Detection and early warning method and device for metering secondary circuit of electric power system

Abstract

The invention discloses an electric power system metering secondary circuit detection early warning method and device, and relates to the technical field of electric power circuit detection early warning, and the method comprises the steps: providing a current signal for a current terminal box merging unit of any secondary circuit in an electric power system through a current test line of a portable calibrator, a voltage signal is provided for the voltage terminal box merging unit of the secondary circuit through the voltage wire barrow; obtaining a current signal sequence, a voltage signal sequence and an electric energy pulse signal; obtaining an iterative interaction current signal sequence and an iterative interaction voltage signal sequence; a target secondary loop comprehensive error is obtained; and when the target secondary loop comprehensive error is greater than a preset error threshold, generating first early warning information. According to the invention, the technical problems of hysteresis and low early warning reliability of secondary circuit detection early warning in the prior art are solved, and the technical effect of improving the accuracy of secondary circuit detection early warning is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of power circuit detection and early warning, and particularly to a method and device for detecting and early warning the metering secondary circuit of a power system. Background Art

[0002] Smart substations have absolute advantages in terms of safety, reliability, power quality, investment cost, maintenance difficulty, etc. Under this background, how to ensure the safe and stable operation of smart substations and measure and guarantee the stability of their metering devices has become an urgent problem to be solved. At present, with the construction of smart substations, the cost of measuring the error of metering devices in their secondary circuits and the economic losses caused by untimely and inaccurate measurement are also increasing. How to accurately, timely and conveniently measure the error of the metering device in the secondary circuit under various states of the smart substation and how to improve the measurement efficiency have attracted more and more attention from power enterprises.

[0003] The prior art has the technical problems of lag in secondary circuit detection and early warning and low reliability of early warning. Summary of the Invention

[0004] The present application provides a method and device for detecting and early warning the metering secondary circuit of a power system, which are used to solve the technical problems of lag in secondary circuit detection and early warning and low reliability of early warning in the prior art.

[0005] In view of the above problems, the present application provides a method and device for detecting and early warning the metering secondary circuit of a power system.

[0006] In the first aspect of the present application, a method for detecting and early warning the metering secondary circuit of a power system is provided. The method includes: Using the current test line of a portable calibrator to provide a current signal for the merging unit of the current terminal box of any secondary circuit in the power system, and using a voltage wire-releasing vehicle to provide a voltage signal for the merging unit of the voltage terminal box of the secondary circuit; Performing simultaneous data acquisition on the merging unit of the current terminal box and the merging unit of the voltage terminal box within a preset detection window, and using a radio frequency synchronization module to receive the power pulse signal of a digital watt-hour meter to obtain a current signal sequence, a voltage signal sequence and a power pulse signal; Performing multi-scale iterative interactive analysis on the current signal sequence and the voltage signal sequence to obtain an iterative interactive current signal sequence and an iterative interactive voltage signal sequence; Performing comprehensive error analysis of the secondary circuit based on the iterative interactive current signal sequence, the iterative interactive voltage signal sequence and the power pulse signal to obtain the comprehensive error of the target secondary circuit; When the comprehensive error of the target secondary circuit is greater than a preset error threshold, generating a first early warning message.

[0007] In the second aspect of the present application, a detection and early warning device for the secondary circuit of power system metering is provided. The device includes: A current signal providing module, configured to provide a current signal for the merging unit of the current terminal box of any secondary circuit in the power system by using the current test line of a portable calibrator, and provide a voltage signal for the merging unit of the voltage terminal box of the secondary circuit by using a voltage wire laying vehicle; A signal sequence obtaining module, configured to perform simultaneous time-series data acquisition on the merging unit of the current terminal box and the merging unit of the voltage terminal box within a preset detection window, and receive the power pulse signal of a digital watt-hour meter by using a radio frequency synchronization module to obtain a current signal sequence, a voltage signal sequence, and a power pulse signal; An iterative interaction analysis module, configured to perform multi-scale iterative interaction analysis on the current signal sequence and the voltage signal sequence to obtain an iterative interaction current signal sequence and an iterative interaction voltage signal sequence; A comprehensive error obtaining module, configured to perform secondary circuit comprehensive error analysis based on the iterative interaction current signal sequence, the iterative interaction voltage signal sequence, and the power pulse signal to obtain the comprehensive error of the target secondary circuit; A first early warning information generating module, configured to generate a first early warning information when the comprehensive error of the target secondary circuit is greater than a preset error threshold.

[0008] One or more technical solutions provided in the present application have at least the following technical effects or advantages: In the present application, a current signal is provided for the merging unit of the current terminal box of any secondary circuit in the power system by using the current test line of a portable calibrator, and a voltage signal is provided for the merging unit of the voltage terminal box of the secondary circuit by using a voltage wire laying vehicle. Then, simultaneous time-series data acquisition is performed on the merging unit of the current terminal box and the merging unit of the voltage terminal box within a preset detection window, and the power pulse signal of a digital watt-hour meter is received by using a radio frequency synchronization module to obtain a current signal sequence, a voltage signal sequence, and a power pulse signal. Furthermore, multi-scale iterative interaction analysis is performed on the current signal sequence and the voltage signal sequence to obtain an iterative interaction current signal sequence and an iterative interaction voltage signal sequence. Secondary circuit comprehensive error analysis is performed based on the iterative interaction current signal sequence and the iterative interaction voltage signal sequence to obtain the comprehensive error of the target secondary circuit. Then, when the comprehensive error of the target secondary circuit is greater than a preset error threshold, a first early warning information is generated. The technical effect of improving the reliability of secondary circuit detection and early warning is achieved. Description of the Drawings

[0009] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for description in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0010] Figure 1 Schematic diagram of the process of the power system metering secondary circuit detection and early warning method provided by the embodiments of the present application; Figure 2 Schematic diagram of the process of obtaining the iterative interaction current signal sequence and the iterative interaction voltage signal sequence in the power system metering secondary circuit detection and early warning method provided by the embodiments of the present application; Figure 3 Schematic diagram of the structure of the power system metering secondary circuit detection and early warning device provided by the embodiments of the present application.

