A method for overvoltage prediction of a power system

Abstract

The present application relates to the technical field of power system management, in particular to a power system overvoltage prediction method, comprising the following steps: collecting voltage time sequence, current component, power flow rate through power system monitoring nodes, extracting ripple amplitude and multi-dimensional data, differentiating ripple and rate change, marking mutation nodes, obtaining associated components along topology, sliding window predicting disturbance path, analyzing virtual power flow direction and change, clustering and screening virtual power kinks, establishing risk node library, sorting node voltage parameters, and generating overvoltage prediction results. In the present application, voltage ripple, zero sequence and capacitor current phase, dynamic power flow rate are collected through multi-dimensional monitoring data fusion, composite threshold is used to accurately identify disturbance nodes, sliding prediction is used to track energy path, dynamic clustering is used to screen risk characteristic nodes, overvoltage trend is reflected in real time, risk screening integrity and resolution are improved, and targeted support is provided for overvoltage hidden danger monitoring.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system management, in particular to a power system overvoltage prediction method. BACKGROUND

[0002] The technical field of power system management involves monitoring, analyzing, controlling and optimizing management of power system operation state, and covers core matters such as power dispatching, load forecasting, power equipment operation state monitoring, power grid safety analysis and evaluation. The stability and reliability of the power system are crucial to social production and life, and with the growth of power demand and the aging of equipment, its safety and stability are facing challenges, and the related technical research in this field is of great significance.

[0003] Among them, the traditional power system overvoltage prediction method refers to predicting the occurrence of overvoltage by analyzing the voltage variation law of the power system, combining with the monitoring data, using statistical analysis based on raw data, simplified power system dynamic model and empirical formula and other ways. This method relies on known voltage fluctuation law and uses simple numerical calculation, and it is difficult to fully consider the dynamic action of multiple factors in the system, and the prediction accuracy is insufficient in complex or sudden situations.

[0004] The traditional method based on traditional prediction method of original statistics and single index analysis is difficult to capture the complex coupling change of power system signal, and in the face of node multi-element disturbance or dynamic change, the monitoring response is easy to be distorted, the abnormal node cannot be effectively identified in real-time operation, and the synchronous analysis of ripple and energy propagation process is lacking, which leads to the overvoltage hidden danger being easily missed in the multi-factor superposition scene, and once the sudden current direction change or local anomaly occurs, the overall safety and emergency decision-making ability of the system will be seriously affected, affecting the safety monitoring effect of the power network. SUMMARY

[0005] In order to solve the technical problem that the traditional method based on traditional prediction method of original statistics and single index analysis is difficult to capture the complex coupling change of power system signal, and in the face of node multi-element disturbance or dynamic change, the monitoring response is easy to be distorted, the abnormal node cannot be effectively identified in real-time operation, and the synchronous analysis of ripple and energy propagation process is lacking, which leads to the overvoltage hidden danger being easily missed in the multi-factor superposition scene, and once the sudden current direction change or local anomaly occurs, the overall safety and emergency decision-making ability of the system will be seriously affected, affecting the safety monitoring effect of the power network, the present application provides a power system overvoltage prediction method, comprising the following steps:

[0006] In order to achieve the above purpose, the present application adopts the following technical scheme: a power system overvoltage prediction method, comprising the following steps:

[0007] S1: Obtain voltage time sequence signals, current components, bus power flow rates through power system monitoring nodes, and perform ripple amplitude extraction processing on the voltage time sequence signals, simultaneously collect zero sequence current components, capacitor current phase angles, power transmission change rates, and form a multi-dimensional monitoring data set;

[0008] S2: Call the multi-dimensional monitoring data set for ripple amplitude difference, calculate the ripple amplitude change and power flow rate change between adjacent time instants, compare the ripple mutation threshold and the rate change threshold, mark the double-threshold out-of-limit nodes, and form a mutation node identification sequence;

[0009] S3: Call the nodes in the mutation node identification sequence, obtain the adjacent node zero sequence current components and capacitor current phase angles along the topological forward and reverse progression, analyze the disturbance amplitude trend using a sliding window prediction algorithm, identify the main energy propagation path and the secondary branch path, and output a disturbance propagation path set;

[0010] S4: Obtain the main path bus virtual power distribution and virtual power flow direction values of the disturbance propagation path set, calculate the flow direction change and flow rate change, identify the flow direction reverse and amplitude out-of-limit nodes, filter the virtual power kink structure using a dynamic clustering analysis algorithm, and establish an overvoltage risk node library.

[0011] As a further scheme of the application, the multi-dimensional monitoring data set includes voltage effective value, current component and bus power flow rate, the mutation node identification sequence includes node number, waveform mutation type and topological position information, the disturbance propagation path set includes main path serial number, zero sequence current trend and energy transmission direction, and the overvoltage risk node library includes virtual power abnormal node, risk level and node distribution characteristics.

[0012] As a further scheme of the application, the specific steps of S1 are:

[0013] S101: Obtain voltage time sequence signals, current components, bus power flow rates through power system monitoring nodes, divide the voltage signals by period and calculate the fluctuation difference value, extract the amplitude mutation points of adjacent periods, combine the local peak values to construct the amplitude interval, and generate the voltage fluctuation amplitude interval;

[0014] S102: Call the voltage fluctuation amplitude interval, collect zero sequence current components and capacitor current phase angles, analyze the amplitude change trend of the section, compare the zero sequence current frequency change of the corresponding period, calculate the correlation and superimpose the capacitor current phase difference value, and generate a phase coupling change coefficient;

[0015] The phase coupling change coefficient is a comprehensive quantitative index reflecting the correlation between the voltage fluctuation amplitude interval and the zero sequence current component and the capacitor current phase angle at the same sampling time;

[0016] S103: Based on the phase coupling change coefficient, call the bus power flow rate data and process its rate differential, analyze the phase coupling and power change trend in a unified sampling time axis, and generate a multi-dimensional monitoring data set.

[0017] As a further scheme of the present application, the specific steps of S2 are:

[0018] S201: Call the voltage waveform value and power flow value in the multi-dimensional monitoring data set, extract the voltage ripple amplitude of adjacent time, calculate the adjacent difference value to obtain the change value, record the power flow rate change value corresponding to the time period, group corresponding according to time period, and generate a differential change control value sequence;

[0019] S202: Based on the number in the differential change control value sequence, call the corresponding ripple amplitude change value and power change value, compare the ripple mutation threshold and flow rate change threshold respectively, filter the synchronous over-limit number, remove the records that do not meet the conditions, and obtain a double-threshold over-limit number set;

[0020] The ripple mutation threshold and flow rate change threshold are set according to the power quality standard and the actual operation experience of the power distribution system. The voltage mutation threshold is set to 1.3% of the rated voltage, and the power change threshold is set to 2.7% of the rated power, to ensure a reasonable balance between detection sensitivity and false alarm rate;

[0021] S203: Call the double-threshold over-limit number set to mark the number, index the node number and time node in the original data, combine the node information that meets the conditions in order, associate the node number and time node, and generate a mutation node identification sequence.

