Multi-modal fusion power network operation state evaluation method and device

Abstract

The invention relates to the related technical field of power network control, in particular to a multi-mode fusion power network operation state evaluation method and device, and the method comprises the steps: receiving power network switch control and operation state monitoring parameters, obtaining prediction parameters through the analysis of a primary network and a secondary network, adding the same attributes, comparing the obtained prediction parameters with the monitoring parameters, and obtaining a prediction result; the abnormal circuit is fed back to the decision-making layer, switch disconnection control is executed, and the technical problems that in a complex power network, influences of operation environment fluctuation and data noise are likely to happen, the operation state of the power network is difficult to respond and adjust in time, and abnormal conditions and sudden faults cannot be effectively handled are solved; the method has the technical effects of improving the accuracy of operation state prediction parameters, accurately isolating and disconnecting abnormal circuits, analyzing prediction deviation of a main network and a secondary network layer by layer, establishing a dynamic adjustment and grading evaluation mechanism, efficiently processing redundancy and abnormal conditions of multi-modal data in the power network, and improving the operation efficiency and safety of the power network.

Description

Technical Field

[0001] The present invention relates to the technical field related to power network control, and in particular to a method and device for evaluating the operating status of a power network using multi-modal fusion. Background Art

[0002] With the continuous growth of energy demand and the large-scale access of renewable energy, the scale and complexity of modern power networks have increased significantly. The power system has gradually developed towards smart grids to cope with the dynamic balance of energy supply and demand and the complex and changing operating environment. In this process, the power network needs to have a higher level of security, reliability and intelligence to ensure the stability of power supply.

[0003] The remote control circuit of power network switches is related to the safe operation of the power network. At present, conventional power network operation status assessments are based on fixed settings of operating modes. Faced with the high complexity and dynamic changes of power network operation, it is difficult to effectively capture the correlation and prediction accuracy of multiple operating states, and it is impossible to dynamically adjust according to the real-time circuit switch status, resulting in large assessment deviations.

[0004] In summary, the existing technology has technical problems that complex power networks are easily affected by operating environment fluctuations and data noise, making it difficult to respond and adjust the operating status of the power network in a timely manner, and unable to effectively deal with abnormal situations and sudden failures. Summary of the invention

[0005] The present application aims to solve the technical problems in the prior art that a complex power network is easily affected by operating environment fluctuations and data noise, it is difficult to respond and adjust the power network operating status in a timely manner, and it is impossible to effectively deal with abnormal situations and sudden failures by providing a multi-modal fusion power network operating status assessment device.

[0006] In view of the above problems, the technical solution to implement this application is: On the one hand, the present application provides a multi-modal fusion power network operation status assessment method, wherein a power network operation status assessment device applied to multi-modal fusion is embedded in a power network switch remote control circuit decision layer, and the method includes: receiving a power network switch control state and a power network operation status monitoring parameter of a target power network; processing the power network switch control state through an operation status analysis primary network to obtain a first operation status prediction parameter; processing the power network switch control state through an operation status analysis secondary network to obtain a second operation status prediction parameter; performing same-attribute summation on the first operation status prediction parameter and the second operation status prediction parameter, and outputting an operation status prediction parameter; comparing the power network operation status monitoring parameter and the operation status prediction parameter to obtain an abnormal circuit of the power network; and feeding back the abnormal circuit of the power network to the power network switch remote control circuit decision layer to execute abnormal circuit switch disconnection control.

[0007] On the other hand, the present application provides a multi-modal fusion power network operation status assessment device, wherein the device is embedded in the power network switch remote control circuit decision layer, and the device includes: a parameter receiving module, which is used to receive the power network switch control state and the power network operation status monitoring parameter of the target power network; a first processing module, which is used to process the power network switch control state through the operation status analysis main network to obtain a first operation status prediction parameter; a second processing module, which is used to process the power network switch control state through the operation status analysis secondary network to obtain a second operation status prediction parameter; an attribute addition module, which is used to perform the same attribute addition on the first operation status prediction parameter and the second operation status prediction parameter, and output the operation status prediction parameter; a parameter comparison module, which is used to compare the power network operation status monitoring parameter and the operation status prediction parameter to obtain the power network abnormal circuit; a disconnection control module, which is used to feed back the power network abnormal circuit to the power network switch remote control circuit decision layer to execute abnormal circuit switch disconnection control.

[0008] In summary, one or more technical solutions provided in the present application receive power network switch control and operation status monitoring parameters, obtain prediction parameters through primary and secondary network analysis, compare the parameters with the monitoring parameters after summing the same attributes, and find abnormal circuits; feed back the abnormal circuits to the decision-making layer, execute switch disconnection control, use the operation status analysis of the primary network and the secondary network to hierarchically process the power network data, improve the accuracy of the operation status prediction parameters, accurately isolate and disconnect the abnormal circuits, analyze the prediction deviations of the primary network and the secondary network layer by layer, establish a dynamic adjustment and hierarchical evaluation mechanism, efficiently handle the redundancy and abnormal conditions of multimodal data in the power network, and improve the operation efficiency and safety of the power network. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A flow chart of a multi-modal fusion power network operation status assessment method is provided for this application; Figure 2 A structural schematic diagram of a multi-modal fusion power network operation status assessment device is provided for this application.

[0010] Description of reference numerals: parameter receiving module M100, first processing module M200, second processing module M300, attribute adding module M400, parameter comparing module M500, disconnection control module M600. DETAILED DESCRIPTION Embodiment 1

[0011] The present application is described in detail below with reference to the accompanying drawings. Figure 1 As shown, the present application provides a multi-modal fusion power network operation status assessment method, wherein a multi-modal fusion power network operation status assessment device is applied, and the device is embedded in the decision layer of the power network switch remote control circuit, and the method includes: S1: Receive the power network switch control state and power network operation state monitoring parameters of the target power network; S2: Process the power network switch control state through the operation state analysis primary network to obtain the first operation state prediction parameter; S3: Process the power network switch control state through the operation state analysis secondary network to obtain the second operation state prediction parameter.

[0012] Specifically, the switch control state of the power network refers to the on and off states of each switch in the power network, which determine the topological structure and current path of the power network; the power network operation status monitoring parameters include voltage, current, frequency and other parameters, which reflect the various status information of the power network in real-time operation and are the basic data for evaluating the health status of the power network; the operation status analysis main network is used to conduct a global analysis of the switch control state of the power network to obtain an overall operation status prediction; the operation status analysis secondary network complements the main network, and the secondary network focuses on compensating and deeply adjusting the deviations of the main network analysis results, especially for more detailed predictive analysis under abnormal conditions; the operation status prediction parameters refer to the parameters obtained after analyzing the network processing switch control state, which are used to predict the future operation status of the power network, including normal operation status and possible abnormal status.

[0013] The switch opening and closing states and operation state monitoring parameters are collected in real time from various nodes and sensors of the power network, and these data are transmitted to the evaluation device through the data acquisition system; the received power network switch control state is input into the operation state analysis main network, and the main network uses its trained model to perform a global analysis of the switch state, comprehensively considers the overall topological structure and operation characteristics of the power network, and outputs the first operation state prediction parameter; the global analysis of the main network can capture the overall operation trend and characteristics of the power network, and provide a preliminary prediction result for the operation state of the power network. The first operation state prediction parameter reflects the overall operation state of the power network under the current switch state.

