A power grid transient instability mode intelligent identification scheduling system

Abstract

The present application relates to the technical field of intelligent identification and scheduling, and relates to a power grid transient instability mode intelligent identification and scheduling system. The system comprises a data collection module, a data screening module, a node matching module, a difference analysis module and a scheduling sorting module. The data collection module is used to collect historical operation data and real-time operation data of the power grid, extract transient instability data from the historical operation data, and extract instability severity values from the transient instability data. The data screening module is used to set a screening data length based on the instability severity values, and then screen the historical operation data according to the instability severity values of the transient instability data in the historical operation data. The historical data of the corresponding length is matched through grade interval division, which ensures that the instability event analysis can cover the complete evolution process before the instability, and provides more targeted historical data support for subsequent instability tracing and risk analysis.

Description

Technical Field

[0001] This invention relates to the field of intelligent identification and scheduling technology, and more specifically, to an intelligent identification and scheduling system for power grid transient instability modes. Background Technology

[0002] In the field of power system dispatch and security control, the technology of power grid transient instability mode identification and dispatch control is the core technology to ensure the safe and stable operation of the power grid. Its core role is to accurately identify transient instability events that occur during power grid operation, locate the source of instability, assess instability risks, and generate targeted defense dispatch strategies.

[0003] Currently, most methods rely on statistical analysis of historical power grid operation data, fault determination based on fixed mechanism models, or local identification of single instability types or single power grid nodes. These methods mainly focus on instability events that have already occurred in the power grid or pre-set typical faults, which is no longer sufficient to meet the core objective of precise prevention and control of power grid transient instability risks. Existing technologies are mostly based on static analysis of historical data and lack the ability to actively explore and identify unknown and complex transient instability modes, which can easily create risk blind spots. When new instability events occur in the power grid, problems such as identification lag and judgment bias are likely to occur. Secondly, there is a disconnect between instability identification and source location. Some technologies can only identify instability events but cannot accurately locate instability nodes, or the location process is not combined with simulation verification of the physical characteristics of the power grid, resulting in a lack of targeted scheduling strategies and difficulty in suppressing instability propagation from the source. In order to reduce this situation, an intelligent identification and scheduling system for power grid transient instability modes is proposed. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent identification and scheduling system for power grid transient instability modes to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, an intelligent identification and scheduling system for power grid transient instability modes is provided, including a data collection module, a data filtering module, a node matching module, a difference analysis module, and a scheduling ranking module;

[0006] The data collection module is used to collect historical and real-time operating data of the power grid, extract transient instability data based on historical operating data, and extract instability severity values ​​from transient instability data;

[0007] The data filtering module is used to set the length of the filtered data based on the instability severity value, and then filter the historical operating data based on the instability severity value of the transient instability data in the historical operating data.

[0008] The node matching module is used to acquire power grid resource data, establish a simulation model based on the power grid resource data, analyze unstable nodes by combining all transient instability data with the simulation model, obtain the corresponding unstable nodes in the simulation model, and match transient instability data with unstable nodes.

[0009] The difference analysis module is used to extract difference values ​​by taking real-time operating data as the center and combining historical operating data corresponding to each transient instability data, and to set risk warning thresholds and instability matching weights for each instability node based on the difference values.

[0010] The scheduling and sorting module is used to retain transient instability data with instability severity values ​​greater than the risk warning threshold in each unstable node, extract instability modes according to instability type, establish an instability mode update library, then randomly reorganize the transient instability data retained in each unstable node to update the instability mode database, and schedule and sort the instability modes in the instability mode database in combination with the corresponding instability severity values ​​and instability matching weights.

[0011] As a further improvement to this technical solution, the data collection module establishes a wireless connection with the power grid management terminal to acquire historical monitoring data of voltage, current, power angle, frequency, power output of new energy power plants, power flow of lines, and equipment operating parameters of each node of the power grid, and summarizes the parameters as historical operating data.

[0012] Consistent with historical operational data, the data collected at the latest time is selected as the real-time operational data;

[0013] Transient instability is identified based on historical operating data. Anomalies are identified in the historical operating data using the electromagnetic identification method of the power system, and the anomalies are selected as transient instability data.