[0011] Explanation of reference numerals: current signal providing module 11, signal sequence obtaining module 12, iterative interaction analysis module 13, comprehensive error obtaining module 14, first early warning information generating module 15. Detailed implementation manners

[0012] The present application provides a power system metering secondary circuit detection and early warning method and device, which are used to solve the technical problems of lag and low warning reliability in secondary circuit detection and early warning in the prior art.

[0013] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts belong to the scope of protection of the present application.

[0014] It should be noted that the terms "including" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, device, product, or server including a series of steps or units does not necessarily have to be limited to those clearly listed steps or units, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products, or devices.

[0015] Embodiment 1, as Figure 1 shown, the present application provides a power system metering secondary circuit detection and early warning method, and the method includes: S1: Use the current test line of the portable calibrator to provide a current signal for the merging unit of the current terminal box of any secondary circuit in the power system, and use the voltage wire laying vehicle to provide a voltage signal for the merging unit of the voltage terminal box of the secondary circuit; In a possible embodiment, the portable calibrator is a portable intelligent substation metering secondary circuit on-site calibrator, which can provide analog signals for various devices in the power system or measure the accuracy of the devices. The portable intelligent substation metering secondary circuit on-site calibrator is convenient for on-site operation and helps verify the accuracy of power system devices by providing accurate current and voltage signals. The current test line is the part of the portable calibrator that provides current signals for the current terminal box merging unit in the power system. The current test line inputs the current signal into the power system by simulating the current signal to test the system response. The voltage wire-releasing vehicle is a device used to provide voltage signals for the power system on-site, and it can transmit the voltage signal to the voltage terminal box merging unit. Through the voltage wire-releasing vehicle, the power equipment can be tested without disturbing the system operation.

[0016] Preferably, the current terminal box merging unit is an important component in the secondary circuit, which is responsible for summarizing and processing multiple current signals to accurately measure the current flowing through the system. The voltage terminal box merging unit is used to summarize multiple voltage signals in the power system to ensure the accuracy of voltage data for power quality analysis in the secondary circuit.

[0017] By providing a current signal to the current terminal box merging unit through the current test line to simulate the current situation in the real working environment, and providing a voltage signal to the voltage terminal box merging unit through the voltage wire-releasing vehicle to accurately simulate the voltage change situation, thus making preliminary preparations for the detection and early warning of the secondary circuit in the power system. Thereby, it is ensured that during the test process, each merging unit of the secondary circuit can receive standard test signals, achieving the technical effect of providing a basis for subsequent data acquisition and error detection.

[0018] S2: Simultaneously collect time-sequential data for the current terminal box merging unit and the voltage terminal box merging unit within a preset detection window, and use a radio frequency synchronization module to receive the power pulse signal of the digital watt-hour meter to obtain a current signal sequence, a voltage signal sequence, and a power pulse signal; In a possible embodiment, the preset detection window is a test period preset within a time range. The length and detection frequency of the preset detection window are set by those skilled in the art according to the operating characteristics of the equipment, test requirements, and system load conditions. The radio frequency synchronization module is a high-precision wireless synchronization technology device, which ensures data synchronization between different devices through radio frequency technology, can effectively receive the power pulse signal from the digital watt-hour meter, and then obtain accurate power data. The digital watt-hour meter outputs the consumed power information in the form of power pulses, and these pulse signals are used to accurately record the power consumption. Each power pulse represents a certain amount of power consumption.

[0019] Preferably, the current signal and the voltage signal are synchronously collected within the same time period, and the wireless radio frequency synchronization module is used to receive the power pulse signal of the digital electric energy meter, and the time sequence alignment of the collected data is ensured through the time synchronization protocol (such as IEEE1588) of the intelligent substation, so as to avoid the situation of data asynchronization.

[0020] The current signal sequence, the voltage signal sequence, and the power pulse signal are respectively time sequence records of the current, voltage, and power pulse signals. The current signal sequence records the change of the current within the preset detection window, the voltage signal sequence records the change of the voltage within the preset detection window, and the power pulse signal is used to record the pulse signal output of the electric energy. Thus, a comprehensive record of the operating state of the secondary circuit of the power system is realized.

[0021] At the same time, multi-dimensional data such as current, voltage, and power pulse are acquired to provide accurate measurement data. Through synchronous data acquisition, the operating states of various metering devices in the power system can be accurately tracked, ensuring that potential electrical faults or abnormalities can be detected in a timely manner, thereby improving the reliability and accuracy of the measurement.

