A power system inertia distribution evaluation method and system and a storage medium

Abstract

The present application relates to the technical field of power system, and more particularly to a kind of power system inertia distribution evaluation method, system and storage medium, method includes: the active power and frequency measurement data of power system power source side grid-connected node are collected, and measurement data is filtered and cleaned, bad data is rejected and interpolation reconstruction, and normalization processing is carried out, form active power standard data sequence and frequency standard data sequence;Combining sliding window technology and using detrended fluctuation analysis active power standard data sequence and frequency standard data sequence, identify the initial time of system disturbance occurrence;Based on generator group power frequency response characteristics, construct inertia evaluation analytical model from energy point of view;Through the continuous input of real-time measurement data and model iterative calculation, dynamically update inertia evaluation result, and determine final inertia evaluation value according to quasi-steady state result.The present application can realize the real-time perception of inertia distribution in power grid, and does not need excessive dependence on existing high-precision model.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and in particular to a method, system and storage medium for evaluating the inertia distribution of a power system. Background Technology

[0002] With the large-scale and high-proportion integration of new energy power generation, represented by wind power and photovoltaics, into the power grid, the proportion of traditional synchronous power sources continues to decline, resulting in a significant downward trend in the overall inertia level of the power system. This poses an increasingly severe challenge to the frequency stability of the power grid. As an inherent physical property of the power system, inertia is the first line of defense against power surges and the maintenance of frequency stability. Its accurate assessment is of vital importance for understanding the dynamic characteristics of the system and ensuring safe operation.

[0003] Currently, the mainstream methods for inertia assessment can be broadly categorized into two types. The first is the method based on mathematical models and simulations. This method relies on detailed mathematical models, accurate parameters, and system topology. However, in practical engineering applications, it often introduces significant errors due to model simplification, inaccurate parameters, or difficulty in obtaining them in real time, leading to distorted assessment results. The second type is the measurement and analysis method based on post-disturbance response. This type of method mostly calculates inertia by capturing the rate of frequency change after a large disturbance event. While this reduces the reliance on accurate models to some extent, its effectiveness is often limited by the intensity of the disturbance event, measurement noise, and data quality.

[0004] Furthermore, existing technologies mostly focus on overall inertia estimation at the network-wide or regional level, lacking the ability to finely and spatially identify the distribution of inertia on the power source side. With increasingly complex power grid structures, flexible and varied operating modes, and random fluctuations in renewable energy output, the distribution of inertia in the power grid exhibits uneven and dynamically evolving characteristics. Traditional assessment methods struggle to provide operation and dispatch personnel with real-time, accurate, and detailed information on the spatial distribution of inertia, thus hindering the optimization of frequency-coordinated control strategies and the improvement of adaptive regulation capabilities. Summary of the Invention

[0005] This invention provides a method, system, and storage medium for evaluating the inertia distribution of a power system, which can effectively solve the problems of uneven inertia distribution in the current power grid, difficulty for operators to perceive it in real time, and excessive reliance on high-precision models in existing methods.

[0006] This invention provides a method for evaluating the inertia distribution of a power system, comprising: The active power and frequency measurement data of the grid-connected nodes on the power supply side of the power system are collected, and the measurement data are filtered, cleaned, bad data is removed and interpolated and reconstructed, and normalized to form a standard data sequence of active power and a standard data sequence of frequency. By combining sliding window technology and employing detrended fluctuation analysis of active power standard data sequences and frequency standard data sequences, the initial moment of system disturbance is identified; Based on the power frequency response characteristics of generator sets, an analytical model for inertia assessment is constructed from an energy perspective. By continuously inputting real-time measurement data and iteratively calculating the model, the inertia assessment results are dynamically updated, and the final inertia assessment value is determined based on the quasi-steady-state results.