[0014] Perform numerical analysis of instability severity on the selected transient instability data to obtain the instability severity value corresponding to each transient instability data;

[0015] Among them, the numerical analysis of instability severity includes a multi-dimensional quantitative indicator system of instability propagation speed, impact range, and parameter offset magnitude.

[0016] As a further improvement to this technical solution, in the data filtering module, the length of the filtered data is set based on the severity value of instability, and different level intervals are divided according to the magnitude of the severity value of instability. Each level interval corresponds to a specific historical data filtering length, and the filtering data length is longer for the severity value of instability in the higher level interval.

[0017] Among them, the higher the instability severity value, the longer the length of the historical operation data to be screened;

[0018] The lower the instability severity value, the shorter the length of the historical data to be filtered.

[0019] Based on the severity values ​​of transient instability data, historical operational data of corresponding lengths are filtered. Using the occurrence time of the instability event corresponding to each transient instability data point as a node, historical operational data of the corresponding length is extracted by tracing back. Each transient instability data point is matched with a unique historical operational data point of the corresponding length.

[0020] As a further improvement to this technical solution, the node matching module obtains power grid resource data through the power grid management terminal and constructs a power grid digital simulation model consistent with the physical characteristics of the power grid based on the power grid resource data.

[0021] All transient instability data are combined with the simulation model to analyze the instability nodes. The transient instability data is input into the simulation model, and the transient instability process is reconstructed through the simulation model to locate the source node that causes the power grid transient instability, i.e., the instability node.

[0022] Among them, the unstable node is the physical node in the power grid that first shows abnormal parameters and triggers the propagation of instability.

[0023] As a further improvement to this technical solution, the node matching module splits and matches transient instability data with unstable nodes. Based on the analysis results of unstable nodes, each transient instability data is classified under its corresponding unstable node. One unstable node can correspond to one or more transient instability data, thus forming a one-to-one matching relationship between each unstable node and its corresponding transient instability data, and obtaining the transient instability dataset independently corresponding to each unstable node.

[0024] As a further improvement to this technical solution, in the difference analysis module, real-time operating data of the power grid is taken as the center in the simulation model, and the real-time operating data of the power grid is input into the simulation model as the benchmark data of the current power grid operating status.

[0025] Then, a difference numerical analysis was performed by combining the historical operating data corresponding to each transient instability data. The historical operating data matched with the transient instability data under each transient instability node were compared with the benchmark data with parameters of the same dimension. At the same time, the difference numerical value representing the similarity between historical data and real-time data was obtained by normalization.

[0026] As a further improvement to this technical solution, the difference analysis module determines a unique risk warning threshold for each unstable node based on the distribution characteristics of the difference values ​​corresponding to each unstable node, which is used to determine the critical value of transient instability risk.

[0027] A weight calculation model is constructed by combining the difference value and the instability severity value. The rule is that the smaller the difference value and the higher the instability severity value, the greater the instability matching weight. The corresponding instability matching weight is calculated for each transient instability data.

[0028] As a further improvement to this technical solution, in the scheduling and sorting module, in the transient instability dataset of each unstable node, the instability severity value of each transient instability data under that node is compared with the risk warning threshold of that node;

[0029] If the instability severity value exceeds the risk warning threshold, the transient instability data will be retained.

[0030] If the instability severity value is less than the risk warning threshold, the transient instability data will be deleted.

[0031] Based on the classification criteria for transient instability of power grids, the retained transient instability data are classified by type, and representative instability characteristics and evolution patterns are extracted from each category to form corresponding instability modes.

[0032] All extracted instability patterns are integrated and stored to form an instability pattern update library for updating the underlying database.

[0033] As a further improvement to this technical solution, in the scheduling and sorting module, the transient instability data retained under all unstable nodes are randomly combined across nodes and types to generate new transient instability data samples. The newly generated data samples are then input into the instability mode update library to achieve iterative updates of the database.