[0022] S3: Perform multi-scale iterative interaction analysis on the current signal sequence and the voltage signal sequence to obtain an iterative interaction current signal sequence and an iterative interaction voltage signal sequence; In one embodiment, the current signal sequence and the voltage signal sequence are subjected to feature analysis at different scales, so as to capture the transient changes and long-term trends of the secondary circuit within the preset detection window, and obtain the more representative iterative interaction current signal and iterative interaction voltage signal, providing more reliable analysis data for subsequent error detection and early warning analysis.

[0023] Preferably, multi-scale analysis is first performed on the current signal sequence and the voltage signal sequence. Signal features at different time scales are extracted and interactively calculated to ensure that short-time and long-time feature information can be fused. For example, a short-time window can identify sudden current pulse anomalies, while a long-time window can detect slow-changing trend deviations. Subsequently, an iterative interaction analysis method is adopted to perform multiple rounds of updates on the features at different time scales. By constructing an iterative interaction matrix, the current and voltage can perform feature propagation between different scales, thereby forming a more robust anomaly detection ability.

[0024] By obtaining iterative interaction current signals and iterative interaction voltage signals that are more stable than the original signals, the technical effect of reducing errors caused by data noise or single-scale analysis is achieved.

[0025] S4: Based on the iterative interaction current signal sequence, iterative interaction voltage signal sequence, and power pulse signal, perform comprehensive error analysis of the secondary circuit to obtain the target comprehensive error of the secondary circuit; S5: When the target comprehensive error of the secondary circuit is greater than the preset error threshold, generate a first warning message.

[0026] In an embodiment of the present application, a standard watt-hour meter measurement module (included in the portable calibrator) is used to calculate the electric energy of the iterative interaction current signal and iterative interaction voltage signal to obtain the standard electric energy. Then, the difference between the calculated standard electric energy and the iterative interaction power pulse signal is calculated, and the calculation result is divided by the standard electric energy to obtain the target comprehensive error of the secondary circuit.

[0027] Preferably, the standard watt-hour meter measurement module is embedded with a standard electric energy calculation formula, where the standard electric energy calculation formula is: ; is the standard electric energy, is the instantaneous power, is the preset detection window, is the iterative interaction voltage signal sequence, is the iterative interaction current signal sequence. The standard electric energy calculation formula is used to calculate the standard electric energy of the iterative interaction current signal sequence and iterative interaction voltage signal sequence.

[0028] In one embodiment, the preset error threshold is the maximum error for the normal operation of the secondary circuit preset by those skilled in the art. When the target comprehensive error of the secondary circuit is greater than the preset error threshold, it indicates that there is an abnormality in the secondary circuit at this time, and the first warning message is generated. Among them, the first warning message is used to remind the staff that there is an abnormality in the operation of the secondary circuit of the power system. By improving the data quality through iterative interaction signals, the technical effects of ensuring the reliability and accuracy of error analysis and the timeliness of warning are achieved.

[0029] Further, as Figure 2 shown, perform multi-scale iterative interaction analysis on the current signal sequence and voltage signal sequence to obtain the iterative interaction current signal sequence and iterative interaction voltage signal sequence. Step S3 of the embodiment of the present application further includes: Taking the current signal and voltage signal as a node respectively, constructing an initialization detection topology network, and constructing a detection topology network sequence in combination with the current signal sequence and voltage signal sequence; Obtaining a set of feature extraction scales, and respectively using the feature extraction scales in the set of feature extraction scales as receptive fields to construct a set of adaptive feature extractors; Use the set of adaptive feature extractors to extract features from the detection topology network sequence respectively, obtaining a set of detection topology network features; Perform iterative interaction analysis on the set of detection topology network features to obtain target iterative interaction detection topology network features; Perform a convolution operation on the target iterative interaction detection topology network features and the last detection topology network in the detection topology network sequence to obtain an iterative interaction detection topology network, and extract an iterative interaction current signal sequence and an iterative interaction voltage signal sequence from the iterative interaction detection topology network.

[0030] In a possible embodiment, taking current signals and voltage signals as nodes, and taking the associations between them (such as temporal correlation, physical connection relationship, etc.) as edges, an initial detection topology structure of the secondary circuit of the power system is constructed. This initial detection topology network is used to construct the synchronization relationship between current and voltage, and perform feature extraction and optimization in subsequent analysis processes. Input the current signal sequence and voltage signal sequence into the initial detection topology network according to the same time sequence to obtain the detection topology network sequence. Among them, the detection topology network sequence is used to reflect the synchronous change situation of current and voltage over time.

[0031] The set of feature extraction scales refers to a set of parameters used to capture features in different time scales and spatial ranges. The receptive field refers to the range that each node can perceive during the feature extraction process. For example, only local changes can be captured within a short time window, while global trends can be recognized in a large range. By setting different receptive fields, features of different scales can be adaptively extracted from signal characteristics.

[0032] Preferably, the set of adaptive feature extractors is composed of multiple feature extractors, each extractor corresponding to a different receptive field and used to adapt to signal patterns of different scales. Through the combination of multiple adaptive feature extractors, more comprehensive signal features can be extracted to obtain the set of detection topology network features. Among them, the set of topology network detection features reflects the operation feature situation of current and voltage within a preset detection window, including temporal features (mean, standard deviation, and skewness), frequency domain features (harmonic components, energy distribution, and main frequency), etc. It achieves the technical effect of improving the accuracy of signal analysis.