[0014] Based on the analysis of the main network, the switch control state is input into the operating state analysis secondary network. The secondary network is used to compensate and adjust the deviation of the main network analysis results, especially to conduct more detailed prediction analysis under abnormal conditions, and output the second operating state prediction parameters. The introduction of the secondary network can effectively make up for the insufficient analysis of the main network in some local or abnormal conditions, and improve the accuracy and reliability of the operating state prediction. The second operating state prediction parameters supplement and optimize the first prediction parameters, making the overall operating state evaluation more comprehensive and accurate, and better able to cope with complex situations and sudden failures in the power network.

[0015] S4: perform same-attribute summation on the first operating state prediction parameter and the second operating state prediction parameter, and output the operating state prediction parameter; S5: compare the power network operating state monitoring parameter and the operating state prediction parameter to obtain the power network abnormal circuit; S6: feed back the power network abnormal circuit to the power network switch remote control circuit decision layer to execute abnormal circuit switch disconnection control.

[0016] Specifically, the same attribute addition refers to the addition processing of parameters with the same attributes in the first operating state prediction parameter and the second operating state prediction parameter. For example, if both parameters contain voltage prediction values, the two voltage prediction values ​​are added to obtain a more accurate voltage prediction result; the operating state prediction parameter is a comprehensive prediction parameter obtained after the same attribute addition processing, which reflects the overall operating state of the power network under the current switching state, including normal operating state and possible abnormal state; the power network abnormal circuit refers to the circuit part that is found to be inconsistent with the normal operating state after comparing the power network operating state monitoring parameters and the operating state prediction parameters. These abnormal circuits may be caused by equipment failure, line problems or other reasons; the power network switch remote control circuit decision layer is a decision layer for remotely controlling switches in the power network. According to the abnormal circuit information provided by the evaluation device, the switch disconnection control operation is performed to isolate and eliminate the abnormal circuit to ensure the safe and stable operation of the power system.

[0017] The parameters of the same attributes in the first operating state prediction parameter and the second operating state prediction parameter are added together, for example, the voltage prediction values ​​in the two parameters are added together to obtain a comprehensive voltage prediction result. The parameters of other attributes are also added together in the same way, and finally the comprehensive operating state prediction parameters are output. In the above steps, the addition of the same attributes can integrate the analysis results of the primary network and the secondary network, make full use of the advantages of the two networks, and improve the accuracy and reliability of the operating state prediction.

[0018] The real-time collected power network operation status monitoring parameters are compared with the output operation status prediction parameters; by analyzing the difference between the two, the circuit part that does not conform to the normal operation status, that is, the abnormal circuit of the power network, is identified. For example, if there is a significant deviation between a voltage value in the monitoring parameter and the voltage value in the prediction parameter, it can be judged that the circuit is abnormal; by comparing the monitoring parameters and the prediction parameters, faults and problems in the power network are discovered in time, providing accurate information support for subsequent abnormal processing and control decisions.

[0019] The detected abnormal circuit information of the power network is fed back to the remote control circuit decision layer of the power network switch. The decision layer executes the corresponding switch disconnection control operation based on the abnormal circuit information to isolate the abnormal circuit from the power network to prevent the fault from spreading and affecting other normally operating circuits; the abnormal circuit information is fed back to the control decision layer and the switch disconnection control is executed to isolate and eliminate the abnormal circuit in time, effectively prevent the further expansion of the fault, reduce the impact on power supply, and improve the reliability and stability of the power system.

[0020] Furthermore, the main network processes the power network switch control state through the operation state analysis to obtain the first operation state prediction parameter. Before that, the method of the present application includes: Obtain a circuit equivalent topology diagram of the target power network; construct an operation status analysis network topology according to the circuit equivalent topology diagram, wherein the operation status analysis network topology has the same topological structure as the circuit equivalent topology diagram, the input nodes of the operation status analysis network topology are switch distribution nodes, the input information includes 0 or 1, 1 represents connection, and 0 represents disconnection, the input nodes of the operation status analysis main network topology are various lines, and the output information is the operation status characteristic value; collect the operation monitoring record parameters of the target power network, perform training on the operation status analysis network topology, and obtain the operation status analysis main network.

[0021] Specifically, the circuit equivalent topology diagram represents the connection relationship between various components (such as generators, transformers, lines, switches, etc.) in the power network. The equivalent topology diagram simplifies the complex power network structure into a form that is easy to analyze and calculate through equivalent processing; the operation status analysis network topology refers to the network topology structure constructed based on the circuit equivalent topology diagram, which is used for operation status analysis. This topology is the same as the topology of the circuit equivalent topology diagram, but the definitions of input nodes and output nodes are different to meet the needs of operation status analysis; the switch distribution node represents the distribution position of the switch in the power network. As the input node of the operation status analysis network topology, the state of the switch (connected or disconnected) is represented by the input information (0 or 1), 1 means connected, and 0 means disconnected; the operation status characteristic value refers to the information output by the operation status analysis main network, which reflects the operation characteristics of the power network in the current switch state, such as the predicted values ​​of parameters such as voltage, current, and power; the operation monitoring record parameters are the historical operation data collected from the operation monitoring system of the power network, including the recorded values ​​of parameters such as voltage, current, and frequency, which are used to train the operation status analysis network topology so that it can accurately predict the operation status of the power network.

[0022] Detailed topological information of the power network is obtained from the planning and design information or real-time operation data of the power network; the complex power network structure is simplified into a circuit equivalent topological diagram through the equivalent processing method. The equivalent processing usually includes steps such as equivalent merging of components and simplifying connection relationships to obtain a topological diagram that is easy to analyze and calculate; the equivalent topological diagram can clearly represent the connection relationship between each component in the power network, and provide structured basic data for subsequent operation status analysis, making the analysis process more efficient and accurate.

[0023] On the basis of the circuit equivalent topology diagram, the operation status analysis network topology is constructed. The topology structure is the same as that of the circuit equivalent topology diagram, but the definitions of input nodes and output nodes are different. The input nodes are switch distribution nodes, and the input information includes 0 or 1, indicating the connection or disconnection state of the switch; the output nodes are individual lines, and the output information is the characteristic value of the operation status, such as the predicted values ​​of parameters such as voltage and current; the operation status analysis network topology is constructed to combine the structural information of the power network with the operation status analysis, so that the analysis process can fully consider the topological structure characteristics of the power network, and accurately simulate the operation of the power network under different switch states by taking the switch state as input information. Historical operation data is collected from the operation monitoring system of the power network, including the recorded values ​​of parameters such as voltage, current, and frequency. These operation monitoring record parameters are used as training data to train the operation status analysis network topology. The training process usually adopts machine learning or deep learning algorithms, and the parameters of the network topology are continuously adjusted to enable it to accurately predict the operation status of the power network; by training the operation status analysis network topology, the operation status analysis main network is obtained. The main network can accurately predict the operation status characteristic values ​​of the power network according to the input switch status information, so that the main network has the ability to globally analyze and predict the operation status of the power network, providing a reliable basis for subsequent abnormality detection and control decisions.