[0034] A comprehensive ranking model is constructed, which includes instability severity values ​​and instability matching weights. All instability modes in the updated instability mode database are weighted and calculated. The models are then ranked from highest to lowest based on the calculation results. The ranking results provide a priority basis for the defense and scheduling of power grid transient instability modes.

[0035] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0036] 1. In this intelligent identification and dispatching system for power grid transient instability modes, historical data of corresponding lengths is matched by dividing the data into level intervals to ensure that the analysis of instability events can cover the complete evolution process before instability. This provides more targeted historical data support for subsequent instability source tracing and risk analysis. At the same time, a digital simulation model consistent with the physical characteristics of the actual power grid is constructed based on the PSCAD / EMTDC simulation platform. The instability node is accurately located and verified by combining the transient trajectory tracing method with the calculation of instability propagation contribution. This solves the problems of vague instability source location and lack of physical verification in existing technologies, and achieves accurate matching between instability data and instability nodes. This allows subsequent risk analysis and dispatching strategy formulation to focus on the instability source, greatly improving the targeting and effectiveness of dispatching.

[0037] 2. In this intelligent identification and dispatching system for power grid transient instability modes, the electromagnetic transient mechanism of the power system is deeply integrated with intelligent algorithms. In the instability identification stage, a threshold-trend joint identification method based on the electromagnetic transient mechanism is adopted. Combined with multi-dimensional feature extraction, the system achieves accurate screening of transient instability data. At the same time, a multi-dimensional index system of instability propagation speed, impact range, and parameter deviation amplitude is used to achieve quantitative analysis of instability severity. This changes the problem of single index evaluation and low identification accuracy in existing technologies, making the judgment of instability events more in line with the physical operation law of the power grid, and significantly improving the accuracy of transient instability data screening and the scientific nature of instability quantitative assessment.

[0038] 3. In this intelligent identification and dispatching system for power grid transient instability modes, the system achieves numerical analysis of the differences between historical instability data and real-time operating status through comparison and normalization of parameters of the same dimension. At the same time, it sets risk warning thresholds according to the distribution characteristics of the differences in the numerical values ​​of each instability node, and constructs an instability matching weight calculation model by combining the difference values ​​and the instability severity values. This achieves node-level refined quantitative assessment of instability risk, which changes the problems of fixed threshold judgment and lack of dynamism in risk assessment in existing technologies. It allows risk judgment to accurately adapt to the real-time operating status of the power grid, effectively reducing the probability of misjudgment and omission. At the same time, it provides a scientific quantitative basis for subsequent dispatching and ranking, avoiding the subjectivity of dispatching priority determination. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the structure of an intelligent identification and scheduling system for power grid transient instability modes according to the present invention;

[0040] Figure 2 This is a flowchart illustrating the data collection module of the present invention;

[0041] Figure 3 This is a flowchart illustrating the data filtering module of the present invention;

[0042] Figure 4This is a flowchart illustrating the node matching module of the present invention;

[0043] Figure 5 This is a flowchart illustrating the difference analysis module of the present invention;

[0044] Figure 6 This is a flowchart illustrating the scheduling and sorting module of the present invention. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] Please see Figures 1-6 As shown, the purpose of this embodiment is to provide an intelligent identification and scheduling system for power grid transient instability modes, including a data collection module, a data filtering module, a node matching module, a difference analysis module, and a scheduling and sorting module;

[0047] The data collection module is used to collect historical and real-time operating data of the power grid, extract transient instability data based on historical operating data, and extract instability severity values ​​from transient instability data;

[0048] In the data collection module, a wireless connection is established with the power grid management terminal to obtain historical monitoring data of voltage, current, power angle, frequency, power output of new energy power plants, power flow of lines, and equipment operating parameters of each node of the power grid. The parameters are then summarized as historical operating data.

[0049] An encrypted wireless connection is established with the power grid management EMS / DMS system through a dedicated power wireless communication protocol (power IoT 5G private network) to achieve two-way data interaction and permission verification. At the same time, power grid operation monitoring data within a preset time span is retrieved from the power grid management terminal. The extracted dimensions include voltage, current, power angle, frequency of each node in the power grid, output of new energy power plants, power flow of lines, and operating parameters of power equipment. The multi-source data is standardized and aligned according to timestamps and node numbers. After removing duplicate and null data, it is summarized into a structured historical operation dataset.