[0033] Preferably, the feature extraction scale set is used as the receptive field set, and corresponding multiple sample detection topology network sequence sets and multiple sample detection topology network feature sets are obtained as training data. Using the receptive field set and the corresponding multiple sample detection topology network sequence sets as inputs, and the multiple sample detection topology network feature sets as outputs, the framework set constructed based on the feedforward neural network is supervised and trained until convergence, and the trained adaptive feature extractor set is obtained. By constructing the adaptive feature extractor set, the technical effect of paving the way for subsequent multi-scale feature extraction is achieved.

[0034] By performing iterative interaction analysis on the detection topology network feature set, the detection topology network features at different scales are calculated and optimized multiple times, so that the feature expression is more stable. Exemplarily, the current fluctuation at a certain scale may affect the stability of the entire circuit. Through iterative analysis, the fluctuation conditions at other scales can be comprehensively considered, so as to optimize the contribution of this fluctuation to the overall error calculation. Furthermore, through convolution operation, the target iterative interaction detection topology network feature is fused with the last detection topology network in the detection topology network sequence. This process can further optimize the feature expression, so that the finally extracted signal sequence not only contains the original data information, but also integrates the signal features, achieving the technical effect of improving the robustness of the signal.

[0035] Preferably, multiple sample iterative interaction detection topology network features, multiple sample detection topology networks, and corresponding multiple sample iterative interaction detection topologies are obtained as training data, and the framework constructed based on the convolutional neural network is supervised and trained until convergence, and the trained convolutional operation network layer is obtained. The convolutional operation network layer is used to perform convolutional analysis on the target iterative interaction detection topology network feature and the last detection topology network in the detection topology network sequence to obtain the iterative interaction detection topology network.

[0036] Furthermore, the iterative interaction current signal sequence and the iterative interaction voltage signal sequence are extracted from the final iterative interaction detection topology network. Compared with the original signal sequence, these signal sequences have higher signal-to-noise ratio and more stable feature expression, which is helpful for subsequent comprehensive error analysis and detection warning.

[0037] Further, in performing iterative interaction analysis on the detection topology network feature set to obtain the target iterative interaction detection topology network feature, step S3 of the embodiment of the present application further includes: Randomly and without replacement, the first detection topology network feature and the second detection topology network feature are extracted from the detection topology network feature set; Performing iterative interaction analysis on the first detection topology network feature and the second detection topology network feature to obtain the first iterative interaction detection topology network feature; Randomly and without replacement, extract the third detected topological network feature from the set of detected topological network features, and perform iterative interaction analysis on the third detected topological network feature and the first iterative interaction detected topological network feature to obtain the second iterative interaction detected topological network feature; And so on, based on the (N - 2)-th iterative interaction detected topological network feature, perform iterative interaction analysis on the N-th detected topological network feature randomly and without replacement extracted from the set of detected topological network features to obtain the target iterative interaction detected topological network feature, where N is the number of detected topological network features in the set of detected topological network features, and N is an integer greater than or equal to 1.

[0038] Further, perform iterative interaction analysis on the first detected topological network feature and the second detected topological network feature to obtain the first iterative interaction detected topological network feature. Step S3 of the embodiment of the present application further includes: Perform inner product mapping on the first detected topological network feature and the second detected topological network feature to obtain the first similarity set; Perform normalization processing on the first similarity set and add the normalization processing result to an initially empty matrix to obtain the first iterative interaction matrix; Perform convolution operation on the first iterative interaction matrix and the second detected topological network feature to obtain the first iterative interaction detected topological network feature.

[0039] In a possible embodiment, by randomly and without replacement extracting the first detected topological network feature and the second detected topological network feature from the set of detected topological network features, it is ensured that the features extracted in each iteration are new, avoiding redundancy. Furthermore, perform the first round of interaction analysis, and obtain the first iterative interaction detected topological network feature by performing feature fusion on the first detected topological network feature and the second detected topological network feature. Wherein, the first iterative interaction detected topological network feature is the feature of the second detected topological network feature after complementary enhancement with the first detected topological network feature, achieving the goal of initially fusing features from different sources.

[0040] Randomly extract the third detected topological network features without replacement again, and based on the results of the first-round interaction, conduct a new interaction analysis with the first iterative interaction detected topological network features. Based on the same principle as obtaining the first iterative interaction detected topological network features, more representative features are obtained, and the second iterative interaction detected topological network features are obtained. Based on continuously conducting iterative interaction analysis until all features are used. Until the N-1th round is reached, based on the N-2 iterative interaction detected topological network features, the Nth detected topological network features are introduced for the final interaction analysis to obtain the target iterative interaction detected topological network features. Wherein, N is the number of detected topological network features in the set of detected topological network features, and N is an integer greater than or equal to 1. During the iterative process, each round of iteration can optimize the features to make them more suitable for error detection and early warning analysis. And through random sampling without replacement, it is ensured that all features are fully utilized and not reused. Preferably, through an interactive method, features of different scales and different categories are fully fused, achieving the technical effect of enhancing the sensitivity to errors and abnormal states.

