Method and device for evaluating operation state of power system

Abstract

The invention discloses an electric power system operation state evaluation method and device, and belongs to the field of electric power systems, and the method comprises the steps: firstly obtaining the real-time power flow data of a target electric power system; inputting the real-time power flow data into an anomaly recognition model, so that the anomaly recognition model performs anomaly recognition according to the real-time power flow data to obtain an operation state evaluation result of the target power system; the abnormal identification model is obtained by taking the historical fault power flow data set as the source domain data and the historical normal power flow data set as the target domain data and carrying out adversarial training, so that the abnormal operation state of the power system can be effectively identified by implementing the method and the device, and the accuracy of the abnormal operation state of the power system is improved. The problem that in the prior art, due to dependence on artificial experience and simple threshold judgment, the abnormal operation state cannot be found and processed in time is effectively solved.

Description

Technical Field

[0001] The present invention relates to the field of power systems, and in particular to a method and device for evaluating the operating status of a power system. Background Art

[0002] In modern power systems, real-time power flow data is crucial to ensuring the stable operation and security of power systems. As power systems continue to expand in scale and become increasingly complex, they are affected by a variety of factors, leading to abnormal operation.

[0003] Traditional power system monitoring and analysis methods rely primarily on manual experience and simple threshold judgments, making it difficult to accurately identify complex anomalies. For example, during power system operation, power fluctuations and voltage anomalies may occur. If these anomalies are not detected and addressed in a timely manner, they can lead to serious consequences such as equipment damage and power outages. Summary of the Invention

[0004] The present invention provides a method and device for evaluating the operating status of an electric power system. The method effectively identifies the abnormal operating status of the electric power system by inputting real-time power flow data into a constructed abnormality recognition model, thereby solving the problem in the prior art that abnormal operating status cannot be discovered and handled in a timely manner due to reliance on manual experience and simple threshold judgment.

[0005] An embodiment of the present invention provides a method for evaluating the operating status of a power system, comprising:

[0006] Obtain real-time power flow data of the target power system;

[0007] The real-time power flow data is input into an anomaly recognition model so that the anomaly recognition model performs anomaly recognition based on the real-time power flow data to obtain an operating status assessment result of the target power system; wherein, the anomaly recognition model is obtained after adversarial training using a historical fault power flow dataset as source domain data and a historical normal power flow dataset as target domain data.

[0008] Furthermore, the construction process of the anomaly recognition model includes:

[0009] Acquire a historical normal power flow dataset of a target power system and a historical fault power flow dataset of the target power system;

[0010] Build a pair of adversarial domain adaptive networks;

[0011] The historical fault power flow dataset is used as source domain data, and the historical normal power flow dataset is used as target domain data, and the datasets are input into the adversarial domain adaptive network, so that the adversarial domain adaptive network repeatedly performs iterative training according to the source domain data and the target domain data until the total loss function value of the adversarial domain adaptive network converges, thereby obtaining the adversarial domain adaptive network in a domain adaptation state;

[0012] The adversarial domain adaptation network in the domain adaptation state is used as the anomaly recognition model.

[0013] Furthermore, the adversarial domain adaptation network includes a feature extractor, a label predictor, and a domain classifier;

[0014] The feature extractor is configured to receive a historical fault power flow dataset as source domain data and a historical normal power flow dataset as target domain data, extract feature data of the historical fault power flow dataset and the historical normal power flow dataset based on a preset neural network, and input the feature data into the label predictor and the domain classifier respectively;

[0015] The label predictor is used to classify the feature data into abnormal situations and obtain a classification result;

[0016] The domain classifier is used to perform domain discriminant analysis on the feature data to obtain a discrimination result.

[0017] Furthermore, the historical fault power flow dataset is used as source domain data, and the historical normal power flow dataset is used as target domain data, and inputted into the adversarial domain adaptive network, so that the adversarial domain adaptive network repeatedly performs iterative training according to the source domain data and the target domain data until the total loss function value of the adversarial domain adaptive network converges, thereby obtaining the adversarial domain adaptive network in a domain adaptation state, including:

[0018] For each iterative training, the classification result output by the label predictor is compared with the corresponding real abnormal situation, and the loss function value of the current label predictor is calculated according to the comparison result; the discrimination result output by the domain classifier is compared with the corresponding real source domain, and the loss function value of the current domain classifier is calculated according to the comparison result; according to the calculated loss function value of the current label predictor and the loss function value of the current domain classifier, it is judged whether the total loss function value of the adversarial domain adaptive network converges. If so, the iterative training is stopped to obtain the adversarial domain adaptive network in the domain adaptation state. If not, the network parameters of the adversarial domain adaptive network are adjusted to obtain the adversarial domain adaptive network for the next iterative training.