[0022] As a further scheme of the present application, the specific steps of S3 are:

[0023] S301: Call the node in the mutation node identification sequence, access the topology connection relationship in the forward and reverse directions of the topology structure in turn, obtain the zero sequence current component and capacitance current phase angle of its adjacent nodes, and obtain the adjacent node current phase parameter group according to the node level combination data;

[0024] S302: Based on the current component sequence and capacitance current phase angle sequence in the adjacent node current phase parameter group, identify the current amplitude change trend in the sliding window, combine the phase angle to deduce the energy gradient, calculate the trend change in the window, and obtain the disturbance energy gradient trend coefficient;

[0025] S303: According to the disturbance energy gradient trend coefficient, judge the energy propagation direction between nodes in the topology path, mark the gradient continuously rising path as the main energy propagation path, and the fluctuation interval path as the secondary path, and output the disturbance propagation path set.

[0026] As a further scheme of the present application, the specific steps of S4 are:

[0027] S401: Obtain the virtual power distribution amount on the main path bus in the set of disturbance propagation paths, calculate the virtual power difference value of the bus node within the path based on the node number and virtual power in-out value of each path node, and combine the bus number information to summarize the node virtual power difference value and generate the path node virtual power distribution value;

[0028] S402: According to the path node virtual power distribution value, calculate the virtual power flow value before and after the disturbance between nodes, judge whether the flow direction is reversed under the same node difference time sequence, compare whether the flow velocity variation amplitude exceeds the virtual power flow velocity amplitude threshold value, and obtain the virtual power flow direction change characteristic quantity;

[0029] S403: Call the virtual power flow direction change characteristic quantity, use a dynamic clustering analysis algorithm to determine the characteristic value distance between nodes, filter nodes that meet the aggregation threshold condition, and merge into a virtual power kink structure to establish an overvoltage risk node library.

[0030] As a further scheme of the present application, the path node virtual power distribution value is a set of bus node virtual power difference values calculated and summarized along the nodes in the disturbance propagation path according to the node number and corresponding virtual power in-out value;

[0031] The virtual power flow direction change characteristic quantity is a difference index calculated based on the path node virtual power distribution value, the virtual power flow direction between nodes before and after the disturbance, and the flow velocity variation amplitude;

[0032] The aggregation threshold condition is a numerical boundary parameter used to determine whether the distance between the virtual power flow direction change characteristic quantities of the nodes meets the merging requirement in the dynamic clustering analysis;

[0033] The virtual power kink structure refers to a node set formed by merging based on the reverse change of the virtual power flow direction between nodes and the flow velocity amplitude threshold condition on the disturbance propagation path through dynamic clustering analysis.

[0034] As a further scheme of the present application, the method further comprises the S5 step:

[0035] S5: Sort the node voltage mutation parameters in the overvoltage risk node library in descending order, extract the first node voltage amplitude in the order, calculate the deviation between the node voltage amplitude and the rated voltage, and when the deviation exceeds the overvoltage threshold value, generate an overvoltage prediction result;

[0036] The overvoltage prediction result includes the node identification number, the voltage deviation, and the warning time stamp.

[0037] As a further scheme of the present application, the specific steps of S5 are:

[0038] S501: Obtain the voltage mutation parameter of the node in the overvoltage risk node library, combine the node voltage time sequence data, calculate the node voltage transient mutation amplitude, sort the first node voltage amplitude in descending order, and obtain the node voltage amplitude;

[0039] S502: Based on the node voltage amplitude, the rated voltage value of the corresponding node is extracted, the difference between the node voltage amplitude and the rated voltage value is calculated, the amplitude deviation degree is analyzed through the direct numerical difference between the two, and the deviation amount of the node voltage amplitude and the rated voltage is obtained;

[0040] S503: The deviation amount of the node voltage amplitude and the rated voltage is called, and the numerical comparison with the overvoltage threshold value is carried out, whether the deviation amount exceeds the threshold boundary range is judged, if the deviation amount is greater than the threshold range, the overvoltage prediction result is generated.

[0041] As a further scheme of the application, the node voltage amplitude is the voltage change amount of the node voltage when the transient mutation occurs, which represents the mutation characteristics of the node voltage in a specific time sequence;

[0042] The deviation amount of the node voltage amplitude and the rated voltage is the absolute difference between the node voltage amplitude and the node rated voltage value;

[0043] The overvoltage threshold value is a preset voltage numerical limit value, which is compared with the deviation amount of the node voltage amplitude and the rated voltage to determine whether there is an overvoltage situation.

[0044] Compared with the prior art, the application has the following advantages and positive effects:

[0045] In the application, multi-dimensional monitoring data fusion acquisition is carried out, including voltage ripple, zero sequence current, capacitor current phase and dynamic power flow rate, the running disturbance characteristics are finely presented, the disturbance node is accurately identified through a composite threshold discrimination mechanism, the primary and secondary energy propagation paths of the disturbance are tracked by a sliding prediction algorithm, and the node set with risk characteristics is screened out by combining the virtual power flow state dynamic clustering, the potential overvoltage distribution trend in the system is reflected in time, the positioning problem of hidden disturbance and flow abnormality is effectively solved, and the integrity, real-time performance and resolution of the power system risk node screening are greatly improved, thereby providing more targeted auxiliary data support for overvoltage hidden danger monitoring and disposal. BRIEF DESCRIPTION OF DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0047] Figure 1 A flow chart of the steps of the present application;

[0048] Figure 2 A detailed flow chart of S1 of the present application;

[0049] Figure 3 A detailed flow chart of S2 of the present application;

[0050] Figure 4 A detailed flow chart of S3 of the present application;

[0051] Figure 5 A detailed flow chart of S4 of the present application;

[0052] Figure 6 A detailed flow chart of S5 of the present application. DETAILED DESCRIPTION

[0053] The technical solutions in the present application will be described below with reference to the drawings.

[0054] In the embodiments of the present application, the words such as "exemplary", "for example", etc. are used to represent an example, illustration or description. Any embodiment or design scheme described as "exemplary" in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the word "exemplary" is intended to present the concept in a specific manner. In addition, in the embodiments of the present application, the meaning expressed by "and / or" can be both, or can be one of the two optionally.