[0041] In an embodiment of the present application, by performing inner product mapping, similarity normalization, constructing an interaction matrix, and convolution operation on the first detected topological network features and the second detected topological network features, the first iterative interaction detected topological network features are finally obtained. Preferably, the cosine similarity calculation formula is used to calculate the corresponding feature similarities of the first detected topological network features and the second detected topological network features respectively to obtain the first similarity set. Wherein, the first similarity set is used to describe the similarity degree between the first detected topological network features and the second detected topological network features. Furthermore, the softmax normalization formula is used to normalize the first similarity set to ensure that the similarity range is between. And the normalization result is added to the initially empty matrix to obtain the first iterative interaction matrix. Wherein, the first iterative interaction matrix reflects the interaction weight between the first detected topological network features and the second detected topological network features.

[0042] Furthermore, using graph convolution operation, the first iterative interaction matrix is used as the adjacency weight and convolved with the second detected topological network features. Preferably, the convolution calculation formula is: , where is the first iterative interaction detected topological network feature, is the activation function (such as ReLU), is the first iterative interaction matrix, is the second detected topological network feature, A weight matrix pre-constructed for those skilled in the art. Through performing convolution operations, the first iterative interaction detection topology network features are obtained. The technical effect of further optimizing the detection topology network features and endowing them with stronger feature expression capabilities is achieved.

[0043] Further, to obtain a set of feature extraction scales, step S3 of the embodiment of the present application further includes: Obtain the set of detection anomaly logs of the secondary circuit within the historical window; Taking the anomaly type as the index, perform the same-type aggregation on the set of detection anomaly logs to obtain L sets of clustered detection anomaly logs, where L is an integer greater than or equal to 1; Perform two-dimensional centralized identification of the anomaly time scale on the L sets of clustered detection anomaly logs to obtain L sets of centralized identification anomaly time scales, where the two-dimensional centralized identification of the anomaly time scale is to perform identification in an equilibrium manner from two dimensions: the similarity degree of time nodes of the clustered detection anomaly logs and the similarity degree of the anomaly time sizes of the clustered detection anomaly logs; Take the union of the L sets of centralized identification anomaly time scales to obtain the set of feature extraction scales.

[0044] Further, to perform two-dimensional centralized identification of the anomaly time scale on the L sets of clustered detection anomaly logs to obtain L sets of centralized identification anomaly time scales, step S3 of the embodiment of the present application further includes: Randomly select a preset number of clustered detection anomaly logs from each of the L sets of clustered detection anomaly logs as L sets of centralized identification center sets; Construct the neighborhoods of the centralized identification centers within each of the L sets of centralized identification center sets according to a preset identification threshold to obtain L sets of centralized identification center neighborhood sets; Perform loss analysis on the L sets of centralized identification center neighborhood sets using a two-dimensional centralized identification loss function to obtain L two-dimensional centralized identification loss amounts; When the L two-dimensional centralized identification loss amounts are less than or equal to a preset loss amount threshold, calculate the neighborhood means of the L sets of centralized identification center neighborhood sets respectively to obtain L sets of centralized identification anomaly time scales.

[0045] Further, the two-dimensional centralized identification loss function is: ; Where is the two-dimensional centralized identification loss amount, is the number of centralized identification centers within a set of centralized identification centers, is the centralized identification center neighborhood of the th centralized identification center within a set of centralized identification centers, is the The time node of the th clustering detection abnormal log within the neighborhood of the centralized recognition center of the centralized recognition center, is the time node of the clustering detection abnormal log corresponding to the th centralized recognition center, is the abnormal time size of the clustering detection abnormal log corresponding to the th centralized recognition center, is the abnormal time size of the th clustering detection abnormal log within the neighborhood of the centralized recognition center of the th centralized recognition center, is the weight for balancing the similarity degree of time nodes and the similarity degree of abnormal time sizes.

[0046] In a possible embodiment, during the operation of the secondary circuit of an intelligent substation, the system records various abnormal situations (such as current abnormality, voltage abnormality, error exceeding the standard, etc.). A historical time window (such as the most recent month or year) is set, and the detection abnormal log set within this time range is extracted to obtain the said detection abnormal log set. Indexed by abnormal types (such as current fluctuation abnormality, voltage imbalance, comprehensive error exceeding the standard, etc.), the logs are clustered to obtain the log set corresponding to each abnormal type, that is, the said L clustering detection abnormal log sets. Wherein, L is the abnormal type that occurred in the secondary circuit within the historical window.

[0047] Furthermore, a two-dimensional centralized recognition of the abnormal time scale is performed on the said L clustering detection abnormal log sets, that is, the recognition is balanced from two dimensions: the similarity degree of time nodes of the clustering detection abnormal logs and the similarity degree of the abnormal time sizes of the clustering detection abnormal logs, to obtain the said L centralized recognition abnormal time scale sets, so as to capture the time scale when the abnormality occurs and provide data support for subsequent analysis. Then, the union of the said L centralized recognition abnormal time scale sets is obtained to avoid scale duplication and waste of computing power, thereby obtaining the said feature extraction scale set.

[0048] Randomly select a preset number of clustering detection abnormal logs from each of the L clustering detection abnormal log sets as the L centralized recognition center sets. The preset number is set by those skilled in the art themselves. Each centralized recognition center is a representative sample initially selected from the clustering logs, used to construct a benchmark for abnormal recognition and improve the accuracy of clustering. The preset recognition threshold is the similarity range with the centralized recognition center when constructing the neighborhood of the centralized recognition center, which is preset by those skilled in the art. When the similarity between the clustering detection abnormal logs in the L clustering detection abnormal log sets and the centralized recognition centers of the corresponding L centralized recognition center sets meets the preset recognition threshold, they are added to the corresponding neighborhood of the centralized recognition center, thereby obtaining the said L centralized recognition center neighborhood sets.