[0019] Furthermore, before inputting the historical fault power flow data set as source domain data and the historical normal power flow data set as target domain data into the adversarial domain adaptive network, the method further includes:

[0020] Obtain the installed capacity of the target power system;

[0021] Based on the installed capacity, simulating the target power system to obtain a simulated failure probability of each transmission line in the target power system;

[0022] Screening and obtaining key transmission lines of the target power system according to the simulated failure probability of each transmission line in the target power system;

[0023] According to the key transmission lines, corresponding weights are assigned to each data in the historical normal power flow dataset and the historical fault power flow dataset to obtain weighted historical normal power flow dataset and historical fault power flow dataset of the target power system.

[0024] An embodiment of the present invention further provides a power system operating status evaluation device, comprising: a data acquisition module and an operating status evaluation module;

[0025] The data acquisition module is used to acquire real-time power flow data of the target power system;

[0026] The operating status assessment module is used to input the real-time power flow data into an anomaly recognition model so that the anomaly recognition model performs anomaly recognition based on the real-time power flow data to obtain an operating status assessment result of the target power system; wherein, the anomaly recognition model is obtained after adversarial training using a historical fault power flow dataset as source domain data and a historical normal power flow dataset as target domain data.

[0027] Furthermore, the construction process of the anomaly recognition model includes:

[0028] Acquire a historical normal power flow dataset of a target power system and a historical fault power flow dataset of the target power system;

[0029] Build a pair of adversarial domain adaptive networks;

[0030] The historical fault power flow dataset is used as source domain data, and the historical normal power flow dataset is used as target domain data, and the datasets are input into the adversarial domain adaptive network, so that the adversarial domain adaptive network repeatedly performs iterative training according to the source domain data and the target domain data until the total loss function value of the adversarial domain adaptive network converges, thereby obtaining the adversarial domain adaptive network in a domain adaptation state;

[0031] The adversarial domain adaptation network in the domain adaptation state is used as the anomaly recognition model.

[0032] Furthermore, the adversarial domain adaptation network includes a feature extractor, a label predictor, and a domain classifier;

[0033] The feature extractor is configured to receive a historical fault power flow dataset as source domain data and a historical normal power flow dataset as target domain data, extract feature data of the historical fault power flow dataset and the historical normal power flow dataset based on a preset neural network, and input the feature data into the label predictor and the domain classifier respectively;

[0034] The label predictor is used to classify the feature data into abnormal situations and obtain a classification result;

[0035] The domain classifier is used to perform domain discriminant analysis on the feature data to obtain a discrimination result.

[0036] Furthermore, the historical fault power flow dataset is used as source domain data, and the historical normal power flow dataset is used as target domain data, and inputted into the adversarial domain adaptive network, so that the adversarial domain adaptive network repeatedly performs iterative training according to the source domain data and the target domain data until the total loss function value of the adversarial domain adaptive network converges, thereby obtaining the adversarial domain adaptive network in a domain adaptation state, including:

[0037] For each iterative training, the classification result output by the label predictor is compared with the corresponding real abnormal situation, and the loss function value of the current label predictor is calculated according to the comparison result; the discrimination result output by the domain classifier is compared with the corresponding real source domain, and the loss function value of the current domain classifier is calculated according to the comparison result; according to the calculated loss function value of the current label predictor and the loss function value of the current domain classifier, it is judged whether the total loss function value of the adversarial domain adaptive network converges. If so, the iterative training is stopped to obtain the adversarial domain adaptive network in the domain adaptation state. If not, the network parameters of the adversarial domain adaptive network are adjusted to obtain the adversarial domain adaptive network for the next iterative training.

[0038] Furthermore, the power system operation status assessment device further includes: a key transmission line determination module;

[0039] The key transmission line determination module is used to obtain the installed capacity of the target power system; based on the installed capacity, simulate the target power system to obtain the simulated failure probability of each transmission line in the target power system; based on the simulated failure probability of each transmission line in the target power system, screen and obtain the key transmission lines of the target power system; based on the key transmission lines, assign corresponding weights to each data in the historical normal power flow data set and the historical fault power flow data set to obtain the weighted historical normal power flow data set and historical fault power flow data set of the target power system.