[0024] Further, the secondary network processes the power network switch control state through the operation state analysis to obtain the second operation state prediction parameter. Before, the method of the present application includes: Obtain a set of main network output deviation vectors whose output deviation of the operation status analysis main network is greater than or equal to the output deviation threshold; when the number of the main network output deviation vector sets is less than the quantity threshold, set the output of the operation status analysis secondary network to 0; when the number of the main network output deviation vector sets is greater than the quantity threshold, count the main network output system deviation vectors of the main network output deviation vector sets; with the main network output system deviation vector as supervision, collect the operation monitoring record parameters of the target power network, perform training on the operation status analysis network topology, and obtain a secondary operation status analysis secondary network; loop analysis until the number of N-level network output deviation vector sets is less than the quantity threshold, perform addition and full connection on the secondary operation status analysis secondary network until the N-level operation status analysis secondary network, and generate the operation status analysis secondary network.

[0025] Specifically, the main network output deviation vector set refers to the set of deviation vectors generated by the main network of the running status analysis during the prediction process. Each deviation vector represents the difference between the predicted value of the main network and the actual value. The number of deviation vectors in the set reflects the accuracy of the main network prediction. The output deviation threshold is a preset value used to determine whether the output deviation of the main network is significant. When the deviation is greater than or equal to the threshold, the deviation is considered significant and requires further analysis and processing. The quantity threshold is a preset value used to determine whether the number of sets of main network output deviation vectors is large enough. When the number of sets is less than the threshold, the deviation of the main network is considered insufficient to start the analysis of the secondary network. When the number of sets is greater than the threshold, the analysis of the secondary network is started. The main network output system deviation vector refers to the deviation vector obtained by statistically analyzing the set of main network output deviation vectors, which reflects the average deviation of the main network in the entire system. The summed full connection refers to summing the outputs of multiple network layers and integrating them through the fully connected layer to generate a running status analysis secondary network, thereby integrating the analysis results of multiple networks to improve the accuracy and reliability of the prediction.

[0026] In the process of the main network predicting the running status, the deviation between each predicted value and the actual value is calculated, and the deviation vectors with deviations greater than or equal to the output deviation threshold are collected into the main network output deviation vector set; by collecting significant deviation vectors, the main errors and deficiencies in the prediction process of the main network are identified, providing a basis for further analysis and optimization; when the number of the main network output deviation vector set is less than the preset quantity threshold, it is considered that the prediction deviation of the main network is insufficient to start the analysis of the secondary network, and the output of the secondary network is set to 0, indicating that the secondary network does not participate in the current analysis process; by setting the quantity threshold, it is possible to avoid starting the secondary network when the main network prediction is relatively accurate, thereby improving the efficiency and stability of the system.

[0027] When the number of the main network output deviation vector set is greater than the preset number threshold, the deviation vectors in the set are statistically analyzed to calculate the main network output system deviation vector, reflecting the average deviation of the main network in the entire system; statistical analysis of the main network output system deviation vector can provide a more comprehensive understanding of the prediction deviation of the main network, and provide accurate supervision information for the analysis and training of the secondary network; using the main network output system deviation vector as supervision information, combined with the operation monitoring record parameters of the target power network, the operation status analysis network topology is trained to obtain a secondary operation status analysis secondary network, which is used to compensate and adjust the deviation of the main network, especially for deep prediction analysis under abnormal conditions; by training the secondary network, the prediction deviation of the main network can be effectively compensated, the accuracy and reliability of the operation status prediction can be improved, and the system's ability to cope with complex and abnormal situations can be enhanced.

[0028] The analysis and training are performed in a loop until the number of deviation vector sets output by the N-level network is less than the quantity threshold. During the loop process, the secondary operation status analysis secondary network until the N-level operation status analysis secondary network is subjected to addition and full connection, and the analysis results of the multi-level network are integrated to generate the final operation status analysis secondary network. Through the loop analysis and addition and full connection, the prediction and analysis of the operation status can be refined and optimized layer by layer, and the advantages of the multi-level network can be fully utilized to improve the overall performance and prediction accuracy of the system, and ensure the accuracy and reliability of the evaluation of the operation status of the power network.

[0029] Furthermore, the power network operation status monitoring parameter and the operation status prediction parameter are compared to obtain the abnormal circuit of the power network. The method of the present application includes: Configure an operation status monitoring attribute deviation threshold; compare the operation status deviation modulus of the power network operation status monitoring parameter and the operation status prediction parameter; extract the circuit whose operation status deviation modulus is greater than the operation status monitoring attribute deviation threshold, and add it to the power network abnormal circuit.

[0030] Specifically, the operating status monitoring attribute deviation threshold is a preset value used to determine whether the deviation between the power network operating status monitoring parameters and the predicted parameters is significant. Each monitoring attribute (such as voltage, current, etc.) has a corresponding deviation threshold, which is used to identify abnormal situations; the operating status deviation modulus represents the deviation between the power network operating status monitoring parameters and the operating status predicted parameters, which is usually a non-negative number. The larger the modulus, the more significant the deviation; the power network abnormal circuit refers to the circuit part that is found to be inconsistent with the normal operating status after comparing the monitoring parameters and the predicted parameters. It is caused by equipment failure, line problems or other reasons.

[0031] According to the operation experience and historical data of the power network, a deviation threshold is configured for each operation status monitoring attribute (such as voltage, current, frequency, etc.). The deviation threshold can be adjusted and optimized according to the actual operation conditions and safety requirements. By setting the threshold reasonably, it is possible to accurately judge whether the deviation between the monitoring parameters and the predicted parameters is significant, so as to timely discover potential abnormal situations and improve the safety and reliability of the system.

[0032] The real-time collected power network operation status monitoring parameters are compared with the operation status prediction parameters, and the operation status deviation modulus of each monitoring attribute is calculated. For example, for the voltage attribute, the absolute value of the difference between the monitored voltage and the predicted voltage is calculated as the voltage deviation modulus. By calculating the deviation modulus, the difference between the monitoring parameters and the prediction parameters can be quantified, providing accurate data support for subsequent abnormality identification and processing.

[0033] During the comparison process, circuits whose operating status deviation modulus is greater than the corresponding monitoring attribute deviation threshold are extracted. For example, if the voltage deviation modulus of a circuit is greater than the voltage deviation threshold, the circuit is identified as an abnormal circuit and added to the list of abnormal circuits in the power network. By identifying and recording abnormal circuits, problems and faults in the power network can be understood in a timely manner, providing accurate information support for subsequent abnormal processing and control decisions, which helps to improve the safety and stability of the power system.

[0034] Further, to configure the running status monitoring attribute deviation threshold, the present application method includes: Perform a separate attribute deviation analysis on the operating status monitoring attribute set to obtain a first-level operating status monitoring attribute deviation threshold and a non-individually sensitive operating status monitoring attribute set; perform a two-item attribute enumeration combination deviation analysis on the non-individually sensitive operating status monitoring attribute set to obtain a second-level operating status monitoring attribute deviation threshold; until M-item attribute enumeration combination deviation analysis is performed on the non-individually sensitive operating status monitoring attribute set to obtain an M-level operating status monitoring attribute deviation threshold, wherein M represents the number of attributes of the non-individually sensitive operating status monitoring attribute set; add the first-level operating status monitoring attribute deviation threshold, the second-level operating status monitoring attribute deviation threshold up to the M-level operating status monitoring attribute deviation threshold into the operating status monitoring attribute deviation threshold, wherein the first-level operating status monitoring attribute deviation threshold, the second-level operating status monitoring attribute deviation threshold up to the M-level operating status monitoring attribute deviation threshold are in a logical AND relationship.