[0049] Analyze the loss of the L neighborhood sets of the centralized recognition centers by using the two-dimensional centralized recognition loss function to obtain L two-dimensional centralized recognition loss amounts. Among them, the two-dimensional centralized recognition loss function is used to analyze the recognition division loss degree of the neighborhood set of the centralized recognition center. When the L two-dimensional centralized recognition loss amounts are less than or equal to the preset loss amount threshold, calculate the neighborhood means of the L neighborhood sets of the centralized recognition centers respectively to obtain L centralized recognition abnormal time scale sets.

[0050] Improve the pertinence of time scale division, reduce false alarms through anomaly type clustering, and ensure that the feature extraction scale adapts to the characteristics of different types of anomalies through two-dimensional centralized recognition, improve the robustness of the model, and then combine the similarity of time nodes and the similarity of anomaly time sizes to improve the rationality of scale division and make the detection more accurate.

[0051] In summary, the embodiments of the present application have at least the following technical effects: The present application uses the current test line of the portable calibrator to provide a current signal for the merging unit of the current terminal box of any secondary circuit in the power system, and uses the voltage wire laying vehicle to provide a voltage signal for the merging unit of the voltage terminal box of the secondary circuit. Then, simultaneous time-series data acquisition is performed on the merging unit of the current terminal box and the merging unit of the voltage terminal box within a preset detection window, and the wireless radio frequency synchronization module is used to receive the power pulse signal of the digital watt-hour meter to obtain the current signal sequence, voltage signal sequence, and power pulse signal. Furthermore, multi-scale iterative interaction analysis is performed on the current signal sequence and the voltage signal sequence to obtain the iterative interaction current signal sequence and the iterative interaction voltage signal sequence. Based on the iterative interaction current signal sequence and the iterative interaction voltage signal sequence, comprehensive error analysis of the secondary circuit is performed to obtain the comprehensive error of the target secondary circuit. Then, when the comprehensive error of the target secondary circuit is greater than the preset error threshold, a first warning message is generated. The technical effect of improving the reliability of secondary circuit detection and warning is achieved.

[0052] Embodiment 2, based on the same inventive concept as the power system metering secondary circuit detection and warning method in the foregoing embodiment, as Figure 3 shown, the present application provides a power system metering secondary circuit detection and warning device. The device in the embodiments of the present application and the method embodiments are based on the same inventive concept. Among them, the device includes: A current signal providing module 11, configured to use the current test line of the portable calibrator to provide a current signal for the merging unit of the current terminal box of any secondary circuit in the power system, and use the voltage wire laying vehicle to provide a voltage signal for the merging unit of the voltage terminal box of the secondary circuit; The signal sequence acquisition module 12 is configured to perform simultaneous time-series data acquisition on the current terminal box merging unit and the voltage terminal box merging unit within a preset detection window, and receive the power pulse signal of the digital electric energy meter by using the radio frequency synchronization module, so as to obtain a current signal sequence, a voltage signal sequence, and a power pulse signal; The iterative interaction analysis module 13 is configured to perform multi-scale iterative interaction analysis on the current signal sequence and the voltage signal sequence to obtain an iterative interaction current signal sequence and an iterative interaction voltage signal sequence; The comprehensive error acquisition module 14 is configured to perform secondary circuit comprehensive error analysis based on the iterative interaction current signal sequence, the iterative interaction voltage signal sequence, and the power pulse signal to obtain the target secondary circuit comprehensive error; The first warning information generation module 15 is configured to generate a first warning information when the target secondary circuit comprehensive error is greater than a preset error threshold.

[0053] Further, the iterative interaction analysis module 13 is configured to perform the following steps: Taking the current signal and the voltage signal as a node respectively, constructing an initial detection topology network, and constructing a detection topology network sequence by combining the current signal sequence and the voltage signal sequence; Obtaining a set of feature extraction scales, and respectively taking the feature extraction scales in the set of feature extraction scales as receptive fields to construct a set of adaptive feature extractors; Using the set of adaptive feature extractors to perform feature extraction on the detection topology network sequence respectively to obtain a detection topology network feature set; Performing iterative interaction analysis on the detection topology network feature set to obtain target iterative interaction detection topology network features; Performing a convolution operation on the target iterative interaction detection topology network features and the last detection topology network in the detection topology network sequence to obtain an iterative interaction detection topology network, and extracting an iterative interaction current signal sequence and an iterative interaction voltage signal sequence from the iterative interaction detection topology network.

[0054] Further, the iterative interaction analysis module 13 is configured to perform the following steps: Randomly and without replacement extracting a first detection topology network feature and a second detection topology network feature from the detection topology network feature set; Performing iterative interaction analysis on the first detection topology network feature and the second detection topology network feature to obtain a first iterative interaction detection topology network feature; Randomly and without replacement extracting a third detection topology network feature from the detection topology network feature set, and performing iterative interaction analysis on the third detection topology network feature and the first iterative interaction detection topology network feature to obtain a second iterative interaction detection topology network feature; And so on, based on the (N-2)-th iterative interaction to detect the topological network features, iterative interaction analysis is performed on the N-th detected topological network features randomly and without replacement extracted from the set of detected topological network features, to obtain the target iterative interaction detected topological network features, where N is the number of detected topological network features in the set of detected topological network features, and N is an integer greater than or equal to 1.