[0040] The following beneficial effects are achieved by implementing the present invention:

[0041] The present invention provides a method and device for evaluating the operating status of an electric power system. The method inputs the acquired real-time power flow data of the target electric power system into an anomaly recognition model, so that the anomaly recognition model performs anomaly recognition based on the real-time power flow data, and obtains an evaluation result of the operating status of the target electric power system. Since the anomaly recognition model is obtained through adversarial training using a historical fault power flow data set as source domain data and a historical normal power flow data set as target domain data, the operating status of the current target electric power system can be effectively evaluated, thereby solving the problem in the prior art that abnormal operating status cannot be discovered and processed in a timely manner due to reliance on manual experience and simple threshold judgment. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solution of the present application, the following is a brief introduction to the drawings required for use in the implementation. Obviously, the drawings described below are only some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0043] Figure 1 This is a flowchart of a method for evaluating the operating status of a power system provided in one embodiment of the present application;

[0044] Figure 2 This is a schematic diagram of the structure of a power system operating status assessment device provided by a certain embodiment of the present application;

[0045] Figure 3 It is a structural diagram of a power system operating status assessment device provided in another embodiment of the present application. DETAILED DESCRIPTION

[0046] To make the objectives, technical solutions, and advantages of this application more clear, the technical solutions in this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of this application.

[0047] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned figure descriptions are intended to cover non-exclusive inclusions.

[0048] In the description of the embodiments of this application, the technical terms "first" and "second" are used only to distinguish different objects and should not be understood to indicate or imply relative importance or implicitly specify the quantity, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, the meaning of "plurality" is more than two, unless otherwise clearly and specifically defined.

[0049] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0050] In the description of the embodiments of this application, the term "and / or" is simply a description of the association relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent the following three situations: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this document generally indicates that the associated objects are in an "or" relationship.

[0051] In the description of the embodiments of the present application, the term "multiple" refers to more than two (including two). Similarly, "multiple groups" refers to more than two groups (including two groups), and "multiple pieces" refers to more than two pieces (including two pieces).

[0052] In the description of the embodiments of the present application, unless otherwise expressly specified or limited, technical terms such as "installed," "connected," "connected," and "fixed" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integration; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; internal connections between two components or interactions between two components. Those skilled in the art can understand the specific meanings of the above terms in the embodiments of the present application based on specific circumstances.

[0053] See also Figure 1 To address the problem in the prior art of failing to promptly detect and address abnormal operating conditions due to reliance on manual experience and simple threshold judgment, an embodiment of the present invention provides a method for evaluating the operating status of a power system, comprising:

[0054] S1. Obtain real-time power flow data of the target power system;

[0055] S2. Inputting the real-time power flow data into an anomaly recognition model so that the anomaly recognition model performs anomaly recognition based on the real-time power flow data to obtain an operating status assessment result of the target power system; wherein the anomaly recognition model is obtained through adversarial training using a historical fault power flow dataset as source domain data and a historical normal power flow dataset as target domain data;

[0056] In a preferred embodiment, before the historical fault power flow dataset is used as source domain data and the historical normal power flow dataset is used as target domain data and inputted into the adversarial domain adaptive network, the method further includes:

[0057] Obtain the installed capacity of the target power system;

[0058] Based on the installed capacity, simulating the target power system to obtain a simulated failure probability of each transmission line in the target power system;

[0059] Screening and obtaining key transmission lines of the target power system according to the simulated failure probability of each transmission line in the target power system;

[0060] According to the key transmission lines, assigning corresponding weights to each data in the historical normal power flow dataset and the historical fault power flow dataset to obtain weighted historical normal power flow dataset and historical fault power flow dataset of the target power system;

[0061] Schematically, the installed capacity of the target power system includes the generator reserve capacity g and the line capacity factor u of the target power system;

[0062] Specifically, assuming that there are N transmission lines in the target power system, the maximum capacity of the generator is The relationship between load and generator capacity can be expressed as:

[0063]

[0064] Where, represents the maximum capacity of the generator of the i-th line; g represents the generator's spare capacity; P {Lj} represents the load on the jth line;

[0065] The capacity factor u of a transmission line can be defined as:

[0066]

[0067] Where, F {ij} represents the active power flow between the i-th line and the j-th line; represents the maximum transmission capacity of the line; u represents the capacity factor of the line;

[0068] Schematically, based on the installed capacity, the target power system is simulated to obtain a simulated failure probability of each transmission line in the target power system;

[0069] Specifically, the failure probability of the kth line in the i-th simulation is defined as p {lk} , and its calculation formula is as follows:

[0070]

[0071] Where S {ik} Indicates whether the kth line has a fault in the i-th simulation. If a fault occurs, S {ik} =1, otherwise 0; N represents the total number of simulations.

