal data embedding coding vector are input into a feature association optimization module based on sequence endogenous correlation analysis to obtain the optimized fault repair work order embedding coding vector, the optimized manually released power outage information embedding coding vector, the optimized distribution terminal power outage event embedding coding vector, the optimized terminal offline data embedding coding vector, the optimized line branch switch data embedding coding vector, the optimized current cross-section curve data embedding coding vector and the optimized line fault signal data embedding coding vector, including: Calculate the semantic association score vector between any two embedded coding vectors in the feature set consisting of the embedded coding vector of the fault repair work order, the embedded coding vector of the manually released power outage information, the embedded coding vector of the power distribution terminal power outage event, the embedded coding vector of the terminal offline data, the embedded coding vector of the line branch switch data, the embedded coding vector of the current cross-section curve data and the embedded coding vector of the line fault signal data to obtain a sequence of semantic association score vectors; Calculating the mean vector of the sequence of semantic association score vectors to obtain a global representation vector of the sequence endogenous correlation; Based on the global representation vector of the endogenous correlation of the sequence, respectively calculating the associated optimization factors of the embedded coding vector of the fault repair work order, the embedded coding vector of the manually released power outage information, the embedded coding vector of the power distribution terminal power outage event, the embedded coding vector of the terminal offline data, the embedded coding vector of the line branch switch data, the embedded coding vector of the current cross-section curve data and the embedded coding vector of the line fault signal data to obtain a sequence of associated optimization factors; Inputting the sequence of association optimization factors into a Softmax activation function to obtain a sequence of association optimization weight factors; Taking each associated optimization weight factor in the sequence of associated optimization weight factors as a weight, the fault report work order embedded coding vector, the manually released power outage information embedded coding vector, the distribution terminal power outage event embedded coding vector, the terminal offline data embedded coding vector, the line branch switch data embedded coding vector, the current cross-section curve data embedded coding vector and the line fault signal data embedded coding vector are weighted respectively to obtain the optimized fault report work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current cross-section curve data embedded coding vector and the optimized line fault signal data embedded coding vector.

4. The multi-dimensional power outage accurate judgment method according to claim 3 is characterized in that: Calculating the semantic association score vector between any two embedded coding vectors in the feature set consisting of the embedded coding vector of the fault repair work order, the embedded coding vector of the manually released power outage information, the embedded coding vector of the power distribution terminal power outage event, the embedded coding vector of the terminal offline data, the embedded coding vector of the line branch switch data, the embedded coding vector of the current cross-section curve data and the embedded coding vector of the line fault signal data to obtain a sequence of semantic association score vectors, including: The fault repair work order embedding coding vector and the manually released power outage information embedding coding vector are cascaded and multiplied by a weight coefficient matrix, and then dotted with a bias vector to obtain the semantic association score vector.

5. The multi-dimensional power outage accurate judgment method according to claim 4 is characterized in that: Based on the sequence endogenous correlation global representation vector, calculating the correlation optimization factor of the fault repair work order embedded coding vector includes: The fault repair work order embedding coding vector and the sequence endogenous correlation global representation vector are multiplied by different weight coefficient vectors respectively and then added to obtain a semantic correlation coefficient; The semantic association coefficient is added with a bias parameter and then passed through a sigmoid activation function to obtain the association optimization factor.

6. The multi-dimensional power outage accurate judgment method according to claim 5 is characterized in that: The optimized fault repair work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current cross-section curve data embedded coding vector and the optimized line fault signal data embedded coding vector are cross-domain query matching coded to obtain a fault cross-dimensional query matching feature vector, including: The optimized fault report work order embedded coding vector is used as the query feature vector, and the feature set composed of the optimized fault report work order embedded coding vector, the optimized manually released power outage information embedded coding vector, the optimized distribution terminal power outage event embedded coding vector, the optimized terminal offline data embedded coding vector, the optimized line branch switch data embedded coding vector, the optimized current section curve data embedded coding vector and the optimized line fault signal data embedded coding vector is used as the sequence of library feature vectors, and the sequence of the query feature vector and the library feature vector is input into the cross-domain attention gate query encoder based on the key matrix local association enhancement to obtain the fault cross-dimensional query matching feature vector.

7. The multi-dimensional power outage accurate judgment method according to claim 6 is characterized in that: Inputting the sequence of the query feature vector and the library feature vector into a cross-domain attention gate query encoder based on key matrix local association enhancement to obtain the fault cross-dimensional query matching feature vector, including: Performing a linear transformation on each library feature vector in the sequence of library feature vectors to obtain an initial key matrix, wherein each row vector in the initial key matrix is ​​a library feature vector after each linear transformation; Calculating feature distribution difference energy coefficients between the query feature vector and each row vector in the initial key matrix to obtain a sequence of feature distribution difference energy coefficients; Comparing each feature distribution difference energy coefficient in the sequence of feature distribution difference energy coefficients with a predetermined threshold, and in response to the feature distribution difference energy coefficient being less than the predetermined threshold, taking a mean vector between a row vector corresponding to the feature distribution difference energy coefficient and the query feature vector as an updated row vector to obtain an updated key matrix; Processing the query feature vector using a value embedding matrix to obtain a value vector; Using the query feature vector as a query vector and each row vector in the update key matrix as a key vector, the query vector, the key vector and the value vector are input into a cross-domain attention gate query module based on a Transformer structure to obtain a sequence of fault feature cross-domain query attention vectors; The position-wise mean vector of the sequence of the fault feature cross-domain query focus vector is calculated to obtain the fault cross-dimensional query matching feature vector.

8. The multi-dimensional power outage accurate judgment method according to claim 7 is characterized in that: Calculating the characteristic distribution difference energy coefficients between the query characteristic vector and each row vector in the initial key matrix to obtain a sequence of characteristic distribution difference energy coefficients, including: Calculate the position point addition, position point subtraction and position point multiplication between the query feature vector and the row vector to obtain a first matching fusion result, a second matching fusion result and a third matching fusion result; After concatenating the first matching fusion result, the second matching fusion result, and the third matching fusion result into a multi-scale semantic matching vector of a fault feature, one-dimensional convolution processing and maximum pooling processing are performed on the vector to obtain a fault feature association representation vector; The energy coefficient of the fault feature association representation vector is calculated to obtain the feature distribution difference energy coefficient.

9. The multi-dimensional power outage accurate judgment method according to claim 8 is characterized in that: Calculating the energy coefficient of the fault feature association representation vector to obtain the feature distribution difference energy coefficient includes: The square of the difference between the maximum eigenvalue of the fault feature association representation vector and its eigenmean, the feature variance of the fault feature association representation vector and the weighted sum of the bias terms are calculated, and the sum of the feature variance of the fault feature association representation vector and the bias term is divided by the weighted sum to obtain the feature distribution difference energy coefficient.

10. The multi-dimensional power outage accurate judgment method according to claim 9 is characterized in that: Based on the fault cross-dimensional query matching feature vector, a fault judgment result is generated, including: The fault cross-dimensional query matching feature vector is input into a fault analysis result generator based on a decoder to obtain the fault judgment result.

Images