[0007] Furthermore, the specific steps for filtering and cleaning the measurement data, removing bad data, interpolating and reconstructing, and normalizing the data include: The active power measurement data are integrated into an active power data sequence in chronological order, and the frequency measurement data are integrated into a frequency data sequence in chronological order. For each data point in the active power data sequence and frequency data sequence, determine its corresponding sliding time window, and calculate the median and absolute deviation of the median of all data within the window. Based on the median and the absolute deviation of the median, it is determined whether each data point in the frequency data sequence and the active power data sequence is an outlier. If the difference between the data point and the corresponding median exceeds a set threshold, it is marked as an outlier data point; otherwise, it is marked as a normal data point. For data points marked as anomalous, linear interpolation is performed using the values ​​of the two nearest normal data points to reconstruct effective alternative data. For all normal data points and reconstructed valid data points in the frequency data sequence and active power data sequence, the difference is calculated and normalized based on the rated frequency and rated power of the unit to complete the data normalization.

[0008] Furthermore, the specific algorithm for calculating the median and absolute deviation of the median of all data within this window is as follows: ; ; In the formula, The median of the active power measurement data sequence obtained by the generator set within the nth sliding window; P i This refers to the active power data acquired by the generator set at the i-th sampling time. MED stands for Median of a data sequence. h represents the number of sampling points contained within a single sliding window; P i,res For data P i The absolute deviation relative to the middle value within the nth sliding window; The median of the frequency measurement data sequence obtained by the generator set within the nth sliding window; f i This refers to the frequency data acquired by the generator set at the i-th sampling time. f i,res For data f i The absolute deviation relative to the middle value within the nth sliding window.

[0009] Furthermore, the specific algorithm for distinguishing between outlier and normal data points is as follows: ; ; In the formula, and These are respectively the abnormal power data points and the abnormal frequency data points of the generator set; and These are normal power data points and normal frequency data points for the generator set. c is the threshold for judging abnormal data.

[0010] Furthermore, the specific algorithm for reconstructing effective alternative data is as follows: ; In the formula, and These are the valid data points obtained after linear interpolation reconstruction of generator power and frequency, respectively. and These are the power anomaly data points. The two nearest preceding and following normal power data points in the time series; and These are the frequency anomaly data points. The two nearest preceding and following normal frequency data points in the time series.

[0011] Furthermore, the specific algorithm for per-unit processing is as follows: ; ; In the formula, and These are the dimensionless values ​​corresponding to the changes in generator power and frequency after being normalized to per-unit. S gen This refers to the rated power of the generator set. f s This is the rated frequency.

[0012] Furthermore, the specific steps for identifying the initial moment of a system disturbance include: For each data point in the active power standard data sequence and the frequency standard data sequence, based on its own corresponding sliding time window, calculate the median and the absolute deviation of the median of all data in the window, and use the time series composed of the absolute deviations of the median corresponding to each window as a first-time processing signal. Based on the first-processed signals obtained under different windows, a second-order polynomial fitting is performed, and the fitted signal is used as the second-order processed signal. The difference between the primary and secondary processed signals under different windows is calculated to obtain the detrended processed signal; Calculate the root mean square value of the detrended signal corresponding to different sliding windows and use it as the criterion for the timing of system disturbance. When the criterion exceeds a preset threshold, it is determined that a system disturbance has occurred within the sliding window, and the starting time corresponding to the window is determined as the initial time of the disturbance.

[0013] Furthermore, the specific algorithm for the power frequency response characteristics of the generator set is as follows: ; In the formula, H gen The inertial time constant of the generator set; This represents the per-unit value of the frequency variation of the generator set; This represents the per-unit value of the change in the mechanical power of the generator set. This represents the per-unit value of the change in active power of the generator set; D gen This is the damping coefficient of the generator set.

[0014] Furthermore, the specific algorithm for constructing an analytical model for inertia assessment from an energy perspective is as follows: ; In the formula, The inertia assessment result of the generator set at time t; t0 is the initial time of data sampling; and Let τ be the per-unit value of the change in mechanical power and the change in active power of the generator set at time τ; The frequency per unit value of the generator set at time t; This represents the per-unit frequency value of the generator set at time t0.