[0072] Risk value R {lk} It can be defined as:

[0073] R {lk} =p {lk} ×I {lk} ;

[0074] Where, I {lk} is the system load loss caused by the kth transmission line failure; R {lk} represents the risk value of the k-th transmission line;

[0075] Specifically, the transmission lines whose risk values ​​exceed a preset threshold are regarded as the key transmission lines of the target power system; then the data of the key transmission lines in the historical normal power flow dataset and the historical fault power flow dataset are given a higher weight, and the weighted historical normal power flow dataset and historical fault power flow dataset of the target power system are obtained.

[0076] In a preferred embodiment, the process of constructing the anomaly recognition model includes:

[0077] Acquire a historical normal power flow dataset of a target power system and a historical fault power flow dataset of the target power system;

[0078] Build a pair of adversarial domain adaptive networks;

[0079] The historical fault power flow dataset is used as source domain data, and the historical normal power flow dataset is used as target domain data, and the datasets are input into the adversarial domain adaptive network, so that the adversarial domain adaptive network repeatedly performs iterative training according to the source domain data and the target domain data until the total loss function value of the adversarial domain adaptive network converges, thereby obtaining the adversarial domain adaptive network in a domain adaptation state;

[0080] Using the adversarial domain adaptation network in the domain adaptation state as the anomaly recognition model;

[0081] In a preferred embodiment, the adversarial domain adaptation network includes a feature extractor, a label predictor, and a domain classifier;

[0082] The feature extractor is configured to receive a historical fault power flow dataset as source domain data and a historical normal power flow dataset as target domain data, extract feature data of the historical fault power flow dataset and the historical normal power flow dataset based on a preset neural network, and input the feature data into the label predictor and the domain classifier respectively;

[0083] The label predictor is used to classify the feature data into abnormal situations and obtain a classification result;

[0084] The domain classifier is used to perform domain discriminant analysis on the feature data to obtain a discrimination result;

[0085] Schematically, an adversarial domain adaptive network is constructed, wherein the adversarial domain adaptive network includes a feature extractor, a label predictor, and a domain classifier; then the historical fault power flow dataset is used as the source domain data, and the historical normal power flow dataset is used as the target domain data, and input into the constructed adversarial domain adaptive network for iterative training;

[0086] Specifically, the feature extractor needs to extract feature data of the historical fault power flow dataset and the historical normal power flow dataset based on a preset neural network during each iterative training;

[0087] Specifically, in this application, a graph neural network (GNN) is used to extract feature data; the graph neural network updates the hidden state of each node by aggregating the features of neighboring nodes; the hidden state update formula of the l+1 layer is:

[0088]

[0089] Where, represents the power flow characteristic data of the i-th transmission line; represents the characteristics of edge k; N(i) represents the set of neighbor nodes of node i; ε(i) represents the set of all edges of node i; W {(l)} 、 and b {(l)} denote the weight matrix and bias of the lth layer respectively; σ(·) denotes the activation function;

[0090] Then, after extracting the feature data, the feature data of the historical fault power flow data set and the historical normal power flow data set are extracted respectively, and the feature data are input into the label predictor so that the label predictor classifies the feature data as abnormal situations to obtain classification results; and the feature data are input into the domain classifier so that the feature data is subjected to domain discriminant analysis to obtain discrimination results.