[0055] In the embodiments of the present application, "image" and "picture" can be used interchangeably at times. It should be pointed out that when the distinction is not emphasized, the meanings expressed are consistent. "Of", "corresponding" and "corresponding" can be used interchangeably at times. It should be pointed out that when the distinction is not emphasized, the meanings expressed are consistent.

[0056] In the embodiments of the present application, sometimes the subscript such as W1 can be written in the form of non-subscript such as W1. When the distinction is not emphasized, the meanings expressed are consistent.

[0057] In order to make the technical problems, technical solutions and advantages of the present application clearer, specific embodiments will be described in detail below with reference to the drawings.

[0058] Please refer to Figure 1 The embodiments of the present application provide a power system overvoltage prediction method, which comprises the following steps:

[0059] S1: Obtain voltage time sequence signal, current component, bus power flow rate through power system monitoring node, and perform ripple amplitude extraction processing on voltage time sequence signal, simultaneously collect zero sequence current component, capacitance current phase angle, power transmission change rate, and form multi-dimensional monitoring data set;

[0060] S2: Call multi-dimensional monitoring data set for ripple amplitude difference, calculate ripple amplitude change quantity and power flow rate change quantity between adjacent time, compare ripple mutation threshold and rate change threshold, mark double threshold overrun node, and form mutation node identification sequence;

[0061] S3: Call nodes in mutation node identification sequence, obtain adjacent node zero sequence current component and capacitance current phase angle along topological forward and reverse progression, analyze disturbance amplitude increase and decrease trend using sliding window prediction algorithm, identify main energy propagation path and secondary branch path, and output disturbance propagation path set;

[0062] S4: Obtain main path bus virtual power distribution amount and virtual power flow direction value of disturbance propagation path set, calculate flow direction change quantity and flow rate change quantity, identify flow direction reverse and amplitude overrun node, screen virtual power kink structure using dynamic clustering analysis algorithm, and establish overvoltage risk node library;

[0063] S5: Call node voltage mutation parameters in overvoltage risk node library for descending order sorting, extract first node voltage amplitude, calculate node voltage amplitude and rated voltage deviation, and generate overvoltage prediction result when deviation exceeds overvoltage threshold;

[0064] The multi-dimensional monitoring data set includes voltage effective value, current component and bus power flow rate, the mutation node identification sequence includes node number, waveform mutation type and topological position information, the disturbance propagation path set includes main path serial number, zero sequence current trend and energy transmission direction, the overvoltage risk node library includes virtual power abnormal node, risk level and node distribution characteristics, and the overvoltage prediction result includes node identification number, voltage deviation and early warning time mark.

[0065] Please refer to Figure 2 , the specific steps of S1 are as follows:

[0066] S101: Obtain voltage time sequence signal, current component, bus power flow rate through power system monitoring node, divide voltage signal according to period and calculate fluctuation difference, extract adjacent period amplitude mutation point, combine local peak value to construct amplitude interval, and generate voltage fluctuation amplitude interval;

[0067] The voltage time sequence signal obtained by the power system monitoring node should be specifically calibrated as the bus voltage value sequence collected continuously at 10ms sampling interval within the set period. For a complete 50Hz AC cycle, 20 voltage sampling value sequences can be obtained within 20ms, and the sequence is divided into continuous cycle groups, for example, from 0ms to 20ms as the first cycle, from 20ms to 40ms as the second cycle, and so on. The maximum value and the minimum value in each cycle group after cycle division are calculated to obtain the voltage fluctuation difference ΔU. The difference calculation action is: for the maximum value of the first cycle , the minimum value , the fluctuation difference is obtained. This step is repeated in adjacent cycles and the fluctuation difference of each cycle is recorded. After the construction of the basic fluctuation sequence, it is necessary to judge the amplitude mutation point between cycles, that is, the fluctuation difference of each two adjacent cycles is compared. If , , the difference is . If the difference exceeds the set mutation threshold, it is determined that the point is an amplitude mutation point. The threshold is set to refer to the standard deviation σ of the average fluctuation level of the system, which is set to . When the actual change exceeds the threshold, it is recorded as a mutation point. For example, if σ is 1.7V after 100 cycles, the mutation threshold is set to . When the mutation difference reaches 2.5V, it is confirmed as an amplitude mutation point. Then the local maximum and minimum values of the fluctuation amplitude near the mutation point are extracted, and the peak point voltage amplitude is recorded. Assuming that in a mutation cycle, the maximum amplitude is , and the minimum amplitude is , the amplitude interval of the mutation cycle is . A plurality of amplitude intervals are generated in this way, and are mapped and matched according to the time axis to obtain the voltage fluctuation amplitude interval data array. The data structure is as follows.

[0068] Table 1: Voltage fluctuation amplitude interval table

[0069]

[0070] As shown in Table 1, the amplitude mutation point occurs in period 2 and 3, and the fluctuation difference exceeds the mutation threshold value 2.0V, respectively. The final voltage fluctuation amplitude interval is constructed by the minimum value and the maximum value, which provides basic input data for subsequent phase correlation analysis and power flow trend. The overall fluctuation amplitude extraction, mutation determination and interval construction links call four specific operation processes respectively, namely time series sampling, voltage maximum and minimum operation, mutation point detection and interval matching. Combined with the actual operation sequence of the sample data, the operation sequence is as follows: sampling point sequence construction, voltage maximum and minimum value search, period fluctuation difference calculation, difference mutation value comparison and comparison with mutation threshold value, finally determine the mutation point and extract its upper and lower limit value to construct the voltage fluctuation amplitude interval.