[0015] Furthermore, the specific algorithm for dynamically updating the inertia assessment results is as follows: ; In the formula, The inertia assessment result of the generator set at time t-1; η is the threshold for judging frequency changes; This is the inertia assessment result of the generator set at the initial moment, which is a preset initial value; The trapezoidal numerical integration method is used to solve the problem, and the specific algorithm is as follows: ; In the formula, M is the number of active power samples collected by the generator set during the time period from the initial time t0 to time t. This refers to the Mth active power change data collected by the generator set during the time period from time t0 to t.

[0016] This invention also provides a power system inertia distribution assessment system, comprising: The data acquisition module is used to collect measurement data of active power and frequency at the grid-connected nodes on the power supply side of the power system, and to filter and clean the measurement data, remove bad data and reconstruct it by interpolation, and then perform normalization processing to form a standard data sequence of active power and a standard data sequence of frequency. The identification module is used to identify the initial moment of system disturbance by combining sliding window technology and detrended fluctuation analysis of active power standard data sequence and frequency standard data sequence. The modeling module is used to construct an analytical model for inertia assessment from an energy perspective based on the power frequency response characteristics of generator sets. The final evaluation module is used to dynamically update the inertia evaluation results through continuous input of real-time measurement data and iterative calculation of the model, and to determine the final inertia evaluation value based on the quasi-steady-state results.

[0017] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the power system inertia distribution assessment method described above.

[0018] The technical solution of this invention can achieve the following technical effects: 1. This method achieves refined perception of inertia distribution: Traditional methods are mostly limited to rough estimation of the overall system inertia, while this method, through the fusion processing of distributed measurement data, can effectively identify the distribution structure of inertia on the power supply side, providing a certain spatial resolution capability and situational awareness depth for operation scheduling.

[0019] 2. This method effectively overcomes the dependence on high-precision physical models: Existing model- and simulation-based methods often lead to evaluation biases due to inaccurate parameters and model mismatch. This method is based entirely on real-time measurement data and adopts a hybrid modeling approach that combines data-driven and physical laws, which significantly improves the reliability and adaptability of the evaluation results. It is especially suitable for actual systems with incomplete model information or time-varying parameters.

[0020] 3. This method has good anti-interference and data fault tolerance capabilities: In response to common problems such as noise, outliers and missing data in field measurements, this invention designs a hierarchical and progressive data cleaning and robust reconstruction mechanism. Combined with detrended fluctuation analysis and adaptive threshold determination, it effectively suppresses the impact of data quality on the evaluation results and improves the applicability and stability of the method.

[0021] 4. This method provides a clear physical interpretation and an adjustable evaluation mechanism: The inertia evaluation model constructed by this method is derived based on the principle of energy conservation and has a clear physical meaning. At the same time, by introducing dynamic update and quasi-steady-state determination strategies, the evaluation rhythm and output form can be adaptively adjusted according to the operating scenario and data characteristics, taking into account both the timeliness of the calculation and the reliability of the results. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 A flowchart illustrating the method for evaluating the inertia distribution of a power system; Figure 2 This is a topology diagram of the power system in an embodiment of the present invention; Figure 3 This is a schematic diagram of the frequency response of unit G4 under different scenarios in an embodiment of the present invention, obtained after simulation and data preprocessing. Figure 4 This is a data illustration of the inertia assessment result update process for three generator sets (G2, G3, and G4) in scenario 1 of this embodiment of the invention. Figure 5 This is a data diagram illustrating the fluctuation index of the frequency detrending processed signal corresponding to each calculation window of unit G4 in scenario 2 of this embodiment of the invention. Figure 6 This is a data diagram illustrating the inertia assessment result update process for three generator sets (G2, G3, and G4) in scenario 2 of this embodiment of the invention. Detailed Implementation