[0091] In a preferred embodiment, the historical fault power flow dataset is used as source domain data, and the historical normal power flow dataset is used as target domain data, and inputted into the adversarial domain adaptive network, so that the adversarial domain adaptive network repeatedly performs iterative training according to the source domain data and the target domain data until the total loss function value of the adversarial domain adaptive network converges, thereby obtaining the adversarial domain adaptive network in a domain adaptation state, including:

[0092] For each iterative training, the classification result output by the label predictor is compared with the corresponding real abnormal situation, and the loss function value of the current label predictor is calculated according to the comparison result; the discrimination result output by the domain classifier is compared with the corresponding real source domain, and the loss function value of the current domain classifier is calculated according to the comparison result; based on the calculated loss function value of the current label predictor and the loss function value of the current domain classifier, it is judged whether the total loss function value of the adversarial domain adaptive network converges. If so, the iterative training is stopped to obtain the adversarial domain adaptive network in the domain adaptation state. If not, the network parameters of the adversarial domain adaptive network are adjusted to obtain the adversarial domain adaptive network for the next iterative training;

[0093] Schematically, in order to improve the effect of domain adversarial learning, the maximum mean difference (MMD) criterion is introduced in the domain discriminator to align the feature distributions of the source domain and the target domain;

[0094] Specifically, in each iterative training, the classification result output by the current label predictor needs to be compared with the corresponding real abnormal situation, and then the loss function value of the current label predictor is calculated based on the comparison result;

[0095] It should be noted that the classification result is the classification of the feature data into fault data or normal data; the corresponding real abnormal situation is the actual abnormal situation of the feature data (normal or fault);

[0096] Specifically, the discrimination result output by the domain classifier is compared with the corresponding true source domain, and the loss function value of the current domain classifier is calculated according to the comparison result;

[0097] It should be noted that the discrimination result, i.e., the feature data that is discriminated is derived from the source domain data or the target domain data; the true source domain, i.e., the feature data actually is derived from the source domain data or the target domain data;

[0098] Then, a weighted sum is performed based on the calculated loss function value of the current label predictor and the loss function value of the current domain classifier to obtain the total loss function value of the adversarial domain adaptive network, and then it is determined whether the total loss function value of the adversarial domain adaptive network converges. If so, the iterative training is stopped to obtain the adversarial domain adaptive network in the domain adaptation state. If not, the network parameters of the adversarial domain adaptive network are adjusted to obtain the adversarial domain adaptive network for the next iterative training.

[0099] Finally, the adversarial domain adaptation network in the domain adaptation state is used as the anomaly recognition model;

[0100] Specifically, when evaluating the operating status of the target power system, the acquired real-time power flow data is input into the abnormality recognition model, so that the abnormality recognition model performs abnormality recognition based on the real-time power flow data to obtain the operating status evaluation result of the target power system;

[0101] Specifically, the operation evaluation result may be that the target power system is in a normal operating state or in a fault abnormal state.

[0102] See Figure 2 , is a power system operating status assessment device provided by an embodiment of the present invention, comprising: a data acquisition module and an operating status assessment module;

[0103] The data acquisition module is used to acquire real-time power flow data of the target power system;

[0104] The operating status assessment module is used to input the real-time power flow data into an anomaly recognition model so that the anomaly recognition model performs anomaly recognition based on the real-time power flow data to obtain an operating status assessment result of the target power system; wherein, the anomaly recognition model is obtained after adversarial training using a historical fault power flow dataset as source domain data and a historical normal power flow dataset as target domain data.

[0105] In a preferred embodiment, the process of constructing the anomaly recognition model includes:

[0106] Acquire a historical normal power flow dataset of a target power system and a historical fault power flow dataset of the target power system;

[0107] Build a pair of adversarial domain adaptive networks;

[0108] The historical fault power flow dataset is used as source domain data, and the historical normal power flow dataset is used as target domain data, and the datasets are input into the adversarial domain adaptive network, so that the adversarial domain adaptive network repeatedly performs iterative training according to the source domain data and the target domain data until the total loss function value of the adversarial domain adaptive network converges, thereby obtaining the adversarial domain adaptive network in a domain adaptation state;

[0109] The adversarial domain adaptation network in the domain adaptation state is used as the anomaly recognition model.

[0110] In a preferred embodiment, the adversarial domain adaptation network includes a feature extractor, a label predictor, and a domain classifier;

[0111] The feature extractor is configured to receive a historical fault power flow dataset as source domain data and a historical normal power flow dataset as target domain data, extract feature data of the historical fault power flow dataset and the historical normal power flow dataset based on a preset neural network, and input the feature data into the label predictor and the domain classifier respectively;

[0112] The label predictor is used to classify the feature data into abnormal situations and obtain a classification result;

[0113] The domain classifier is used to perform domain discriminant analysis on the feature data to obtain a discrimination result.