[0071] S102: call the voltage fluctuation amplitude interval, collect the zero sequence current component and the capacitance current phase angle, analyze the amplitude change trend of the section, compare the frequency change of the zero sequence current in the corresponding period, calculate the correlation and superimpose the phase difference value of the capacitance current, and generate the phase coupling change coefficient;

[0072] After calling the voltage fluctuation amplitude interval, first, the zero sequence current component data in the corresponding interval is obtained in the order of cycle number. The collection method is to measure the zero sequence current instantaneous value at 10ms intervals in each cycle. The sampling data of period 2 is , and the sampling data of period 3 is . The average value of the zero sequence current component of each cycle is calculated to obtain the average zero sequence current of period 2 , and the average zero sequence current of period 3 . Then, according to the capacitance branch current component, the capacitance current is obtained by instantaneous voltage and capacitance value calculation. Assuming that the system capacitance is , the voltage change rate is , then the capacitance current is:

[0073] ;

[0074] The phase difference between the zero sequence current and the capacitance current at the corresponding sampling time is calculated. Assuming that the phase of the zero sequence current is , the phase of the capacitance current is , and the phase difference is . The phase difference value is analyzed in association with the amplitude interval in the previous period. The analysis process is to correspond the phase difference value 90° to the voltage fluctuation amplitude interval . The phase difference is quantized to the relative amplitude change rate of the interval . The frequency change of the zero sequence current in the adjacent cycle is calculated, taking the time difference 10ms between the sampling points of period 2 and period 3 as the reference. The frequency change rate is:

[0075] ;

[0076] The frequency change value and the phase difference value are superimposed, and the superimposition method is weighted superimposition. The weight is set to 0.7 of the zero sequence current frequency change proportion and 0.3 of the capacitor current phase difference value according to the contribution degree of the capacitance branch current, to obtain a phase coupling change coefficient , the amplitude interval is extracted, the zero sequence current component and the capacitor current phase difference are obtained, the average value, the frequency change value and the phase difference value are calculated, and the phase coupling change coefficient is obtained by superimposing the weighted proportion. The coefficient will be used as a core correlation factor for subsequent bus power flow trend analysis.

[0077] S103: Based on the phase coupling change coefficient, call the bus power flow rate data and perform rate differentiation processing, analyze the phase coupling and power change trend synchronously according to the unified sampling time axis, and generate a multi-dimensional monitoring data set;

[0078] After obtaining the phase coupling change coefficient, the bus power flow rate data is called, and the process of collecting the bus power flow rate is to measure the instantaneous value of the bus active power and the reactive power at 10 ms intervals in the corresponding period. The bus power sampling value sequence of period 3 is , When calculating the power flow rate, the active power rate and the reactive power rate are divided by the sampling interval respectively, and the following is obtained:

[0079] , ;

[0080] Synchronous analysis is performed on the rate data and the phase coupling change coefficient . The analysis steps are as follows: align the rate data with the phase coupling change coefficient sequence according to the time axis, multiply the active power flow rate and the phase coupling change coefficient respectively, and set the synchronous data pair as , , respectively. Record in the multi-dimensional monitoring data set. The set structure records the active power flow rate, the reactive power flow rate, the phase coupling change coefficient, and the synchronous product result according to the time sequence dimension, and saves the multi-dimensional data items of the period in array form. In the actual execution process, the period corresponding data needs to be differentiated, matched, multiplied and archived according to the above steps according to the time sequence to generate a complete multi-dimensional monitoring data set. The set will be used as the core input data for subsequent overvoltage prediction and trend analysis.

[0081] Please refer to Figure 3 , the specific steps of S2 are as follows:

[0082] ​S201: Call the voltage waveform value and power flow rate value in the multi-dimensional monitoring data set, extract the adjacent time voltage ripple amplitude, calculate the adjacent difference value to obtain the change value, record the corresponding period power flow rate change value, group according to the time period, and generate the differential change control value sequence;

[0083] The voltage waveform array is established based on 10ms time resolution:

[0084] ;

[0085] The active power array:

[0086] ;

[0087] For voltage waveform data, the voltage difference between adjacent sampling points is calculated in sequence to obtain the ripple change value sequence , and the power change value sequence is extracted in sequence , each pair is grouped and the differential change control value sequence is generated, such as , in order to ensure the logicality of subsequent grouping processing, the starting time number of each group of data needs to be identified according to the time axis, such as , and recorded in the structured sequence, the calculation process of voltage difference is to perform subtraction operation on each pair of adjacent voltage values in the array, that is , the calculation method of power difference is the same, that is , the two sequences correspond one by one and establish an index array for time dimension , forming a three-tuple structure , such as , in the actual system, if the sampling frequency is 100Hz, 100 groups of data can be obtained per second, then 2 groups of data can be obtained per cycle (assuming 20ms), such as the voltage value in cycle 1 is [230.2V, 232.7V], the ripple change value is 2.5V, and the power difference value is 10kW, the complete data grouping structure is shown below.

[0088] Table 2: Monitoring cycle differential change data table

[0089]

[0090] As shown in Table 2, the differential change values listed in the table are calculated by adjacent time period sampling data, and have a one-to-one corresponding relationship with the cycle number.

[0091] S202: Based on the number in the differential change reference value sequence, call the corresponding ripple amplitude change value and power change value, compare the ripple sudden change threshold and flow rate change threshold respectively, filter out the synchronous over-limit numbers, remove the records that do not meet the conditions, and obtain the set of dual threshold over-limit numbers.

[0092] Based on the numbering information in the differential change control value sequence, the voltage ripple change value and power change value of each group are extracted one by one. The set ripple abrupt change threshold and power change threshold are then used for judgment. The ripple abrupt change threshold is set as follows: The power change threshold is set to , each set of data and If the ripple is greater than or equal to the threshold, it is determined to be out of limit; otherwise... If the power change is greater than or equal to the power change threshold, it is determined to be an excessive power change. Data numbers that meet the "double exceedance" condition are selected. The specific process is as follows: In Group 1 , The conditions are not met; in group 2 , The condition satisfies the double overlimit requirement; in group 3 , It also satisfies the double exceedance condition, thus the set of time numbers that satisfy the synchronization exceedance condition is obtained by filtering. The action is determined by a conditional expression based on absolute value comparison. and Then the time number will be... The results were added to the set, and the dual threshold settings were used to reference the allowable fluctuation range of the system voltage standard and the allowable adjustment rate of the power control system. The voltage mutation threshold of 3.0V corresponds to a 1.3% rated voltage fluctuation in a 10ms sampling period, and the power change threshold of 12kW corresponds to a 2.7% transient change rate in a rated 450kW system. Both of these settings are in line with the reasonable settings for the short-term disturbance response analysis of the power distribution system. The time number sequence obtained by the final screening is the effective number set used to extract mutation nodes.

[0093] S203: Call the set of double threshold over-limit numbers to mark the number, index the node number and time node in the original data, combine the node information that meets the conditions in order, associate the node number and time node, and generate a mutation node identifier sequence;

[0094] Retrieve the time number from the set of double threshold exceedance numbers. For the number in the set, such as... The system sequentially locates the record entries in the original multidimensional monitoring data set, indexes out the corresponding node number and time node information, and sets... Corresponding to bus number Node05, the recording time is 10:45:00. Corresponding to Node08, time is 10:45:10, the combination constitutes structured node identification information pair, such as (Node05, 10:45:00), (Node08, 10:45:10), the combination process is to call the numbering index function first to extract the original record line, extract the field "node number" and "time stamp" in the record line, and then combine to generate node identification sequence in key-value pair mode, all time numbers exceeding the double threshold value should perform the index and combination operation, if the original data set is a table structure sorted by time, the corresponding data row can be directly located by row number offset to complete the extraction operation, if there are multiple node synchronous records in the original data, the data record entries of multiple nodes corresponding to each time number also need to be found, to form a multiple node corresponding relationship table, which can be further reordered according to time sequence to form a structured output, such as [Node05@10:45:00, Node08@10:45:10], the sequence is the final mutation node identification sequence used to describe the voltage and power synchronous mutation event in the system, which has complete space-time positioning information.