[0050] Consistent with historical operational data, the data collected at the latest time is selected as the real-time operational data;

[0051] The latest data collected at the power grid management terminal is retrieved at a preset collection frequency (50Hz / 100Hz is preferred for transient analysis). The data is then used as the real-time operation dataset after the data validity is verified.

[0052] Transient instability is identified based on historical operating data. Anomalies are identified in the historical operating data using the electromagnetic identification method of the power system, and the anomalies are selected as transient instability data.

[0053] Electromagnetic transient features are extracted from historical operation datasets. The extracted features are node voltage phasors, line current phasors, power angle change rate, and frequency fluctuation value. A transient feature matrix is ​​constructed. Then, a threshold and trend joint identification method based on electromagnetic transient mechanism is adopted. Combined with the electromagnetic transient stability judgment standard of power system, the steady-state threshold range and trend change threshold of each feature are set.

[0054] If the feature value is not within the steady-state threshold range at a certain moment, it is marked as a preliminary anomaly. At the same time, the feature change rate is calculated for the preliminary anomaly data. When the change rate exceeds the trend change threshold, it is judged as transient instability anomaly data. Then, all the data judged as anomalies are clustered according to instability events. Each instability event corresponds to a continuous time series of anomaly data. The complete data sequence of each instability event is extracted as the transient instability dataset.

[0055] Numerical analysis of the instability severity is performed on the selected transient instability data to obtain the instability severity value corresponding to each transient instability data. For each instability event n in the transient instability dataset, the quantitative values ​​of the three core indicators are calculated. All indicators are normalized to the [0,1] interval to eliminate the influence of dimensions. At the same time, the normalized indicator values ​​of each instability event are weighted and summed with the corresponding weights to obtain the instability severity value of each instability event.

[0056] Among them, the numerical analysis of instability severity includes a multi-dimensional quantitative indicator system of instability propagation speed (0.4), impact range (0.3), and parameter offset amplitude (0.3).

[0057] The data filtering module is used to set the length of the filtered data based on the instability severity value, and then filter the historical running data based on the instability severity value of the transient instability data.

[0058] In the data filtering module, the length of the filtered data is set based on the severity of instability. Different level intervals are divided according to the magnitude of the severity of instability. Each level interval corresponds to a specific historical data filtering length. The higher the severity of instability, the longer the filtering data length.

[0059] Among them, the higher the instability severity value, the longer the length of the historical operation data to be screened;

[0060] The lower the instability severity value, the shorter the length of the historical data to be filtered.

[0061] Based on the instability severity value of the transient instability data, historical operating data of corresponding lengths are filtered. Taking the occurrence time of the instability event corresponding to each transient instability data as a node, the data is traced back and historical operating data of corresponding lengths are extracted. Each transient instability data is matched with a unique historical operating data of corresponding length.

[0062] The node matching module is used to acquire power grid resource data, establish a simulation model based on the power grid resource data, analyze unstable nodes by combining all transient instability data with the simulation model, obtain the corresponding unstable nodes in the simulation model, and match transient instability data with unstable nodes.

[0063] In the node matching module, power grid resource data is obtained through the power grid management terminal, and a digital simulation model of the power grid that is consistent with the physical characteristics of the power grid is constructed based on the power grid resource data;

[0064] Complete power grid resource data is retrieved from the power grid management terminal (EMS / SCADA system), covering four major categories of data: grid structure, equipment parameters, operating constraints, and geographical environment; the parameter units are unified according to the IEEE standard, and topology matrix, equipment parameter library, constraint matrix, and environmental parameter set are generated respectively to complete the data standardization process and eliminate the differences in the formats of multi-source data;

[0065] A three-tier architecture model was built based on the PSCAD / EMTDC simulation platform:

[0066] At the physical layer, standardized network topology and equipment parameters are imported to construct a 1:1 physical model of the nodes and branches of the actual power grid.