[0035] Specifically, the operating status monitoring attribute set refers to the set of all operating status attributes that need to be monitored in the power network, such as voltage, current, frequency, power, etc.; the individual attribute deviation is to perform deviation analysis on each operating status monitoring attribute separately to determine the degree of its impact on the operating status of the power network when it changes separately; the first-level operating status monitoring attribute deviation threshold is the deviation threshold obtained by the individual attribute deviation analysis, which indicates the maximum allowable deviation range of each attribute when it changes separately; the non-individually sensitive operating status monitoring attribute set refers to the attribute set whose individual changes do not significantly affect the operating status of the power network in the individual attribute deviation analysis.

[0036] Binary attribute enumeration combination deviation analysis refers to the deviation analysis of the attributes in the non-individual sensitive attribute set in pairs, so as to determine the degree of influence of these attribute combinations when they change on the operation status of the power network; the secondary operation status monitoring attribute deviation threshold is the deviation threshold obtained by the binomial attribute enumeration combination deviation analysis, which indicates the maximum allowable deviation range of the attribute combination when it changes; M-item attribute enumeration combination deviation analysis refers to the deviation analysis of M-item combinations of attributes in the non-individual sensitive attribute set, where M is the number of attributes, so as to determine the degree of influence of these attribute combinations when they change on the operation status of the power network; the M-level operation status monitoring attribute deviation threshold is the deviation threshold obtained by the M-item attribute enumeration combination deviation analysis, which indicates the maximum allowable deviation range of the attribute combination when it changes; the logical and relationship refers to a relationship that is established only when multiple conditions are met at the same time. In the setting of the deviation threshold, it means that the operation status of the power network is considered normal only when the deviations of all relevant attributes are less than or equal to their corresponding deviation thresholds.

[0037] Each operating status monitoring attribute is disturbed individually to observe its impact on the operating status of the power network. By analyzing the operating monitoring logs after the disturbance, the proportion of negative samples (circuit abnormality samples) is counted. According to the proportion of negative samples and the disturbance step, the first-level deviation threshold of each attribute is determined. For those attributes whose individual changes have no significant impact on the operating status of the power network, they are classified into the set of non-individually sensitive operating status monitoring attributes; individual attribute deviation analysis can identify attributes that have a significant impact on the operating status of the power network and set reasonable deviation thresholds for them.

[0038] Binary attributes are randomly extracted from the set of non-individual sensitive attributes, and the disturbance step corresponding to the binomial attributes is set to perform joint disturbance; the operation monitoring log after the joint disturbance is collected, and the number of negative samples of the joint disturbance is counted; according to the number of negative samples of the joint disturbance and the disturbance step, the secondary deviation threshold of the combination of these two attributes is determined; the binomial attribute enumeration combination deviation analysis can identify the attribute pairs that have a significant impact on the operating status of the power network when the combination changes, and set a reasonable deviation threshold for it, so as to more comprehensively evaluate the operating status of the power network when multiple attributes change at the same time.

[0039] Continue to enumerate and combine the attributes in the non-individual sensitive attribute set, from two-item combination to M-item combination, and gradually increase the number of combined attributes; for each attribute combination, repeat the process of disturbance, collection of operation monitoring logs, and statistical analysis of the proportion of negative samples, and determine the deviation threshold of the corresponding combination until the deviation analysis of the M-item attribute enumeration combination is completed; by gradually increasing the number of attribute combinations, it is possible to more carefully analyze the impact of simultaneous changes in multiple attributes on the operating status of the power network, obtain a more comprehensive deviation threshold, and more accurately evaluate the operating status of the power network in a complex operating environment, thereby improving the accuracy of anomaly detection.

[0040] The first-level operation status monitoring attribute deviation threshold, the second-level operation status monitoring attribute deviation threshold, and the M-level operation status monitoring attribute deviation threshold are added into the operation status monitoring attribute deviation threshold, and all deviation thresholds obtained through single attribute deviation analysis and multi-level attribute enumeration combined deviation analysis are integrated into the operation status monitoring attribute deviation threshold; in the actual operation status monitoring process, the operation status of the power network is considered normal only when the deviations of all relevant attributes are less than or equal to their corresponding deviation thresholds; integrating the deviation thresholds of different levels into a logical and relationship can ensure that the operation status of the power network meets the requirements in multiple dimensions, improve the strictness and accuracy of anomaly detection, timely discover potential abnormal situations, and ensure the safe and stable operation of the power network.

[0041] Further, performing a separate attribute deviation analysis on the set of operating status monitoring attributes to obtain a first-level operating status monitoring attribute deviation threshold and a non-separate sensitive operating status monitoring attribute set, the method of the present application includes: Extract a first operating state monitoring attribute from the operating state monitoring attribute set; set a first disturbance step size of the first operating state monitoring attribute; perform a unique disturbance on the first operating state monitoring attribute with the first disturbance step size, and collect the operating monitoring log of the target power network; count the proportion of negative samples in the operating monitoring log, wherein negative samples represent circuit abnormality samples; when the proportion of negative samples is greater than or equal to a sample proportion threshold, and the first disturbance step size is less than or equal to a step size threshold, set the first disturbance step size to a first operating state monitoring attribute deviation threshold; when the negative samples When the number proportion is greater than or equal to the sample proportion threshold, and the first disturbance step length is greater than the step length threshold, the first disturbance step length is multiplied by 0.5 to obtain a second disturbance step length, and a separate attribute deviation cyclic analysis is performed according to the second disturbance step length; when the number proportion of negative samples is less than the sample proportion threshold, the first disturbance step length is multiplied by 3 to obtain a third disturbance step length to perform a separate attribute deviation cyclic analysis; when the number proportion of negative samples is less than the sample proportion threshold, and the first disturbance step length is greater than the zero correlation step length threshold, the first operating status monitoring attribute is added to the non-separate sensitive operating status monitoring attribute set.

[0042] Specifically, the operating state monitoring attribute set refers to the set of all operating state attributes that need to be monitored in the power network, such as voltage, current, frequency, etc.; the first operating state monitoring attribute refers to a specific attribute extracted from the operating state monitoring attribute set, which is used for individual attribute deviation analysis; the disturbance step size refers to the amplitude of the attribute value change when the attribute is disturbed, and the size of the disturbance step size will affect the degree of change in the state of the power network after the disturbance; the unique disturbance refers to a single disturbance of a certain attribute without changing the state of other attributes; the operation monitoring log refers to the detailed information recording the operating state of the power network after the disturbance, including the changes in parameters such as voltage, current, and frequency; the proportion of negative samples refers to the proportion of the number of circuit abnormal samples (negative samples) to the total number of samples in the operation monitoring log; the sample proportion threshold is a preset value used to determine whether the proportion of negative samples is significant; the step threshold is a preset value used to determine whether the disturbance step size is reasonable; the zero correlation step threshold means that when the disturbance step size exceeds the threshold, it is considered that the attribute has no significant correlation with the abnormal state of the power network, and it is classified into the non-individual sensitive operating state monitoring attribute set.