[0095] Please refer to Figure 4 The specific steps of S3 are:

[0096] S301: Call the node in the mutation node identification sequence, access the topology connection relationship in the positive and negative directions of the topology structure in turn, obtain the zero sequence current component and capacitance current phase angle of its adjacent nodes, and obtain the adjacent node current phase parameter group according to the node level combination data;

[0097] The calling mutation node identifies the nodes in the node sequence, that is, the node number sequence known to have a disturbance, for each node, the first-order adjacent connection nodes in the positive direction and the reverse direction of the power grid topology are accessed in turn, the real-time zero sequence current data collected by the monitoring device connected to the node is read, the imaginary part information in the complex component of the zero sequence current is extracted, the imaginary part corresponds to the capacitive current in the capacitive system, then the capacitive current is compared with the standard reference voltage phase, the phase angle is calculated, and then the phase angle and the current amplitude corresponding to the node form a parameter group. In this process, the data needs to be combined and processed according to the actual hierarchical structure of the power grid, for example, if a node number is N1, its positive connection nodes are N2, N3, and its reverse connection nodes are N0, N-1, then the zero sequence current of N2, N3, N0 and N-1 is collected, and the phase angle of the capacitive current is formed. The current phase parameter group centered on N1, for example: if the current of N2 is 3.5A, the phase angle is-15°, the current of N3 is 4.1A, the phase angle is-18°, the current of N0 is 2.9A, the phase angle is-12°, and the current of N-1 is 3.2A, the phase angle is-14°, then the parameter group can be expressed as {N2:[3.5,-15°], N3:[4.1,-18°], N0:[2.9,-12°], N-1:[3.2,-14°]}. The above sampling frequency can be set to 10ms, which is determined according to the actual data sampling frequency of the power distribution terminal of the transformer substation. When combining data, the nodes need to be arranged in order from low to high in the topological hierarchical structure. If a node corresponds to multiple path connections, the path weight priority needs to be identified. This priority can be set according to the node stability level in the original operation data, for example, the path weight of the connection frequently disturbed node can be set to 0.7, and the rest can be set to 0.3, as shown in Table 3.

[0098] Table 3: adjacent node current parameter group sample table

[0099]

[0100] As shown in Table 3, the adjacent node current parameter group collects adjacent data centered on the current mutation node N1, and identifies the main propagation direction through the path weight, which provides a basis for subsequent judgment of the current disturbance direction.

[0101] S302: Based on the current component sequence and the capacitive current phase angle sequence in the adjacent node current phase parameter group, the current amplitude variation trend is identified in the sliding window, the energy gradient is deduced combined with the phase angle, the trend change in the window is calculated, and the disturbance energy gradient trend coefficient is obtained.

[0102] Based on the current component sequence in the adjacent node current phase parameter group and the capacitance current phase angle sequence, the node current amplitude is first arranged in a time sliding window manner, such as setting the window width to 100 ms, sampling once every 10 ms, then each window contains 10 groups of data, forming a current amplitude sequence and a corresponding phase angle sequence. For example, the N2 node records the current as [3.5, 3.6, 3.9, 4.0, 4.2, 4.1, 3.8, 3.6, 3.5, 3.4] A within 100 ms, and the corresponding phase angle is [-15, -14, -13, -12, -11, -12, -13, -14, -15, -16] °. When identifying the current amplitude variation trend, the variation amplitude of adjacent two groups of data is compared point by point, for example, the difference between the first group and the second group of current sequence is +0.1 A, and the difference between the second group and the third group is +0.3 A. If the variation direction of the three groups is consistent, it can be determined as an upward trend section. Combined with the sign direction of the difference value of the adjacent angles in the phase angle sequence, if the angle gradually decreases, it indicates that the capacitance current is ahead of the reference voltage, and the combination index of the current variation direction and the phase angle variation trend needs to be constructed accordingly. The energy gradient change is defined as ΔI x Δθ, where ΔI is the current amplitude difference between adjacent points, and Δθ is the adjacent phase angle difference. For example: ΔI = 4.2-4.0 = 0.2 A, Δθ = -11-(-12) = 1 °, then the energy gradient of this step is 0.2 x 1 = 0.2. After calculating the sum of ΔI x Δθ in the sliding window, the total energy gradient value of the window is obtained, and then the average disturbance energy gradient trend coefficient is obtained by dividing the number of data groups. If the cumulative ΔI x Δθ of the 10 groups of data in the window is 1.5, the average trend coefficient is 1.5 / 9≈0.167. Set the disturbance energy gradient interval as follows: less than 0.05 is regarded as a stable area, 0.05-0.15 is a weak disturbance area, and above 0.15 is an obvious disturbance path. Therefore, 0.167 belongs to the obvious disturbance area, which is used to identify whether the path is the main disturbance direction.

[0103] S303: According to the disturbance energy gradient trend coefficient, the energy propagation direction between nodes in the topology path is judged, the gradient continuously rising path is marked as the main energy propagation path, the fluctuation interval path is the secondary path, and the disturbance propagation path set is output.

[0104] According to the disturbance energy gradient trend coefficient, the gradient coefficients between the node pairs in the topological path are traversed, comparison and segmentation processing are performed according to the path continuity, if the trend coefficients of three consecutive node connection pairs in a path are all greater than 0.15, the path segment is marked as the main energy propagation path, if a node pair lower than 0.05 appears in the middle, it is judged as an energy transmission breakpoint, the path is cut off according to this boundary, the part with the coefficient between 0.05 and 0.15 in the fluctuation section is marked as the secondary path, for example, in the node path N1→N2→N3→N4, the trend coefficients of adjacent node pairs are 0.167, 0.185, 0.142, 0.045 and 0.087 respectively, N1→N2→N3 is marked as the main path, N4 and the subsequent nodes are regarded as the secondary path, the path marking is realized through the Boolean state matrix, if there is a branch in the node path chain, a trend coefficient chain table is established for each branch, the main and secondary path weight attribution is determined by comparing the average trend values of the chain paths, and finally the disturbance propagation path set is output in the following form: main path: [N1→N2→N3], secondary path: [N4→N5], untriggered path: [N6→N7], the path set is used as the calculation basis for subsequent propagation source positioning and boundary determination links.