[0055] Further, the iterative interaction analysis module 13 is configured to perform the following steps: Perform inner product mapping on the first detected topological network feature and the second detected topological network feature to obtain a first similarity set; Perform normalization processing on the first similarity set, and add the normalization processing result to an initially empty matrix to obtain a first iterative interaction matrix; Perform convolution operation on the first iterative interaction matrix and the second detected topological network feature to obtain a first iterative interaction detected topological network feature.

[0056] Further, the iterative interaction analysis module 13 is configured to perform the following steps: Obtain the set of detection abnormal logs of the secondary circuit within the historical window; Taking the abnormal type as the index, perform aggregation of the same type on the set of detection abnormal logs to obtain L sets of clustered detection abnormal logs, where L is an integer greater than or equal to 1; Perform two-dimensional centralized identification of abnormal time scales on the L sets of clustered detection abnormal logs to obtain L sets of centralized identification abnormal time scales, where the two-dimensional centralized identification of abnormal time scales is to perform identification evenly from two dimensions: the similarity degree of time nodes of the clustered detection abnormal logs and the similarity degree of the sizes of abnormal times of the clustered detection abnormal logs; Perform union operation on the L sets of centralized identification abnormal time scales to obtain a set of feature extraction scales.

[0057] Further, the iterative interaction analysis module 13 is configured to perform the following steps: Randomly select a preset number of clustered detection abnormal logs from each of the L sets of clustered detection abnormal logs as L sets of centralized identification center sets; Construct neighborhoods of the centralized identification centers within the L sets of centralized identification center sets according to a preset identification threshold to obtain L sets of centralized identification center neighborhood sets; Perform loss analysis on the L sets of centralized identification center neighborhood sets by using a two-dimensional centralized identification loss function to obtain L two-dimensional centralized identification loss amounts; When the loss amounts of the L two-dimensional centralized identifications are less than or equal to a preset loss amount threshold, calculate the neighborhood means of the L neighborhood sets of the centralized identification centers respectively to obtain L sets of centralized identification abnormal time scales.

[0058] Further, the two-dimensional centralized identification loss function is: ; Wherein, is the loss amount of the two-dimensional centralized identification, is the number of centralized identification centers in a set of centralized identification centers, is the centralized identification center neighborhood of the th centralized identification center in a set of centralized identification centers, is the th time node of the clustering detection abnormal log in the centralized identification center neighborhood of the th centralized identification center, is the time node of the clustering detection abnormal log corresponding to the th centralized identification center, is the abnormal time size of the clustering detection abnormal log corresponding to the th centralized identification center, is the abnormal time size of the th clustering detection abnormal log in the centralized identification center neighborhood of the th centralized identification center, is the weight for balancing the similarity degree of time nodes and the similarity degree of abnormal time sizes.

[0059] It should be noted that the above sequence of embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above description of specific embodiments of this specification is made. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0060] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included within the protection scope of the present application.

[0061] This specification and the drawings are only exemplary descriptions of the present application and are considered to have covered any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. A detection and early warning method for secondary circuit measurement in a power system, characterized in that: Methods include: Use the current test line of the portable calibrator to provide current signals for the current terminal box merging unit of any secondary circuit in the power system, and use the voltage pay-out vehicle to provide voltage signals for the voltage terminal box merging unit of the secondary circuit; Performing simultaneous data acquisition on the current terminal box merging unit and the voltage terminal box merging unit within a preset detection window, and receiving the electric energy pulse signal of the digital electric energy meter using a wireless radio frequency synchronization module to obtain a current signal sequence, a voltage signal sequence and an electric energy pulse signal; Perform multi-scale iterative interactive analysis on the current signal sequence and the voltage signal sequence to obtain an iterative interactive current signal sequence and an iterative interactive voltage signal sequence; Based on the iterative interactive current signal sequence, iterative interactive voltage signal sequence and electric energy pulse signal, the comprehensive error analysis of the secondary circuit is performed to obtain the target comprehensive error of the secondary circuit; When the target secondary loop comprehensive error is greater than a preset error threshold, a first warning message is generated.

2. The electric power system metering secondary circuit detection and early warning method according to claim 1, characterized in that: Perform multi-scale iterative interactive analysis on the current signal sequence and the voltage signal sequence to obtain an iterative interactive current signal sequence and an iterative interactive voltage signal sequence, including: The current signal and the voltage signal are respectively used as a node to construct an initialization detection topology network, and the detection topology network sequence is constructed by combining the current signal sequence and the voltage signal sequence; Acquire a feature extraction scale set, use the feature extraction scales in the feature extraction scale set as receptive fields respectively, and construct an adaptive feature extractor set; Using the adaptive feature extractor set to extract features from the detection topology network sequence respectively to obtain a detection topology network feature set; Performing iterative interactive analysis on the detection topology network feature set to obtain target iterative interactive detection topology network features; The target iterative interactive detection topology network feature is convolved with the last detection topology network in the detection topology network sequence to obtain an iterative interactive detection topology network, and an iterative interactive current signal sequence and an iterative interactive voltage signal sequence are extracted from the iterative interactive detection topology network.