[0024] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0026] This invention relates to a method for evaluating the inertia distribution of a power system, such as... Figure 1 As shown, the process includes multiple steps S1 to S4. These steps construct a complete processing chain from raw measurement data to inertia distribution patterns, ultimately outputting the real-time inertia assessment value for each generator unit, thus forming a spatial distribution map of inertia on the power supply side. The specific details of each step are as follows: S1: Collect active power and frequency measurement data from grid-connected nodes on the power supply side of the power system, and perform filtering, cleaning, defective data removal, and interpolation reconstruction on the measurement data. This step is mainly used to remove common noise, outliers, and missing data in field measurements to ensure the quality of input data. After data processing, normalization is performed to eliminate the influence of data dimensions, making different data comparable, thereby forming a standard data sequence for active power and a standard data sequence for frequency.

[0027] S2: Combining sliding window technology with detrended fluctuation analysis of active power standard data series and frequency standard data series, by analyzing the local fluctuation characteristics of the data, the occurrence of disturbance events is automatically detected, the initial moment of system disturbance is identified, and an accurate starting time reference is provided for inertia assessment.

[0028] S3: Based on the power frequency response characteristics of generator sets, an analytical model for inertia evaluation is constructed from an energy perspective, establishing the mathematical relationship between inertia and active power and frequency change.

[0029] S4: By continuously inputting real-time measurement data and iteratively calculating the model, the inertia assessment results are dynamically updated, and the final inertia assessment value is determined based on the quasi-steady-state results.

[0030] Preferably, the specific steps for filtering and cleaning the measurement data, removing bad data, interpolating and reconstructing, and normalizing the data include: S11: Integrate the active power measurement data into an active power data sequence in chronological order, and integrate the frequency measurement data into a frequency data sequence in chronological order.

[0031] S12: For each data point in the active power data sequence and frequency data sequence, determine its corresponding sliding time window, and calculate the median and median absolute deviation of all data within the window; compared with the mean and standard deviation, the median and median absolute deviation are not sensitive to outliers, and can more accurately reflect the central trend and dispersion of the data when there is noise and outliers.

[0032] S13: Based on the median and the absolute deviation of the median, determine whether each data point in the frequency data sequence and the active power data sequence is an outlier. If the difference between the data point and the corresponding median exceeds the set threshold, it means that the data point deviates from the local central trend by a much greater degree than the normal fluctuation range, and then it is marked as an outlier data point; otherwise, it is marked as a normal data point.

[0033] S14: For data points marked as anomalous, in order to ensure the continuity and integrity of the data sequence, data reconstruction is required. At this time, linear interpolation can be performed using the values ​​of the two nearest normal data points to reconstruct effective replacement data. S15: Perform difference calculation and per-unit processing on all normal data points and reconstructed valid data points in the frequency data sequence and active power data sequence, respectively, based on the rated frequency and rated power of the unit, to complete data normalization.

[0034] Preferably, the specific algorithm for calculating the median and absolute deviation of the median of all data within the window in S12 is as follows: ; ; In the formula, The median of the active power measurement data sequence obtained by the generator set within the nth sliding window; P i This refers to the active power data acquired by the generator set at the i-th sampling time. MED stands for Median of a data sequence. h represents the number of sampling points contained within a single sliding window; P i,res For data P i The absolute deviation relative to the middle value within the nth sliding window; The median of the frequency measurement data sequence obtained by the generator set within the nth sliding window; f iThis refers to the frequency data acquired by the generator set at the i-th sampling time. f i,res For data f i The absolute deviation relative to the middle value within the nth sliding window.

[0035] Preferably, the specific algorithm for judging abnormal data points and normal data points in S13 is as follows: ; ; In the formula, and These are respectively the abnormal power data points and the abnormal frequency data points of the generator set; and These are normal power data points and normal frequency data points for the generator set. c is the threshold for judging outliers, which can be adjusted according to the actual data quality. The larger the value of c, the more lenient the judgment of outliers.