[0043] Select a specific attribute from the set of operating status monitoring attributes as the analysis object, for example, select the voltage attribute as the first operating status monitoring attribute; extract the first operating status monitoring attribute, and perform a separate deviation analysis on it to determine its impact on the operating status of the power network and the deviation threshold, so as to provide a basis for subsequent abnormality detection; set an initial disturbance step for the extracted first operating status monitoring attribute, for example, set the disturbance step of the voltage attribute to 0.1V; in the simulation environment of the power network, perform a unique disturbance on the first operating status monitoring attribute, that is, only change the value of the attribute without changing the states of other attributes, for example, increase the value of the voltage attribute by 0.1V, and collect the operating monitoring log of the power network after the disturbance; through unique disturbance and collection of operating monitoring logs, use the disturbance step to control the amplitude of the attribute value change, analyze the specific impact of each attribute change on the operating status of the power network, and provide data support for subsequent deviation analysis.

[0044] From the collected operation monitoring log, the number of circuit abnormal samples (negative samples) is counted, and its proportion in the total number of samples is calculated. The proportion of negative samples is further counted to evaluate the degree of abnormality of the power network after the disturbance. The higher the proportion of negative samples, the greater the impact of the change in this attribute on the operation state of the power network. If the proportion of negative samples after the disturbance is greater than or equal to the preset sample proportion threshold, and the first disturbance step is less than or equal to the step threshold, the first disturbance step is used as the deviation threshold of the first operation state monitoring attribute, and the deviation threshold is further set to determine the maximum allowable range of variation of this attribute under normal operation. When the change of this attribute in actual operation exceeds the deviation threshold, it can be judged that there may be an abnormality in the power network.

[0045] In the first case, when the proportion of negative samples is greater than or equal to the sample proportion threshold, and the first disturbance step is greater than the step threshold, the first disturbance step is multiplied by 0.5 to obtain the second disturbance step, and the individual attribute deviation analysis is performed again according to the second disturbance step, and the individual attribute deviation cycle analysis is performed. Furthermore, if the number of negative samples is greater than the sample proportion threshold, it means that the corresponding step must be greater than or equal to the deviation threshold. At this time, try to reduce the step to find a more accurate deviation threshold. If the step itself is small enough, you can directly set the deviation threshold. By reducing the disturbance step, the impact of the change in the attribute on the operating state of the power network can be carefully analyzed, which helps to more accurately determine the deviation threshold.

[0046] In the second case, when the proportion of negative samples after disturbance is less than the sample proportion threshold, the first disturbance step is multiplied by 3 to obtain the third disturbance step, and the individual attribute deviation cyclic analysis is performed again according to the third disturbance step. By gradually increasing the disturbance step, further, if the number of negative samples is less than the sample proportion threshold, it means that the corresponding step must be less than or equal to the deviation threshold. At this time, try to increase the step to find a more accurate deviation threshold. By gradually increasing the disturbance step and performing cyclic analysis, the deviation threshold of the operating status monitoring attribute can be more accurately determined, thereby improving the accuracy and reliability of anomaly detection.

[0047] In the third case, if the proportion of negative samples after the disturbance is less than the sample proportion threshold, and the first disturbance step is greater than the zero-correlation step threshold, the first operating status monitoring attribute is classified into the non-individually sensitive operating status monitoring attribute set; the non-individually sensitive attributes are classified into the set, and then a combined deviation analysis is performed to evaluate the impact of these attributes on the operating status of the power network when the combination changes, so as to further improve the accuracy of anomaly detection.

[0048] Further, performing a binary attribute enumeration combination deviation analysis on the non-individually sensitive running state monitoring attribute set to obtain a secondary running state monitoring attribute deviation threshold, the present application method includes: Randomly extract a first non-individually sensitive operating state monitoring attribute and a second non-individually sensitive operating state monitoring attribute from the non-individually sensitive operating state monitoring attribute set; set a first non-individually sensitive attribute disturbance step of the first non-individually sensitive operating state monitoring attribute and a second non-individually sensitive attribute disturbance step of the second non-individually sensitive operating state monitoring attribute; use the first non-individually sensitive attribute disturbance step and the second non-individually sensitive attribute disturbance step to jointly disturb the first non-individually sensitive operating state monitoring attribute and the second non-individually sensitive operating state monitoring attribute, and collect the joint disturbance operation monitoring log of the target power network; count the proportion of the number of joint disturbance negative samples in the joint disturbance operation monitoring log; when the proportion of the number of joint disturbance negative samples is greater than or equal to the sample proportion threshold, and the first non-individually sensitive attribute disturbance step is less than or equal to the first non-individually sensitive attribute disturbance step threshold, and the second non-individually sensitive attribute disturbance step is greater than or equal to the first non-individually sensitive attribute disturbance step threshold, When it is less than or equal to the second non-individual sensitive attribute disturbance step length threshold, the first non-individual sensitive attribute disturbance step length and the second non-individual sensitive attribute disturbance step length are configured as a logical AND relationship and added into the secondary operation status monitoring attribute deviation threshold; when the proportion of the number of negative samples of the joint disturbance is greater than or equal to the sample proportion threshold, and the first non-individual sensitive attribute disturbance step length is greater than the first non-individual sensitive attribute disturbance step length threshold, and at least one of the second non-individual sensitive attribute disturbance step length is greater than the second non-individual sensitive attribute disturbance step length threshold is triggered, at least one of the triggered first non-individual sensitive attribute disturbance step length and the second non-individual sensitive attribute disturbance step length is multiplied by 0.5 and then a binomial attribute deviation cyclic analysis is performed; when the proportion of the number of negative samples of the joint disturbance is less than the sample proportion threshold, the first non-individual sensitive attribute disturbance step length and the second non-individual sensitive attribute disturbance step length are multiplied by 3 and then a binomial attribute deviation cyclic analysis is performed.

[0049] Specifically, the set of non-individually sensitive operating status monitoring attributes refers to a set of attributes that have no significant impact on the operating status of the power network when changed individually, and their impact on the power network needs to be evaluated through combined analysis; the first non-individually sensitive operating status monitoring attribute and the second non-individually sensitive operating status monitoring attribute are two attributes randomly extracted from the set of non-individually sensitive attributes, which are used for joint disturbance analysis; the non-individually sensitive attribute disturbance step refers to the disturbance amplitude of each non-individually sensitive attribute when performing a joint disturbance.

[0050] Joint disturbance refers to the simultaneous disturbance of two or more non-individually sensitive attributes to observe the impact of their combined changes on the operating status of the power network; the joint disturbance operation monitoring log refers to the detailed information recording the operating status of the power network after the joint disturbance, including the changes in each attribute and the abnormal status of the circuit; the proportion of negative samples in the joint disturbance refers to the proportion of the number of abnormal circuit samples (negative samples) to the total number of samples in the joint disturbance operation monitoring log; the sample proportion threshold is a preset value used to determine whether the proportion of negative samples is significant; the non-individually sensitive attribute disturbance step threshold is a preset value used to determine whether the disturbance step is reasonable. Different attributes may have different disturbance step thresholds.

[0051] A first non-individually sensitive operating state monitoring attribute (such as voltage) and a second non-individually sensitive operating state monitoring attribute (such as frequency) are randomly selected from a set of non-individually sensitive operating state monitoring attributes. The two attributes are randomly extracted for joint disturbance analysis to evaluate the impact of the combined changes of the two attributes on the operating state of the power network; a first non-individually sensitive attribute disturbance step size of the first non-individually sensitive operating state monitoring attribute and a second non-individually sensitive attribute disturbance step size of the second non-individually sensitive operating state monitoring attribute are set, for example, the voltage disturbance step size is set to 0.2V, and the frequency disturbance step size is set to 0.1Hz; the disturbance step size is set to control the variation amplitude of the attribute value, so as to observe its specific impact on the operating state of the power network during the joint disturbance process.