3. The electric power system metering secondary circuit detection and early warning method according to claim 2, characterized in that: Performing iterative interactive analysis on the detection topology network feature set to obtain target iterative interactive detection topology network features, including: randomly extracting a first detection topology network feature and a second detection topology network feature from the detection topology network feature set without replacement; Perform iterative interactive analysis on the first detection topology network feature and the second detection topology network feature to obtain a first iterative interactive detection topology network feature; randomly extracting a third detection topology network feature from the detection topology network feature set without replacement, performing iterative interactive analysis on the third detection topology network feature and the first iterative interactive detection topology network feature, and obtaining a second iterative interactive detection topology network feature; By analogy, based on the N-2th iterative interactive detection topology network feature, an iterative interactive analysis is performed on the Nth detection topology network feature randomly extracted from the detection topology network feature set without replacement to obtain the target iterative interactive detection topology network feature, where N is the number of detection topology network features in the detection topology network feature set, and N is an integer greater than or equal to 1.

4. The electric power system metering secondary circuit detection and early warning method according to claim 3 is characterized in that: Performing iterative interactive analysis on the first detection topology network feature and the second detection topology network feature to obtain a first iterative interactive detection topology network feature includes: Perform inner product mapping on the first detection topology network feature and the second detection topology network feature to obtain a first similarity set; Normalizing the first similarity set, and adding the normalized result to the initially empty matrix to obtain a first iterative interaction matrix; A convolution operation is performed on the first iterative interaction matrix and the second detection topology network feature to obtain a first iterative interaction detection topology network feature.

5. The electric power system metering secondary circuit detection and early warning method according to claim 2, characterized in that: Get the feature extraction scale set, including: Obtaining a set of detection anomaly logs of the secondary circuit within a historical window; Taking the anomaly type as an index, the detection anomaly log set is aggregated of the same type to obtain L clustered detection anomaly log sets, where L is an integer greater than or equal to 1; Performing dual-dimensional centralized identification of abnormal time scales on the L cluster detection abnormal log sets to obtain L centralized identification abnormal time scale sets, wherein the dual-dimensional centralized identification of abnormal time scales is to perform identification in a balanced manner from two dimensions: the similarity of time nodes of cluster detection abnormal logs and the similarity of abnormal time sizes of cluster detection abnormal logs; The L concentrated abnormal identification time scale sets are unioned to obtain a feature extraction scale set.

6. The electric power system metering secondary circuit detection and early warning method according to claim 5, characterized in that: Performing two-dimensional centralized identification of abnormal time scales on the L cluster detection abnormal log sets to obtain L centralized identification abnormal time scale sets, including: Randomly select a preset number of cluster detection anomaly logs from the L cluster detection anomaly log sets as L centralized identification center sets; According to the preset recognition threshold, the neighborhoods of the centralized recognition centers in the L centralized recognition center sets are respectively constructed to obtain L centralized recognition center neighborhood sets; Using a two-dimensional centralized recognition loss function to perform loss analysis on the L centralized recognition center neighborhood sets, obtaining L two-dimensional centralized recognition loss amounts; When the L two-dimensional concentrated identification loss amounts are less than or equal to a preset loss amount threshold, the neighborhood means of the L concentrated identification center neighborhood sets are calculated respectively to obtain L concentrated identification abnormal time scale sets.

7. The electric power system metering secondary circuit detection and early warning method according to claim 6, characterized in that: The loss function of two-dimensional centralized recognition is: ； in, To identify the loss amount in two-dimensional concentration, is the number of centralized identification centers in a centralized identification center set, For a centralized identification center set The neighborhood of the centralized identification center of the centralized identification center, For the The first Clustering detects the time nodes of abnormal logs, For the The time node of the cluster detection anomaly log corresponding to the centralized identification center, For the The abnormal time size of the cluster detection abnormal log corresponding to the centralized identification center, For the The first The abnormal time size of the abnormal log detected by clustering, The weight is used to balance the similarity between time nodes and the similarity between abnormal time sizes.

8. The power system metering secondary circuit detection and early warning device is characterized by: The device is used to implement the power system metering secondary circuit detection and early warning method according to any one of claims 1 to 7, comprising: A current signal providing module is used to provide a current signal to a current terminal box merging unit of any secondary circuit in the power system by using a current test line of a portable calibrator, and to provide a voltage signal to a voltage terminal box merging unit of a secondary circuit by using a voltage pay-out vehicle; A signal sequence acquisition module is used to collect simultaneous data of the current terminal box merging unit and the voltage terminal box merging unit within a preset detection window, and to receive the electric energy pulse signal of the digital electric energy meter using a wireless radio frequency synchronization module to obtain a current signal sequence, a voltage signal sequence and an electric energy pulse signal; An iterative interactive analysis module is used to perform multi-scale iterative interactive analysis on the current signal sequence and the voltage signal sequence to obtain an iterative interactive current signal sequence and an iterative interactive voltage signal sequence; A comprehensive error acquisition module is used to perform a comprehensive error analysis of the secondary circuit based on an iterative interactive current signal sequence, an iterative interactive voltage signal sequence and an electric energy pulse signal to obtain a target comprehensive error of the secondary circuit; The first warning information generating module is used to generate the first warning information when the target secondary circuit comprehensive error is greater than a preset error threshold.

Images