[0114] In a preferred embodiment, the historical fault power flow dataset is used as source domain data, and the historical normal power flow dataset is used as target domain data, and inputted into the adversarial domain adaptive network, so that the adversarial domain adaptive network repeatedly performs iterative training according to the source domain data and the target domain data until the total loss function value of the adversarial domain adaptive network converges, thereby obtaining the adversarial domain adaptive network in a domain adaptation state, including:

[0115] For each iterative training, the classification result output by the label predictor is compared with the corresponding real abnormal situation, and the loss function value of the current label predictor is calculated according to the comparison result; the discrimination result output by the domain classifier is compared with the corresponding real source domain, and the loss function value of the current domain classifier is calculated according to the comparison result; according to the calculated loss function value of the current label predictor and the loss function value of the current domain classifier, it is judged whether the total loss function value of the adversarial domain adaptive network converges. If so, the iterative training is stopped to obtain the adversarial domain adaptive network in the domain adaptation state. If not, the network parameters of the adversarial domain adaptive network are adjusted to obtain the adversarial domain adaptive network for the next iterative training.

[0116] See Figure 3 ,In a preferred embodiment, the power system operating state assessment device further includes: a key transmission line determination module;

[0117] The key transmission line determination module is used to obtain the installed capacity of the target power system; based on the installed capacity, simulate the target power system to obtain the simulated failure probability of each transmission line in the target power system; based on the simulated failure probability of each transmission line in the target power system, screen and obtain the key transmission lines of the target power system; based on the key transmission lines, assign corresponding weights to each data in the historical normal power flow data set and the historical fault power flow data set to obtain the weighted historical normal power flow data set and historical fault power flow data set of the target power system.

[0118] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for evaluating the operating status of a power system, characterized in that: include: Obtain real-time power flow data of the target power system; The real-time power flow data is input into an anomaly recognition model so that the anomaly recognition model performs anomaly recognition based on the real-time power flow data to obtain an operating status assessment result of the target power system; wherein, the anomaly recognition model is obtained after adversarial training using a historical fault power flow dataset as source domain data and a historical normal power flow dataset as target domain data.

2. The method for evaluating the operating status of a power system according to claim 1, wherein: The construction process of the anomaly recognition model includes: Acquire a historical normal power flow dataset of a target power system and a historical fault power flow dataset of the target power system; Build a pair of adversarial domain adaptive networks; The historical fault power flow dataset is used as source domain data, and the historical normal power flow dataset is used as target domain data, and are input into the adversarial domain adaptive network, so that the adversarial domain adaptive network repeatedly performs iterative training according to the source domain data and the target domain data until the total loss function value of the adversarial domain adaptive network converges, thereby obtaining the adversarial domain adaptive network in a domain adaptation state; The adversarial domain adaptation network in the domain adaptation state is used as the anomaly recognition model.

3. The method for evaluating the operating status of a power system according to claim 2, wherein: The adversarial domain adaptation network includes a feature extractor, a label predictor, and a domain classifier; The feature extractor is configured to receive a historical fault power flow dataset as source domain data and a historical normal power flow dataset as target domain data, extract feature data of the historical fault power flow dataset and the historical normal power flow dataset based on a preset neural network, and input the feature data into the label predictor and the domain classifier respectively; The label predictor is used to classify the feature data into abnormal situations and obtain a classification result; The domain classifier is used to perform domain discriminant analysis on the feature data to obtain a discrimination result.

4. The method for evaluating the operating status of a power system according to claim 3, wherein: The method uses the historical fault power flow dataset as source domain data and the historical normal power flow dataset as target domain data to input the data into the adversarial domain adaptive network, so that the adversarial domain adaptive network repeatedly performs iterative training according to the source domain data and the target domain data until the total loss function value of the adversarial domain adaptive network converges, thereby obtaining the adversarial domain adaptive network in a domain adaptation state, including: For each iterative training, the classification result output by the label predictor is compared with the corresponding real abnormal situation, and the loss function value of the current label predictor is calculated according to the comparison result; the discrimination result output by the domain classifier is compared with the corresponding real source domain, and the loss function value of the current domain classifier is calculated according to the comparison result; according to the calculated loss function value of the current label predictor and the loss function value of the current domain classifier, it is judged whether the total loss function value of the adversarial domain adaptive network converges. If so, the iterative training is stopped to obtain the adversarial domain adaptive network in the domain adaptation state. If not, the network parameters of the adversarial domain adaptive network are adjusted to obtain the adversarial domain adaptive network for the next iterative training.