[0036] Preferably, the specific algorithm for reconstructing effective alternative data is as follows: ; In the formula, and These are the valid data points obtained after linear interpolation reconstruction of generator power and frequency, respectively. and These are the power anomaly data points. The two nearest preceding and following normal power data points in the time series; and These are the frequency anomaly data points. The two nearest preceding and following normal frequency data points in the time series.

[0037] Preferably, the specific algorithm for per-unit processing is as follows: ; ; In the formula, and These are the dimensionless values ​​corresponding to the changes in generator power and frequency after being normalized to per-unit. S gen This refers to the rated power of the generator set. f s This is the rated frequency.

[0038] Preferably, the specific steps for identifying the initial moment of a system disturbance include: S21: For each data point in the active power standard data sequence and the frequency standard data sequence, based on its own corresponding sliding time window, calculate the median and the absolute deviation of the median of all data within the window, and use the time series composed of the absolute deviations of the median corresponding to each window as a first-order processing signal. This signal reflects the evolution of the local fluctuation intensity of the data.

[0039] S22: Based on the obtained primary processing signals under different windows, a quadratic polynomial is fitted. The quadratic polynomial can fit the local gradual trend well. The fitted signal is used as the secondary processing signal, which can reflect the changing trend of the primary processing signal within the window.

[0040] S23: Calculate the difference between the primary and secondary processed signals under different windows to obtain the detrended processed signal. The detrended processed signal removes the overall trend of the primary processed signal and can effectively highlight the local fluctuation characteristics caused by sudden events such as system disturbances.

[0041] S24: Calculate the root mean square value of the detrended processing signal corresponding to different sliding windows, and use it as the criterion for the timing of system disturbance; when the criterion exceeds the preset threshold, it is determined that a system disturbance has occurred within the sliding window, and the starting time corresponding to the window is determined as the initial time of the disturbance.

[0042] Preferably, the specific algorithm for the power frequency response characteristics of the generator set is as follows: ; In the formula, H gen The inertial time constant of the generator set is used to characterize the inertia level of the unit. This represents the per-unit value of the frequency variation of the generator set; This is the per-unit value of the mechanical power change of the generator set, and its change in the initial stage of the inertial response is usually negligible. This represents the per-unit value of the change in active power of the generator set; D gen This is the damping coefficient of the generator set, which is usually small and can be ignored in inertia assessment.

[0043] Preferably, the specific algorithm for constructing an analytical model for inertia assessment from an energy perspective is as follows: ; In the formula, The inertia assessment result of the generator set at time t; t0 is the initial time of data sampling; and Let τ be the per-unit value of the change in mechanical power and the change in active power of the generator set at time τ; The frequency per unit value of the generator set at time t; This represents the per-unit frequency value of the generator set at time t0.

[0044] Based on the power frequency response characteristics of the generator set, and taking into account the change in mechanical power... With damping term D gen During the dominant phase of the inertial response, its value is relatively small and can be ignored. Therefore, the analytical model for inertial evaluation is simplified accordingly to: .

[0045] Preferably, the specific algorithm for dynamically updating the inertia assessment results is as follows: ; In the formula, The inertia assessment result of the generator set at time t-1; η is the threshold for judging frequency change. It is very small and is set to a very small positive number close to zero to avoid singular problems in the model solution when the frequency change is extremely small. This is the inertia assessment result of the generator set at the initial moment, which is a preset initial value; For the power integral term in the formula The trapezoidal numerical integration method is used to solve the problem, and the specific algorithm is as follows: ; In the formula, M is the number of active power samples collected by the generator set during the time period from the initial time t0 to time t. This refers to the Mth active power change data collected by the generator set during the time period from time t0 to t.