[0067] The control layer incorporates control logic for generator excitation / speed regulation, new energy power station PCS, and power grid stability control devices.

[0068] The data interaction layer develops a real-time data interface with the power grid management end, supporting the import of instability data and the export of simulation results;

[0069] All transient instability data are combined with the simulation model to analyze the instability nodes. The transient instability data is input into the simulation model, and the transient instability process is reconstructed through the simulation model to locate the source node that causes the power grid transient instability, i.e., the instability node.

[0070] Among them, the unstable node is the physical node in the power grid that first shows abnormal parameters and triggers the propagation of instability.

[0071] The transient instability dataset is imported into the power grid digital simulation model in batches according to the time series. The simulation step size is set to 10ms. The simulation model is run to restore the complete transient evolution process of each instability event. The dynamic trajectory data of parameters such as voltage, power angle, frequency, and power of all power grid nodes are output. At the same time, based on the power grid operation procedures, the steady-state thresholds of each parameter are set (voltage deviation ≤ ±10% of rated value, power angle deviation ≤ 15°, frequency deviation ≤ ±0.2Hz). The dynamic trajectory of parameters of all nodes is traversed, the time of the first occurrence of parameter abnormality of each node is recorded, and all abnormal nodes are selected to form an abnormal node set.

[0072] Extract the node with the earliest first abnormal time from the abnormal node set to form a candidate node set for the instability source. If multiple nodes have the same first abnormal time (time difference ≤ 10ms), they are all included in the candidate set. Then, calculate the instability propagation contribution of each candidate node to quantify its role in causing instability in the whole network. The higher the contribution, the greater the probability that it is the instability source node.

[0073] The candidate node with the largest contribution to the propagation is selected as the unstable node of the unstable event. If the difference in contribution of multiple candidate nodes is ≤0.1, it is determined to be a multi-source unstable node. At the same time, "fault isolation" is set for the determined unstable node in the simulation model, and the unstable event simulation is rerun. If the unstable phenomenon of the whole network disappears / is significantly alleviated, the node determination is verified to be effective; otherwise, the node is extracted again.

[0074] In the node matching module, transient instability data is split and matched with unstable nodes. Based on the analysis results of unstable nodes, each transient instability data is classified under its corresponding unstable node. One unstable node can correspond to one or more transient instability data, and finally a matching relationship between each unstable node and its corresponding transient instability data is formed, so as to obtain the transient instability dataset that is independently corresponding to each unstable node.

[0075] Using the unstable node as the grouping dimension, the transient instability data of each instability event is aggregated to its corresponding unstable node, following the rule that an unstable event belongs to only one node and a node can correspond to multiple unstable events. At the same time, the aggregated data of each node is sorted from high to low according to the instability severity value.

[0076] The difference analysis module is used to extract difference values ​​by taking real-time operating data as the center and combining historical operating data corresponding to each transient instability data. Based on the difference values, risk warning thresholds and instability matching weights are set for each instability node.

[0077] In the difference analysis module, the simulation model is centered on real-time operating data. The real-time operating data of the power grid is input into the simulation model as the benchmark data for the current operating status of the power grid.

[0078] The real-time operating data is input in batches into the verified power grid digital simulation model. The initial operating state of the model is set to match the real-time data. The simulation model is run to steady state (duration ≥ 5s), and the steady-state parameter set of each node is output as the benchmark data of the current power grid operating state.

[0079] Then, the historical operating data corresponding to each transient instability data is combined to perform difference numerical analysis. The historical operating data matched with the transient instability data under each transient instability node is compared with the benchmark data with the same dimension parameters. At the same time, the difference numerical value representing the similarity between historical data and real-time data is obtained through normalization.

[0080] From each unstable node, extract the historical data matching each unstable event, align the historical data of each unstable event with the benchmark data according to the dimensions of nodes and parameters, calculate the parameter difference under each dimension, and eliminate time series differences (take the mean of the steady state period of historical data and compare it with the benchmark data).