[0105] Please refer to Figure 5 , the specific steps of S4 are:

[0106] S401: Obtain the virtual power distribution on the main path bus in the disturbance propagation path set, calculate the virtual power difference value of the bus nodes in the path based on the node number and virtual power in-out value of each path node, and combine the bus number information to summarize the node virtual power difference value, and generate the virtual power distribution value of the path node;

[0107] The virtual power distribution on the main path bus in the disturbance propagation path set is obtained. First, the node numbers of the nodes on each path in the disturbance propagation path set are extracted by analyzing the information of multiple paths, and the bus attribution relationship of the numbers in the topology structure is matched in turn. For example, when the path P1 contains the node numbers [5, 8, 12], the bus numbers where the nodes are located are [101, 104, 107]. Then, based on the virtual power input and output values of the nodes in the disturbance process, the virtual power injection and output values of the buses to which the nodes are connected are counted. For multiple nodes corresponding to any bus number, the difference between the virtual power injection value and the output value of the node is calculated as the virtual power difference value of the node. For example, the virtual power of node 8 connected to bus number 104 is 1.2 MVar before the disturbance process and 0.9 MVar after the disturbance process, so the virtual power difference value of the node is -0.3 MVar. For example, the virtual power input and output values of node 12 on bus 107 are 0.8 MVar and 1.0 MVar respectively, so the difference is +0.2 MVar. Then, the virtual power difference values of the nodes in the path are classified by bus number, and the virtual power difference values of the differentiated nodes on the bus are accumulated to obtain the virtual power distribution value of the bus node corresponding to the path. For example, the virtual power difference value of node 5 corresponding to bus number 101 in path P1 is +0.1 MVar, the virtual power difference value of node 8 corresponding to bus number 104 is -0.3 MVar, and the virtual power difference value of node 12 corresponding to bus number 107 is +0.2 MVar. The virtual power distribution value of path P1 is represented as {101:+0.1, 104:-0.3, 107:+0.2}. The distribution value is used for subsequent analysis of the correlation between the virtual power flow direction and the disturbance evolution.

[0108] S402: According to the virtual power distribution value of the path node, the virtual power flow value between the nodes before and after the disturbance is calculated, whether the flow direction is reversed under the same node differentiation time sequence is judged, whether the flow velocity variation amplitude exceeds the virtual power flow velocity amplitude threshold value is compared, and the virtual power flow change characteristic quantity is obtained.

[0109] According to the virtual power distribution value of the path node, the virtual power flow value between the nodes before and after the disturbance is calculated. First, select several time points t=1, 2, …, T in the disturbance time sequence T, for example T=3, t1=0.1s, t2=0.2s, t3=0.3s. For each time point, extract the virtual power flow values of node i and node j before and after the disturbance and and calculate the difference value at each time point At the same time, the dimensionless weight coefficient of node i and node j at this time point is called and The weight coefficient represents the normalized value of the node in the overall virtual power, and the value range is set to [0, 1]. If the virtual power of node i accounts for 30% at time t=0.1s, then The normalized coefficient of node i corresponding to the disturbance intensity is calculated , which is set as the ratio of disturbance amplitude to the maximum disturbance value of the whole system. For example, the disturbance amplitude of node i at t = 0.1s is 0.06 MVar, and the maximum disturbance amplitude of the whole system is 0.2 MVar, then , set a minimum positive number , avoid zero denominator, finally bring in the following parameters:

[0110] ;

[0111] ;

[0112] ;

[0113] Calculate the virtual power flow change characteristic quantity of the node, using the formula:

[0114] ;

[0115] Where, represents the virtual power flow change characteristic quantity of node before and after the disturbance, after the weighted disturbance intensity correction, represents the virtual power flow value of node before the disturbance at time , represents the virtual power flow value of node after the disturbance at time , represents the dimensionless weight coefficient of node at time , represents the dimensionless weight coefficient of node at time , represents the normalized disturbance intensity coefficient of node at time , represents the total number of statistical time steps, represents a minimum positive number to avoid zero denominator. Substitute to get:

[0116]

[0117]

[0118] ;

[0119] ;

[0120] ​The results show that the disturbance after the change degree of the virtual power flow between nodes i and j is 0.75, if the preset virtual power flow amplitude threshold is 0.6, it shows that the flow rate changes between the two nodes exceeds the threshold, and further identifies whether the flow direction is reversed in the differential time, for example, in t1 and t2 is positive and negative in t3, it can be determined that the flow direction is reversed, thereby marking the virtual power flow change characteristics between the nodes.

[0121] Table 4: Virtual power flow change characteristic data table

[0122]

[0123] As shown in Table 4, the virtual power values of nodes 5 and 8 before and after disturbance at three time points, the weight coefficient and the disturbance intensity normalization coefficient are listed, and is calculated based on this to evaluate the degree of virtual power flow change.

[0124] S403: Call the virtual power flow change characteristic, use the dynamic clustering analysis algorithm to determine the node distance, filter the nodes that meet the aggregation threshold condition, merge into the virtual power torsion knot structure, and establish the overvoltage risk node library;

[0125] Call the virtual power flow change characteristic, calculate the difference degree of the characteristic value between each pair of nodes, and use the distance measurement method to calculate the absolute value difference of the value between the differential nodes, for example, in the node pair , the values are obtained respectively, then the difference values between any two pairs of nodes can be calculated, for example, |0.75-0.78|=0.03, |0.78-0.35|=0.43, etc. The difference value results form a characteristic distance matrix, then set the aggregation threshold, for example, set the threshold δ=0.1, which means that when the characteristic value difference between any two pairs of nodes is not more than 0.1, they can be classified into the same aggregation class, for example, the difference between (i, j) and (j, k) is 0.03, which is less than δ, so they are classified into the same class, and then judge the difference between (k, l) and the above clustering center is 0.43, which is greater than δ, so (k, l) is classified alone. Finally, multiple aggregation groups are formed in the nodes, and the nodes with high frequency and frequent virtual power flow change in the aggregation group are extracted as the key nodes of the torsion knot structure, for example, the node set of aggregation group G1 is {5, 7, 9}, and G2 is {12, 14}, then mark nodes 5, 7 and 9 as the virtual power torsion knot structure, and record them in the overvoltage risk node library.