5. The method for evaluating the operating status of a power system according to claim 4, wherein: Before inputting the historical fault power flow dataset as source domain data and the historical normal power flow dataset as target domain data into the adversarial domain adaptive network, the method further includes: Obtain the installed capacity of the target power system; Based on the installed capacity, simulating the target power system to obtain a simulated failure probability of each transmission line in the target power system; Screening and obtaining key transmission lines of the target power system according to the simulated failure probability of each transmission line in the target power system; According to the key transmission lines, corresponding weights are assigned to each data in the historical normal power flow dataset and the historical fault power flow dataset to obtain weighted historical normal power flow dataset and historical fault power flow dataset of the target power system.

6. A power system operating status assessment device, characterized in that: include: Data acquisition module and operation status evaluation module; The data acquisition module is used to acquire real-time power flow data of the target power system; The operating status assessment module is used to input the real-time power flow data into an anomaly recognition model so that the anomaly recognition model performs anomaly recognition based on the real-time power flow data to obtain an operating status assessment result of the target power system; wherein, the anomaly recognition model is obtained after adversarial training using a historical fault power flow dataset as source domain data and a historical normal power flow dataset as target domain data.

7. The power system operating status evaluation device according to claim 6, characterized in that: The construction process of the anomaly recognition model includes: Acquire a historical normal power flow dataset of a target power system and a historical fault power flow dataset of the target power system; Build a pair of adversarial domain adaptive networks; The historical fault power flow dataset is used as source domain data, and the historical normal power flow dataset is used as target domain data, and are input into the adversarial domain adaptive network, so that the adversarial domain adaptive network repeatedly performs iterative training according to the source domain data and the target domain data until the total loss function value of the adversarial domain adaptive network converges, thereby obtaining the adversarial domain adaptive network in a domain adaptation state; The adversarial domain adaptation network in the domain adaptation state is used as the anomaly recognition model.

8. The power system operating status evaluation device according to claim 7, characterized in that: The adversarial domain adaptation network includes a feature extractor, a label predictor, and a domain classifier; The feature extractor is configured to receive a historical fault power flow dataset as source domain data and a historical normal power flow dataset as target domain data, extract feature data of the historical fault power flow dataset and the historical normal power flow dataset based on a preset neural network, and input the feature data into the label predictor and the domain classifier respectively; The label predictor is used to classify the feature data into abnormal situations and obtain a classification result; The domain classifier is used to perform domain discriminant analysis on the feature data to obtain a discrimination result.

9. The power system operating status evaluation device according to claim 8, characterized in that: The method uses the historical fault power flow dataset as source domain data and the historical normal power flow dataset as target domain data to input the data into the adversarial domain adaptive network, so that the adversarial domain adaptive network repeatedly performs iterative training according to the source domain data and the target domain data until the total loss function value of the adversarial domain adaptive network converges, thereby obtaining the adversarial domain adaptive network in a domain adaptation state, including: For each iterative training, the classification result output by the label predictor is compared with the corresponding real abnormal situation, and the loss function value of the current label predictor is calculated according to the comparison result; the discrimination result output by the domain classifier is compared with the corresponding real source domain, and the loss function value of the current domain classifier is calculated according to the comparison result; according to the calculated loss function value of the current label predictor and the loss function value of the current domain classifier, it is judged whether the total loss function value of the adversarial domain adaptive network converges. If so, the iterative training is stopped to obtain the adversarial domain adaptive network in the domain adaptation state. If not, the network parameters of the adversarial domain adaptive network are adjusted to obtain the adversarial domain adaptive network for the next iterative training.

10. The power system operating status evaluation device according to claim 9, characterized in that: Also includes: Key transmission line identification module; The key transmission line determination module is used to obtain the installed capacity of the target power system; Based on the installed capacity, simulating the target power system to obtain a simulated failure probability of each transmission line in the target power system; Screening and obtaining key transmission lines of the target power system according to the simulated failure probability of each transmission line in the target power system; According to the key transmission lines, corresponding weights are assigned to each data in the historical normal power flow dataset and the historical fault power flow dataset to obtain weighted historical normal power flow dataset and historical fault power flow dataset of the target power system.

Images