[0081] The differences in parameters across all dimensions are normalized using a min-max method to eliminate the influence of dimensions. Then, a weighted average is used to calculate the comprehensive difference value of a single instability event, quantifying the similarity between historical instability scenarios and the current power grid state. Finally, the difference values ​​of all instability events under the same instability node are summarized to form a set of difference values ​​for each instability node, providing a data basis for subsequent threshold setting. The formula is as follows:

[0082] ;

[0083] in, Let be the absolute difference between the nth instability event, the i-th node, and the p-th parameter. Let be the mean value of the steady-state segment of the i-th node and the p-th parameter in the historical data corresponding to the n-th instability event. Let be the steady-state value of the i-th node and the p-th parameter in the baseline data;

[0084] ;

[0085] in, This represents the normalized one-dimensional difference value, ranging from [0,1]. The minimum value of the p-th parameter difference among all instability events. The maximum value of the difference of the p-th parameter in the instability event;

[0086] ;

[0087] in, This represents the overall difference value for the nth instability event, ranging from [0,1]. A smaller value indicates a greater similarity between historical and real-time data. The weights for the p-th parameter are: voltage 0.3, power angle 0.25, frequency 0.2, new energy output 0.15, and line power flow 0.1. This refers to the unstable node corresponding to the unstable event;

[0088] In the difference analysis module, based on the distribution characteristics of the difference values ​​corresponding to each unstable node, a unique risk warning threshold is determined for each unstable node, which is used to determine the critical value for transient instability risk. Nodes are divided into normally distributed nodes and skewed distributed nodes. The formula for normally distributed nodes is as follows:

[0089] ;

[0090] in, Let i be the initial threshold value for the unstable node. Let be the mean of the differences at node i. Let be the standard deviation of the difference values ​​at node i;

[0091] The formula for skewed distribution nodes is as follows:

[0092] ;

[0093] in, Let be the median of the differences at node i. The interquartile range of the difference values ​​at node i;

[0094] ;

[0095] in, The final risk warning threshold for node i. The adjustment factor is (-1 / -2);

[0096] A weight calculation model is constructed by combining the difference value and the instability severity value. Following the rule that the smaller the difference value and the higher the instability severity value, the greater the instability matching weight, the corresponding instability matching weight is calculated for each transient instability data point. The steps are as follows:

[0097] The difference value and the instability severity value are normalized to the [0,1] interval respectively (the difference value is reverse normalized, and the smaller the value, the higher the weight), and the weight coefficients of the difference value and the instability severity value are determined (the difference value weight = 0.6, the instability severity value weight = 0.4). Substitute them into the weight model to calculate the instability matching weight of each instability event. The larger the weight value, the higher the risk of the instability scenario to the current power grid.

[0098] If the weight value exceeds the range [0,1], it will be truncated (<0 is taken as 0, >1 is taken as 1).

[0099] For events with abnormal weight values ​​at the same node (deviation from the mean by 3 times the standard deviation), recheck the difference values / severity values ​​and then make corrections;

[0100] The weights of all unstable events under each unstable node are aggregated to form a sub-node weight set.

[0101] The scheduling and sorting module is used to retain transient instability data with instability severity values ​​greater than the risk warning threshold in each unstable node, extract instability modes according to instability type, establish an instability mode update library, then randomly reorganize the transient instability data retained in each unstable node to update the instability mode database, and sort the instability modes in the instability mode database by combining the corresponding instability severity values ​​and instability matching weights.

[0102] In the scheduling and sorting module, in the transient instability dataset of each unstable node, the instability severity value of each transient instability data under that node is compared with the risk warning threshold of that node;

[0103] If the instability severity value exceeds the risk warning threshold, the transient instability data will be retained.

[0104] If the instability severity value is less than the risk warning threshold, the transient instability data will be deleted.

[0105] Based on the classification criteria for transient instability of power grids, the retained transient instability data are classified by type, and representative instability characteristics and evolution patterns are extracted from each category to form corresponding instability modes.

[0106] According to national standards, high-risk data are classified into categories based on the types of power grid transient instability. These categories include voltage instability, power angle instability, frequency instability, and instability caused by new energy disconnection from the grid. For each type of instability data, representative features (such as voltage drop amplitude, power angle change rate, and frequency fluctuation period) and evolution patterns (such as instability development time, propagation path, and key inflection points) are extracted. The extracted features and patterns are then encapsulated into structured instability patterns. All types of instability patterns are integrated to generate an updated instability pattern library.