[0052] The first non-individually sensitive operating state monitoring attribute and the second non-individually sensitive operating state monitoring attribute are jointly disturbed with the first non-individually sensitive attribute disturbance step size and the second non-individually sensitive operating state monitoring attribute. Exemplarily, in a simulated environment of an electric power network, the first non-individually sensitive operating state monitoring attribute and the second non-individually sensitive operating state monitoring attribute are simultaneously disturbed, for example, the voltage is increased by 0.2 V, the frequency is increased by 0.1 Hz, and then the operating monitoring log of the electric power network after the joint disturbance is collected; the joint disturbance operating monitoring log is collected, and the specific impact of the combined changes of the first non-individually sensitive operating state monitoring attribute and the second non-individually sensitive operating state monitoring attribute on the operating state of the electric power network is analyzed, so as to provide data support for subsequent deviation analysis.

[0053] From the collected joint disturbance operation monitoring log, count the joint disturbance negative samples, and calculate the proportion of the number of joint disturbance negative samples in the total number of samples; count the proportion of the number of joint disturbance negative samples to evaluate the abnormality of the power network after the joint disturbance. The higher the proportion of negative samples, the greater the impact of the combined changes of the two attributes on the operating state of the power network; if the proportion of the number of joint disturbance negative samples is greater than or equal to the sample proportion threshold, and the first non-individual sensitive attribute disturbance step and the second non-individual sensitive attribute disturbance step are within their respective threshold ranges, then configure the first non-individual sensitive attribute disturbance step and the second non-individual sensitive attribute disturbance step as a logical AND relationship, and add them to the secondary operating state monitoring attribute deviation threshold to determine the maximum allowable variation range of the two attribute combinations under normal operating conditions. When the combined variation of the two attributes in actual operation exceeds the deviation threshold, it can be judged that the power network is abnormal.

[0054] When the proportion of the number of negative samples of the joint disturbance is greater than or equal to the sample proportion threshold, and the first non-individual sensitive attribute disturbance step length is greater than the first non-individual sensitive attribute disturbance step length threshold, and the second non-individual sensitive attribute disturbance step length is less than or equal to the second non-individual sensitive attribute disturbance step length threshold; and, when the proportion of the number of negative samples of the joint disturbance is greater than or equal to the sample proportion threshold, and the first non-individual sensitive attribute disturbance step length is less than or equal to the first non-individual sensitive attribute disturbance step length threshold, and the second non-individual sensitive attribute disturbance step length is greater than the second non-individual sensitive attribute disturbance step length threshold; and, when the proportion of the number of negative samples of the joint disturbance is greater than or equal to the sample proportion threshold, and the first non-individual sensitive attribute disturbance step length is greater than The first non-single sensitive attribute disturbance step length threshold, at the same time, the second non-single sensitive attribute disturbance step length is greater than the second non-single sensitive attribute disturbance step length threshold. Further, the first non-single sensitive attribute disturbance step length greater than the first non-single sensitive attribute disturbance step length threshold and the second non-single sensitive attribute disturbance step length greater than the second non-single sensitive attribute disturbance step length threshold are used as trigger conditions, and at least one step length of the first non-single sensitive attribute disturbance step length and the second non-single sensitive attribute disturbance step length corresponding to the trigger is multiplied by 0.5, and the binomial attribute deviation cyclic analysis is re-executed according to the calculated disturbance step length; by reducing the disturbance step length, the impact of attribute combination changes on the operating state of the power network is analyzed in detail.

[0055] When the proportion of the number of negative samples of the joint disturbance is less than the sample proportion threshold, the disturbance step size of the first non-individual sensitive attribute and the disturbance step size of the second non-individual sensitive attribute are multiplied by 3, and the binomial attribute deviation cyclic analysis is performed again based on the calculated disturbance step size; by increasing the disturbance step size, the degree of its impact on the operating status of the power network is comprehensively evaluated.

[0056] In summary, the beneficial effects of the embodiments of the present application are: Receive the power network switch control state and power network operation state monitoring parameters of the target power network; process the power network switch control state through the operation state analysis main network to obtain the first operation state prediction parameter; process the power network switch control state through the operation state analysis secondary network to obtain the second operation state prediction parameter; perform same-attribute summation on the first operation state prediction parameter and the second operation state prediction parameter, output the operation state prediction parameter, and compare it with the power network operation state monitoring parameter to obtain the power network abnormal circuit; feed back the power network abnormal circuit to the power network switch remote control circuit decision layer to execute abnormal circuit switch disconnection control. This application provides a multi-modal fusion power network operation state evaluation method and device, uses the operation state analysis main network and the secondary network to hierarchically process the power network data, improves the accuracy of the operation state prediction parameter, accurately isolates and disconnects the abnormal circuit, analyzes the prediction deviation of the main network and the secondary network layer by layer, establishes a dynamic adjustment and hierarchical evaluation mechanism, efficiently handles the redundancy and abnormal conditions of multi-modal data in the power network, and improves the operation efficiency and safety of the power network. Embodiment 2

[0057] Based on the same inventive concept as the multi-modal fusion power network operation status assessment method in the aforementioned embodiment, Figure 2 As shown, an embodiment of the present application provides a multi-modal fusion power network operation status assessment device, wherein the device includes: The parameter receiving module M100 is used to receive the power network switch control state and power network operation state monitoring parameters of the target power network; A first processing module M200 is used to process the power network switch control state through an operation state analysis main network to obtain a first operation state prediction parameter; The second processing module M300 is used to process the power network switch control state through the operation state analysis secondary network to obtain a second operation state prediction parameter; An attribute summing module M400, configured to sum the first operating state prediction parameter and the second operating state prediction parameter by the same attribute, and output the operating state prediction parameter; The parameter comparison module M500 is used to compare the power network operation status monitoring parameters with the operation status prediction parameters to obtain abnormal circuits of the power network; The disconnection control module M600 is used to feed back the abnormal circuit of the power network to the decision layer of the remote control circuit of the power network switch to execute the abnormal circuit switch disconnection control.

[0058] Furthermore, the first processing module M200 is also used to execute the following method: Obtaining a circuit equivalent topology diagram of the target power network; According to the circuit equivalent topology diagram, construct an operation status analysis network topology, wherein the operation status analysis network topology has the same topological structure as the circuit equivalent topology diagram, the input node of the operation status analysis network topology is a switch distribution node, the input information includes 0 or 1, 1 represents connection, and 0 represents disconnection, the input node of the operation status analysis main network topology is each line, and the output information is the operation status characteristic value; The operation monitoring record parameters of the target power network are collected, and the operation status analysis network topology is trained to obtain the operation status analysis main network.