[0126] Please refer to Figure 6 , the specific steps of S5 are as follows:

[0127] S501: Obtain the voltage mutation parameter of the node in the overvoltage risk node library, combine the node voltage time series data, calculate the node voltage transient mutation amplitude, sort the first node voltage amplitude in descending order, and obtain the node voltage amplitude;

[0128] Obtain the voltage mutation parameter of the node in the overvoltage risk node library, first, for the overvoltage risk node information library registered in the power grid monitoring system, extract the node number and its original recorded voltage mutation characteristic parameter, which is derived from the voltage time series data collected during the operation of the node. By identifying the obvious change segment in the node voltage time series point by point, the voltage values at adjacent sampling time points are called, the voltage difference between adjacent time points is calculated, and the calculation operation is to subtract the voltage values of the two adjacent time points. For example, for the voltage time series of node A {227.5V, 228.0V, 235.3V, 236.1V, 228.5V, 229.0V}, first calculate the mutation sequence {0.5V, 7.3V, 0.8V, 7.6V, 0.5V}, then filter the mutation amplitude in the sequence, sort the absolute value of the amplitude, and arrange the obtained difference sequence in descending order to get the sorting result {7.6V, 7.3V, 0.8V, 0.5V, 0.5V}. Extract the maximum mutation amplitude at the top of the sorting to get the maximum voltage mutation amplitude of the current node in this time period, which is used as a representative mutation index for subsequent analysis. The voltage mutation amplitude of node A in this period is 7.6V. The above sorting process uses the absolute value operation on the mutation difference to compare the data. The specific execution process is to traverse each difference value from the difference value list and call the math.abs() function to process and store it in a new list. Then arrange the mutation amplitude values from large to small by using the list sorting function, so as to finally extract the maximum voltage mutation amplitude as the representative voltage mutation value for subsequent analysis.

[0129] S502: Based on the node voltage amplitude, extract the rated voltage value of the corresponding node, calculate the difference between the node voltage amplitude and the rated voltage value, analyze the amplitude deviation degree through the direct numerical difference between the two, and obtain the deviation amount of the node voltage amplitude and the rated voltage;

[0130] Based on the node voltage amplitude, extract the rated voltage value of the corresponding node, first call the node configuration table in the power grid planning database to identify the device rated voltage value corresponding to the node number. For example, node A is a transformer connection node in a 10kV distribution network system, and its rated voltage is 10000V. Further, calculate the difference between the mutation amplitude 7.6V obtained in S501 and the rated voltage value. The specific operation is to subtract the mutation amplitude from the rated voltage value directly, that is, execute the formula:

[0131] ;

[0132] Wherein, , , substitution formula:

[0133] ;

[0134] But the value is significantly larger, indicating that the rated voltage unit and the mutation voltage unit are inconsistent, and the unit needs to be unified. The mutation amplitude is derived from the instantaneous fluctuation, which reflects the short-time voltage disturbance value, which needs to be calculated with the relative amplitude of the rated voltage, so the percentage method is used for processing, and the formula is changed to:

[0135] ;

[0136] After substitution, we get:

[0137] ;

[0138] This percentage represents the proportion of the current mutation voltage to the rated voltage, representing the degree of voltage fluctuation deviation. This value will be compared with the set overvoltage threshold in the next step. To standardize the extraction of rated voltage and deviation calculation operation in the implementation process, the following table lists the rated voltage of the differentiated type node and the calculated mutation percentage deviation:

[0139] Table 5: Voltage mutation deviation table of power grid nodes

[0140]

[0141] As shown in Table 5, the deviation value of node C is 0.553%, which is significantly higher than that of nodes A and B, indicating a higher degree of short-time voltage disturbance.

[0142] S503: Call the deviation of the node voltage amplitude and the rated voltage, and compare the numerical value with the overvoltage threshold to determine whether the deviation is outside the threshold boundary range. If the deviation is greater than the threshold range, an overvoltage prediction result is generated;

[0143] Call the deviation of the node voltage amplitude and the rated voltage, and compare the numerical value with the overvoltage threshold to determine whether the deviation is outside the threshold boundary range. First, the threshold range of voltage mutation deviation needs to be clearly defined. According to the recommendations in the "Power System Design Specification GB50062", the short-time voltage deviation in the 10kV below distribution network should not exceed ±5%, so the deviation threshold is set to ±5%, that is, the deviation exceeds 5% is considered to occur overvoltage risk. If the current node deviation value is 0.076% (from node A), compare the deviation value with the threshold value through numerical judgment operation, and execute the judgment:

[0144] ;

[0145] The judgment result is not exceeding the threshold range, and it does not constitute an overvoltage warning. However, for node C, the deviation is 0.553%, and the same judgment is performed:

[0146] ;

[0147] It still does not exceed the threshold value. If the mutation amplitude of a certain node is 950V, and the rated voltage is 10,000V, then the deviation is:

[0148] ;

[0149] Compared with 5.0%:

[0150] ;

[0151] Then the deviation is judged to exceed the threshold value, and it is determined that the current node has an overvoltage condition, and the corresponding overvoltage prediction result is generated. The processing action is to store the node number, mutation amplitude, deviation percentage value and judgment result to the overvoltage prediction record library, and mark the node as a high-risk node for subsequent scheduling or maintenance operation. In this judgment process, the threshold value is a fixed configuration item, which needs to be set in combination with the system voltage level, such as 5% for a 10kV distribution system, and 2% for a 220kV transmission system. The differentiated level is gradually reduced according to the national standard.

[0152] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method of overvoltage prediction for an electric power system, characterized by, The method comprises the following steps: S1: obtaining voltage time sequence signals, current components, bus power flow rates through power system monitoring nodes, and performing ripple amplitude extraction processing on the voltage time sequence signals, while collecting zero sequence current components, capacitor current phase angles, power transmission change rates to form a multi-dimensional monitoring data set; S2: calling the multi-dimensional monitoring data set to perform ripple amplitude difference, calculating the ripple amplitude change quantity and the power flow rate change quantity between adjacent time points, comparing the ripple mutation threshold and the rate change threshold, marking the double-threshold overrun nodes to form a mutation node identification sequence; S3: calling the nodes in the mutation node identification sequence, obtaining adjacent node zero sequence current components and capacitor current phase angles along the topological forward and reverse progression, using a sliding window prediction algorithm to analyze the disturbance amplitude increase and decrease trend, identifying the main energy propagation path and the secondary branch path, and outputting a disturbance propagation path set; S4: obtaining the main path bus reactive power distribution and the virtual power flow direction value of the disturbance propagation path set, calculating the flow direction change quantity and the flow rate change quantity, identifying the flow direction reverse and amplitude super threshold nodes, using a dynamic clustering analysis algorithm to screen the virtual power torsion knot structure, and establishing an overvoltage risk node library.