[0107] All extracted instability patterns are integrated and stored to form an instability pattern update library for updating the base database.

[0108] In the scheduling and sorting module, the transient instability data retained under all unstable nodes are randomly combined across nodes and types to generate new transient instability data samples. The newly generated data samples are then input into the instability mode update library to achieve iterative updates of the database.

[0109] Aggregate high-risk data retained across all nodes, construct a cross-node, cross-type reorganized data pool, and set reorganization parameters:

[0110] Combination dimensions, instability characteristics (such as voltage drop + power angle change), and node attributes (such as new energy nodes + load nodes).

[0111] The combination ratio is as follows: select 2 to 3 types of unstable data in a single combination, and select 1 to 2 features for each type;

[0112] The random seed is set based on the historical fault probability distribution of the power grid to ensure the engineering rationality of the recombined samples;

[0113] Data is randomly extracted from the data pool according to the settings, merged to generate new transient instability data samples, supplemented with metadata such as feature labels and risk levels, and then the new samples are imported into the instability mode update library to update the statistical characteristics of the data in the library (such as the number of various instability events and the mean of features), thus completing the iteration of the basic database.

[0114] A comprehensive ranking model is constructed, which includes the instability severity value and instability matching weight. All instability modes in the updated instability mode database are weighted and calculated. The models are then ranked from high to low according to the calculation results. The ranking results provide priority basis for the defense and scheduling of power grid transient instability modes.

[0115] Retrieve the updated instability mode database, extract the instability severity value and instability matching weight corresponding to each instability mode, determine the weight coefficient (instability severity weight = 0.4, matching weight = 0.6), and then calculate the comprehensive ranking value for each instability mode to quantify its defense scheduling priority.

[0116] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A smart identification and dispatching system for power grid transient instability modes, characterized in that: It includes a data collection module, a data filtering module, a node matching module, a difference analysis module, and a scheduling and sorting module; The data collection module is used to collect historical and real-time operating data of the power grid, extract transient instability data based on historical operating data, and extract instability severity values ​​from transient instability data; The data filtering module is used to set the length of the filtered data based on the instability severity value, and then filter the historical operating data based on the instability severity value of the transient instability data in the historical operating data. The node matching module is used to acquire power grid resource data, establish a simulation model based on the power grid resource data, analyze unstable nodes by combining all transient instability data with the simulation model, obtain the corresponding unstable nodes in the simulation model, and match transient instability data with unstable nodes. The difference analysis module is used to extract difference values ​​by taking real-time operating data as the center and combining historical operating data corresponding to each transient instability data, and to set risk warning thresholds and instability matching weights for each instability node based on the difference values. The scheduling and sorting module is used to retain transient instability data with instability severity values ​​greater than the risk warning threshold in each unstable node, extract instability modes according to instability type, establish an instability mode update library, then randomly reorganize the transient instability data retained in each unstable node to update the instability mode database, and schedule and sort the instability modes in the instability mode database in combination with the corresponding instability severity values ​​and instability matching weights.

2. The intelligent identification and dispatching system for power grid transient instability modes according to claim 1, characterized in that: In the data collection module, a wireless connection is established with the power grid management terminal to obtain historical monitoring data of voltage, current, power angle, frequency, power output of new energy power plants, power flow of lines, and equipment operating parameters of each node of the power grid, and the parameters are summarized as historical operating data. Consistent with historical operational data, the data collected at the latest time is selected as the real-time operational data; Transient instability is identified based on historical operating data. Anomalies are identified in the historical operating data using the electromagnetic identification method of the power system, and the anomalies are selected as transient instability data. Perform numerical analysis of instability severity on the selected transient instability data to obtain the instability severity value corresponding to each transient instability data; Among them, the numerical analysis of instability severity includes a multi-dimensional quantitative indicator system of instability propagation speed, impact range, and parameter offset magnitude.