[0059] Furthermore, the second processing module M300 is also used to execute the following method: Obtain a set of main network output deviation vectors whose output deviation of the operation status analysis main network is greater than or equal to an output deviation threshold; When the number of the deviation vector sets output by the main network is less than the number threshold, setting the output of the operation status analysis secondary network to 0; When the number of the main network output deviation vector sets is greater than a number threshold, counting the main network output system deviation vectors of the main network output deviation vector sets; Taking the main network output system deviation vector as supervision, collecting the operation monitoring record parameters of the target power network, performing training on the operation status analysis network topology, and obtaining a secondary operation status analysis secondary network; The analysis is repeated until the number of deviation vector sets output by the N-level network is less than the number threshold, and then the secondary operation status analysis secondary network is subjected to addition and full connection until the N-level operation status analysis secondary network to generate the operation status analysis secondary network.

[0060] Furthermore, the parameter comparison module M500 is used to perform the following method: Configure the running status monitoring attribute deviation threshold; Comparing the operating state deviation modulus of the power network operating state monitoring parameter and the operating state prediction parameter; The circuits whose operating state deviation modulus is greater than the operating state monitoring attribute deviation threshold are extracted and added into the power network abnormal circuit.

[0061] Furthermore, the parameter comparison module M500 is also used to execute the following method: Performing a separate attribute deviation analysis on the operation status monitoring attribute set to obtain a first-level operation status monitoring attribute deviation threshold and a non-separately sensitive operation status monitoring attribute set; Performing a binomial attribute enumeration combination deviation analysis on the non-individually sensitive running status monitoring attribute set to obtain a secondary running status monitoring attribute deviation threshold; Until performing M-item attribute enumeration combination deviation analysis on the non-individually sensitive running state monitoring attribute set to obtain an M-level running state monitoring attribute deviation threshold, wherein M represents the number of attributes of the non-individually sensitive running state monitoring attribute set; The first-level operating status monitoring attribute deviation threshold, the second-level operating status monitoring attribute deviation threshold, and up to the M-level operating status monitoring attribute deviation threshold are added into the operating status monitoring attribute deviation threshold, wherein the first-level operating status monitoring attribute deviation threshold, the second-level operating status monitoring attribute deviation threshold, and up to the M-level operating status monitoring attribute deviation threshold are in a logical AND relationship.

[0062] Furthermore, the parameter comparison module M500 is also used to execute the following method: Extracting a first operating status monitoring attribute from the operating status monitoring attribute set; Setting a first disturbance step size of the first operating state monitoring attribute; Performing unique perturbations on the first operation status monitoring attribute with the first perturbation step length, and collecting operation monitoring logs of the target power network; Counting the proportion of negative samples in the operation monitoring log, wherein negative samples represent circuit abnormality samples; When the proportion of the number of negative samples is greater than or equal to the sample proportion threshold, and the first disturbance step length is less than or equal to the step length threshold, setting the first disturbance step length as the first operating state monitoring attribute deviation threshold; When the proportion of the number of negative samples is greater than or equal to the sample proportion threshold, and the first perturbation step length is greater than the step length threshold, the first perturbation step length is multiplied by 0.5 to obtain a second perturbation step length, and a separate attribute deviation cyclic analysis is performed according to the second perturbation step length; When the proportion of the number of negative samples is less than the sample proportion threshold, the first disturbance step is multiplied by 3 to obtain a third disturbance step to perform a separate attribute deviation cyclic analysis; When the proportion of the number of negative samples is less than the sample proportion threshold and the first disturbance step is greater than the zero-correlation step threshold, the first operating status monitoring attribute is added to the non-individually sensitive operating status monitoring attribute set.

[0063] Furthermore, the parameter comparison module M500 is also used to execute the following method: Randomly extracting a first non-individually sensitive operating state monitoring attribute and a second non-individually sensitive operating state monitoring attribute from the non-individually sensitive operating state monitoring attribute set; Setting a first non-individually sensitive attribute disturbance step length of the first non-individually sensitive operating state monitoring attribute and a second non-individually sensitive attribute disturbance step length of the second non-individually sensitive operating state monitoring attribute; Performing a joint disturbance on the first non-individually sensitive operating state monitoring attribute and the second non-individually sensitive operating state monitoring attribute with the first non-individually sensitive attribute disturbance step length and the second non-individually sensitive attribute disturbance step length, and collecting a joint disturbance operation monitoring log of the target power network; Counting the proportion of the number of negative samples of the joint disturbance in the joint disturbance operation monitoring log; When the proportion of the number of negative samples of the joint disturbance is greater than or equal to the sample proportion threshold, and the first non-individual sensitive attribute disturbance step length is less than or equal to the first non-individual sensitive attribute disturbance step length threshold, and the second non-individual sensitive attribute disturbance step length is less than or equal to the second non-individual sensitive attribute disturbance step length threshold, the first non-individual sensitive attribute disturbance step length and the second non-individual sensitive attribute disturbance step length are configured as a logical AND relationship, and added into the secondary operation status monitoring attribute deviation threshold; When the proportion of the number of negative samples of the joint disturbance is greater than or equal to the sample proportion threshold, and at least one of the first non-individual sensitive attribute disturbance step length is greater than the first non-individual sensitive attribute disturbance step length threshold, and the second non-individual sensitive attribute disturbance step length is greater than the second non-individual sensitive attribute disturbance step length threshold is triggered, at least one of the triggered first non-individual sensitive attribute disturbance step length and the second non-individual sensitive attribute disturbance step length is multiplied by 0.5 and then a binomial attribute deviation cyclic analysis is performed; When the proportion of the number of negative samples of the joint disturbance is less than the sample proportion threshold, the first non-individual sensitive attribute disturbance step length and the second non-individual sensitive attribute disturbance step length are multiplied by 3 and then a binomial attribute deviation cyclic analysis is performed.

[0064] In summary, any step can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor, and no unnecessary restrictions are made here.

[0065] Furthermore, the above-mentioned technical scheme only reflects the preferred technical scheme of the technical scheme of the embodiment of the present application. Some changes that may be made to some parts thereof by technicians in this technical field all reflect the novel principles of the embodiment of the present application. Obviously, technicians in this field can make various changes and modifications to the present application without departing from the scope of the present application.

Claims

1. A multi-modal fusion power network operation status assessment method, characterized in that: A power network operation status assessment device applied to multi-modal fusion, the device is embedded in the decision layer of the power network switch remote control circuit, and the method includes: Receiving a power network switch control state and a power network operation state monitoring parameter of a target power network; The main network processes the switch control state of the power network through the operation state analysis to obtain a first operation state prediction parameter; Processing the power network switch control state through the operation state analysis secondary network to obtain a second operation state prediction parameter; Performing same-attribute summation on the first running state prediction parameter and the second running state prediction parameter, and outputting the running state prediction parameter; Comparing the power network operation status monitoring parameter with the operation status prediction parameter to obtain an abnormal circuit of the power network; The abnormal circuit of the power network is fed back to the decision layer of the remote control circuit of the power network switch to execute the abnormal circuit switch disconnection control.

2. The method according to claim 1, characterized in that The main network processes the switch control state of the power network through the operation state analysis to obtain the first operation state prediction parameter, which includes: Obtaining a circuit equivalent topology diagram of the target power network; According to the circuit equivalent topology diagram, construct an operation status analysis network topology, wherein the operation status analysis network topology has the same topological structure as the circuit equivalent topology diagram, the input node of the operation status analysis network topology is a switch distribution node, the input information includes 0 or 1, 1 represents connection, and 0 represents disconnection, the input node of the operation status analysis main network topology is each line, and the output information is the operation status characteristic value; The operation monitoring record parameters of the target power network are collected, and the operation status analysis network topology is trained to obtain the operation status analysis main network.