2. The overvoltage prediction method of a power system according to claim 1, characterized by, The multi-dimensional monitoring data set includes voltage effective value, current component and bus power flow rate, the mutation node identification sequence includes node number, waveform mutation type and topological position information, the disturbance propagation path set includes main path serial number, zero sequence current trend and energy transfer direction, and the overvoltage risk node library includes virtual power abnormal node, risk level and node distribution characteristics.

3. The overvoltage prediction method of a power system according to claim 1, wherein, The specific steps of S1 are: S101: obtaining voltage time sequence signals, current components, bus power flow rates through power system monitoring nodes, dividing the voltage signals according to the period and calculating the fluctuation difference value, extracting the amplitude mutation points of adjacent periods, combining the local peak values to construct the amplitude interval, and generating the voltage fluctuation amplitude interval; S102: calling the voltage fluctuation amplitude interval, collecting zero sequence current components and capacitor current phase angles, analyzing the amplitude change trend of the section, comparing the zero sequence current frequency change of the corresponding period, calculating the correlation and superimposing the capacitor current phase difference value, and generating the phase coupling change coefficient; S103: based on the phase coupling change coefficient, calling the bus power flow rate data and performing rate differentiation processing, synchronously analyzing the phase coupling and power change trend according to the unified sampling time axis, and generating a multi-dimensional monitoring data set.

4. The overvoltage prediction method of a power system according to claim 3, wherein The specific steps of S2 are: S201: calling the voltage waveform value and power flow rate value in the multi-dimensional monitoring data set, extracting the voltage ripple amplitude of adjacent time points, calculating the change quantity by calculating the adjacent difference value, recording the power flow rate change value of the corresponding period, grouping according to the time period, and generating a difference change reference value sequence; S202: based on the number in the difference change reference value sequence, calling the corresponding ripple amplitude change value and power change value, comparing the ripple mutation threshold and the flow rate change threshold respectively, selecting the synchronous overrun number that meets the condition, eliminating the records that do not meet the condition, and obtaining a double-threshold overrun number set; S203: Call the double-threshold over-limit number set marker number, index the node number and time node in the original data, combine the node information that meets the conditions in order, associate the node number and time node, and generate a mutation node identification sequence.

5. The overvoltage prediction method of a power system according to claim 4, wherein The specific steps of S3 are: S301: Call the nodes in the mutation node identification sequence, access the topology connection relationship in the positive and negative directions of the topology structure in turn, obtain the zero sequence current component and capacitance current phase angle of the adjacent nodes, and obtain the adjacent node current phase parameter group according to the node level combination data; S302: Based on the current component sequence and capacitance current phase angle sequence in the adjacent node current phase parameter group, identify the current amplitude trend in the sliding window, combine the phase angle to deduce the energy gradient, calculate the trend change in the window, and obtain the disturbance energy gradient trend coefficient; S303: According to the disturbance energy gradient trend coefficient, judge the energy propagation direction between nodes in the topology path, mark the gradient continuously rising path as the main energy propagation path, and the fluctuation interval path as the secondary path, and output the disturbance propagation path set.

6. The overvoltage prediction method of a power system according to claim 5, wherein The specific steps of S4 are: S401: Obtain the virtual power distribution on the main path bus in the disturbance propagation path set, calculate the virtual power difference value of the bus node in the path based on the node number and virtual power in-out value of each path node, and combine the bus number information to summarize the node virtual power difference value, and generate the path node virtual power distribution value; S402: According to the path node virtual power distribution value, calculate the virtual power flow direction value before and after the disturbance between nodes, judge whether the flow direction is reversed under the same node difference time sequence, compare whether the flow velocity variation amplitude exceeds the virtual power flow velocity amplitude threshold value, and obtain the virtual power flow direction change characteristic quantity; S403: Call the virtual power flow direction change characteristic quantity, use a dynamic clustering analysis algorithm to judge the characteristic value distance between nodes, filter nodes that meet the aggregation threshold condition, and merge them into a virtual power kink structure, and establish an overvoltage risk node library.

7. The overvoltage prediction method of a power system according to claim 6, wherein, The path node virtual power distribution value is a bus node virtual power difference value set calculated and summarized along the nodes in the disturbance propagation path according to the node number and corresponding virtual power in-out value; The virtual power flow direction change characteristic quantity is a difference index calculated based on the path node virtual power distribution value before and after the disturbance between nodes; The aggregation threshold condition is a numerical boundary parameter used to judge whether the distance between the virtual power flow direction change characteristic quantities of the nodes meets the merging requirement in the dynamic clustering analysis; The virtual power kink structure refers to a node set formed by merging based on the reverse change of the virtual power flow direction between nodes and the flow velocity amplitude threshold condition on the disturbance propagation path.

8. The overvoltage prediction method of a power system according to claim 1, wherein, The method further includes the S5 step: S5: Sort the node voltage mutation parameters in the overvoltage risk node library in descending order, extract the first node voltage amplitude in the order, calculate the deviation between the node voltage amplitude and the rated voltage, and generate an overvoltage prediction result when the deviation exceeds the overvoltage threshold value; The overvoltage prediction result includes the node identification number, the voltage deviation, and the warning time tag.

9. The overvoltage prediction method of a power system according to claim 8, wherein, The specific steps of S5 are: S501: Obtain the voltage mutation parameter of the node in the overvoltage risk node library, combine the node voltage time series data, calculate the node voltage transient mutation amplitude, sort the first node voltage amplitude in descending order, and obtain the node voltage amplitude; S502: Based on the node voltage amplitude, extract the rated voltage value of the corresponding node, calculate the difference between the node voltage amplitude and the rated voltage value, analyze the amplitude deviation degree through the direct numerical difference between the two, and obtain the deviation amount of the node voltage amplitude and the rated voltage; S503: Call the deviation amount of the node voltage amplitude and the rated voltage, and compare the numerical value with the overvoltage threshold value to determine whether the deviation amount exceeds the threshold boundary range. If the deviation amount is greater than the threshold range, an overvoltage prediction result is generated.

10. The overvoltage prediction method of a power system according to claim 9, wherein, The node voltage amplitude is the voltage change amount of the node voltage during transient mutation, which represents the mutation characteristics of the node voltage in a specific time sequence; The deviation amount of the node voltage amplitude and the rated voltage is the absolute difference between the node voltage amplitude and the node rated voltage value; The overvoltage threshold value is a preset voltage numerical limit value, which is compared with the deviation amount of the node voltage amplitude and the rated voltage to determine whether there is an overvoltage situation.

Images