3. The intelligent identification and dispatching system for power grid transient instability modes according to claim 1, characterized in that: In the data filtering module, the length of the filtered data is set based on the severity value of instability. Different level intervals are divided according to the magnitude of the severity value of instability. Each level interval corresponds to a specific historical data filtering length. The higher the severity value of instability is, the longer the filtering data length is. Among them, the higher the instability severity value, the longer the length of the historical operation data to be screened; The lower the instability severity value, the shorter the length of the historical data to be filtered. Based on the instability severity value of the transient instability data, historical operating data of corresponding lengths are filtered. Taking the occurrence time of the instability event corresponding to each transient instability data as a node, the data is traced back and historical operating data of corresponding lengths are extracted. Each transient instability data is matched with a unique historical operating data of corresponding length.

4. The intelligent identification and dispatching system for power grid transient instability modes according to claim 1, characterized in that: In the node matching module, power grid resource data is obtained through the power grid management terminal, and a power grid digital simulation model consistent with the physical characteristics of the power grid is constructed based on the power grid resource data. All transient instability data are combined with the simulation model to analyze the instability nodes. The transient instability data is input into the simulation model, and the transient instability process is reconstructed through the simulation model to locate the source node that causes the power grid transient instability, i.e., the instability node. Among them, the unstable node is the physical node in the power grid that first shows abnormal parameters and triggers the propagation of instability.

5. The intelligent identification and dispatching system for power grid transient instability modes according to claim 1, characterized in that: In the node matching module, transient instability data is split and matched with unstable nodes. Based on the analysis results of unstable nodes, each transient instability data is classified under its corresponding unstable node. One unstable node can correspond to one or more transient instability data, and finally a matching relationship between each unstable node and its corresponding transient instability data is formed to obtain the transient instability dataset that is independently corresponding to each unstable node.

6. The intelligent identification and dispatching system for power grid transient instability modes according to claim 1, characterized in that: In the difference analysis module, the simulation model is centered on real-time operating data, and the real-time operating data of the power grid is input into the simulation model as the benchmark data for the current operating status of the power grid. Then, a difference numerical analysis was performed by combining the historical operating data corresponding to each transient instability data. The historical operating data matched with the transient instability data under each transient instability node were compared with the benchmark data with parameters of the same dimension. At the same time, the difference numerical value representing the similarity between historical data and real-time data was obtained by normalization.

7. The intelligent identification and dispatching system for power grid transient instability modes according to claim 1, characterized in that: In the difference analysis module, based on the distribution characteristics of the difference values ​​corresponding to each unstable node, a unique risk warning threshold is determined for each unstable node, which is used to determine the critical value of transient instability risk. A weight calculation model is constructed by combining the difference value and the instability severity value. The rule is that the smaller the difference value and the higher the instability severity value, the greater the instability matching weight. The corresponding instability matching weight is calculated for each transient instability data.

8. The intelligent identification and dispatching system for power grid transient instability modes according to claim 1, characterized in that: In the scheduling and sorting module, in the transient instability dataset of each unstable node, the instability severity value of each transient instability data under that node is compared with the risk warning threshold of that node; If the instability severity value exceeds the risk warning threshold, the transient instability data will be retained. If the instability severity value is less than the risk warning threshold, the transient instability data will be deleted. Based on the classification criteria for transient instability of power grids, the retained transient instability data are classified by type, and representative instability characteristics and evolution patterns are extracted from each category to form corresponding instability modes. All extracted instability patterns are integrated and stored to form an instability pattern update library for updating the base database.

9. The intelligent identification and dispatching system for power grid transient instability modes according to claim 1, characterized in that: In the scheduling and sorting module, the transient instability data retained under all unstable nodes are randomly combined across nodes and types to generate new transient instability data samples. The newly generated data samples are then input into the instability mode update library to achieve iterative updates of the database. A comprehensive ranking model is constructed, which includes instability severity values ​​and instability matching weights. All instability modes in the updated instability mode database are weighted and calculated. The models are then ranked from highest to lowest based on the calculation results. The ranking results provide a priority basis for the defense and scheduling of power grid transient instability modes.

Images