3. The method according to claim 2, characterized in that The secondary network processes the power network switch control state through the operation state analysis to obtain the second operation state prediction parameter, which includes: Obtain a set of main network output deviation vectors whose output deviation of the operation status analysis main network is greater than or equal to an output deviation threshold; When the number of the deviation vector sets output by the main network is less than the number threshold, setting the output of the operation status analysis secondary network to 0; When the number of the main network output deviation vector sets is greater than a number threshold, counting the main network output device deviation vectors of the main network output deviation vector sets; Taking the deviation vector of the main network output device as supervision, collecting the operation monitoring record parameters of the target power network, performing training on the operation status analysis network topology, and obtaining a secondary operation status analysis secondary network; The analysis is repeated until the number of deviation vector sets output by the N-level network is less than the number threshold, and then the secondary operation status analysis secondary network is subjected to addition and full connection until the N-level operation status analysis secondary network to generate the operation status analysis secondary network.

4. The method according to claim 3, characterized in that Comparing the power network operation status monitoring parameter with the operation status prediction parameter to obtain the power network abnormal circuit, including: Configure the running status monitoring attribute deviation threshold; Comparing the operating state deviation modulus of the power network operating state monitoring parameter and the operating state prediction parameter; The circuits whose operating state deviation modulus is greater than the operating state monitoring attribute deviation threshold are extracted and added into the power network abnormal circuit.

5. The method according to claim 4, characterized in that Configure the deviation thresholds for running status monitoring attributes, including: Performing a separate attribute deviation analysis on the operation status monitoring attribute set to obtain a first-level operation status monitoring attribute deviation threshold and a non-separately sensitive operation status monitoring attribute set; Performing a binomial attribute enumeration combination deviation analysis on the non-individually sensitive running status monitoring attribute set to obtain a secondary running status monitoring attribute deviation threshold; Until performing M-item attribute enumeration combination deviation analysis on the non-individually sensitive running state monitoring attribute set to obtain an M-level running state monitoring attribute deviation threshold, wherein M represents the number of attributes of the non-individually sensitive running state monitoring attribute set; The first-level operating status monitoring attribute deviation threshold, the second-level operating status monitoring attribute deviation threshold, and up to the M-level operating status monitoring attribute deviation threshold are added into the operating status monitoring attribute deviation threshold, wherein the first-level operating status monitoring attribute deviation threshold, the second-level operating status monitoring attribute deviation threshold, and up to the M-level operating status monitoring attribute deviation threshold are in a logical AND relationship.

6. The method according to claim 5, characterized in that Performing a separate attribute deviation analysis on the operation status monitoring attribute set to obtain a first-level operation status monitoring attribute deviation threshold and a non-separate sensitive operation status monitoring attribute set, including: Extracting a first operating status monitoring attribute from the operating status monitoring attribute set; Setting a first disturbance step size of the first operating state monitoring attribute; Performing unique perturbations on the first operation status monitoring attribute with the first perturbation step length, and collecting operation monitoring logs of the target power network; Counting the proportion of negative samples in the operation monitoring log, wherein negative samples represent circuit abnormality samples; When the proportion of the number of negative samples is greater than or equal to the sample proportion threshold, and the first disturbance step length is less than or equal to the step length threshold, the first disturbance step length is set as the first operating state monitoring attribute deviation threshold; When the proportion of the number of negative samples is greater than or equal to the sample proportion threshold, and the first perturbation step length is greater than the step length threshold, the first perturbation step length is multiplied by 0.5 to obtain a second perturbation step length, and a separate attribute deviation cyclic analysis is performed according to the second perturbation step length; When the proportion of the number of negative samples is less than the sample proportion threshold, the first disturbance step is multiplied by 3 to obtain a third disturbance step to perform a separate attribute deviation cyclic analysis; When the proportion of the number of negative samples is less than the sample proportion threshold and the first disturbance step is greater than the zero-correlation step threshold, the first operating status monitoring attribute is added to the non-individually sensitive operating status monitoring attribute set.

7. The method according to claim 5, characterized in that Performing a binomial attribute enumeration combination deviation analysis on the non-individually sensitive running status monitoring attribute set to obtain a secondary running status monitoring attribute deviation threshold includes: Randomly extracting a first non-individually sensitive operating state monitoring attribute and a second non-individually sensitive operating state monitoring attribute from the non-individually sensitive operating state monitoring attribute set; Setting a first non-individually sensitive attribute disturbance step length of the first non-individually sensitive operating state monitoring attribute and a second non-individually sensitive attribute disturbance step length of the second non-individually sensitive operating state monitoring attribute; Using the first non-individually sensitive attribute disturbance step size and the second non-individually sensitive attribute disturbance step size, jointly disturb the first non-individually sensitive operating state monitoring attribute and the second non-individually sensitive operating state monitoring attribute, and collect a joint disturbance operation monitoring log of the target power network; Counting the proportion of the number of negative samples of the joint disturbance in the joint disturbance operation monitoring log; When the proportion of the number of negative samples of the joint disturbance is greater than or equal to the sample proportion threshold, and the first non-individual sensitive attribute disturbance step length is less than or equal to the first non-individual sensitive attribute disturbance step length threshold, and the second non-individual sensitive attribute disturbance step length is less than or equal to the second non-individual sensitive attribute disturbance step length threshold, the first non-individual sensitive attribute disturbance step length and the second non-individual sensitive attribute disturbance step length are configured as a logical AND relationship, and added into the secondary operation status monitoring attribute deviation threshold; When the proportion of the number of negative samples of the joint disturbance is greater than or equal to the sample proportion threshold, and at least one of the first non-individual sensitive attribute disturbance step length is greater than the first non-individual sensitive attribute disturbance step length threshold, and the second non-individual sensitive attribute disturbance step length is greater than the second non-individual sensitive attribute disturbance step length threshold is triggered, at least one of the triggered first non-individual sensitive attribute disturbance step length and the second non-individual sensitive attribute disturbance step length is multiplied by 0.5 and then a binomial attribute deviation cyclic analysis is performed; When the proportion of the number of negative samples of the joint disturbance is less than the sample proportion threshold, the first non-individual sensitive attribute disturbance step length and the second non-individual sensitive attribute disturbance step length are multiplied by 3 and then a binomial attribute deviation cyclic analysis is performed.

8. A multi-modal fusion power network operation status assessment device, characterized in that: A method for evaluating the operation status of a power network using multi-modal fusion according to any one of claims 1 to 7, wherein the device is embedded in a decision layer of a remote control circuit of a power network switch, and the device comprises: A parameter receiving module, used to receive the power network switch control state and power network operation state monitoring parameters of the target power network; A first processing module, configured to process the switch control state of the power network through an operation state analysis main network to obtain a first operation state prediction parameter; A second processing module, configured to process the power network switch control state through an operation state analysis secondary network to obtain a second operation state prediction parameter; an attribute summing module, used for summing the first running state prediction parameter and the second running state prediction parameter with the same attribute, and outputting the running state prediction parameter; A parameter comparison module, used for comparing the power network operation status monitoring parameter with the operation status prediction parameter to obtain an abnormal circuit of the power network; The disconnection control module is used to feed back the abnormal circuit of the power network to the decision layer of the remote control circuit of the power network switch to execute the disconnection control of the abnormal circuit switch.

Images