[0046] The method will be further illustrated below through an example: Based on the above implementation steps, this embodiment utilizes the power system simulation software PSD-BPA to build an improved IEEE-39 node system as a test system, and performs simulation verification on the proposed power system inertia distribution evaluation method based on power frequency response characteristics and energy perspective. The improved IEEE-39 node system topology is as follows: Figure 2As shown, the system includes seven synchronous generator units (G2, G3, G4, G6, G8, G9, and G10) and three new energy generator units (N1 to N3) using virtual inertia control. The total installed capacity of the system is 9490MW, with 2100MW of new energy installed capacity. The system's rated frequency and sampling frequency are both set to 50Hz in the simulation. The inertia reference value H for each unit is... gen and its rated capacity S gen as follows: Unit number N1: S gen =600MW, H gen =3.00s; Unit number N2:S gen =700MW, H gen =3.50s; Unit number N3:S gen =800MW, H gen =4.00s; Unit number G2:S gen =1300MW, H gen =3.00s; Unit number G3:S gen =918MW, H gen =4.36s; Unit number G4:S gen =800MW, H gen =3.58s; Unit number G6:S gen =1000MW, H gen =3.80s; Unit number G8:S gen =1000MW, H gen =4.98s; Unit number G9:S gen =1173MW, H gen =2.94s; Unit number G10:S gen =1200MW, H gen =4.15s.

[0047] The disturbance fault scenario in the simulation test is as follows: Scenario 1 (Disturbance test of a single node): At t=0s, a load step disturbance is introduced at node 25, with a total disturbance power of 400MW.

[0048] Scenario 2 (Disturbance test with multiple nodes and different disturbance occurrence times): At t=2s, load disturbances are simultaneously introduced at nodes 18, 20 and 25, with a total disturbance power of 300MW.

[0049] The initial value for inertia assessment is set to 2s. Figure 3 The frequency deviation response curves of unit G4 under the above two scenarios are shown after simulation and data preprocessing.

[0050] Disturbance test based on scenario 1: Taking units G2, G3, and G4 as observation objects, the dynamic update process of their inertia assessment results under scenario 1 is as follows: Figure 4 As shown in the figure. Analysis reveals a certain deviation between the initial inertia assessment value set at the initial moment and the actual inertia reference value of the unit. With the continuous input of real-time measurement data and the iterative update of the assessment model, the inertia assessment error gradually converges. At t≈0.5s, the assessment results of each unit have basically converged to near the inertia reference value and maintained a stable state in the subsequent period, demonstrating that this method has good dynamic convergence and steady-state maintenance capabilities. Based on this method, the steady-state inertia assessment values ​​H of each unit obtained in scenario 1 are shown. est The evaluation error (err) is as follows: Unit number N1: H gen =3.00s, H est =3.008s, err=0.267%; Unit number N2: H gen =3.50s, H est =3.514s, err=0.400%; Unit number N3: H gen =4.00s, H est =4.019s, err=0.475%; Unit number G2: H gen =3.00s, H est =3.001s, err=0.033%; Unit number G3: H gen =4.36s, H est =4.358s, err=0.046%; Unit number G4: H gen =3.58s, H est =3.581s, err=0.028%; Unit number G6: H gen =3.80s, H est =3.803s, err=0.079%; Unit number G8: H gen =4.98s, H est =5.041s, err=1.255%; Unit number G9:H gen=2.94s, H est =2.956s, err=0.544%; Unit number G10: H gen =4.15s, H est =4.172s, err=0.530%.

[0051] It can be seen that when the disturbance occurs at a single location, the inertia distribution assessment method based on this approach has high accuracy, all within 1.5%, with a maximum error of 1.225% and a minimum error of 0.028%. The inertia assessment error for most units is less than 1%, demonstrating high precision in accurately estimating the inertia of both synchronous and renewable energy units. In the scenario of a single-location disturbance, the inertia assessment value obtained using this method shows a very high degree of agreement with the preset benchmark inertia value for the unit, with the relative assessment error for each unit not exceeding 1.5%. Specifically, the maximum assessment error is 1.225% (corresponding to unit G8), and the minimum assessment error reaches 0.028% (corresponding to unit G4), with the assessment error for the vast majority of units being less than 1%. Furthermore, the evaluation results show that this method not only has excellent assessment accuracy for traditional synchronous units but also achieves accurate and reliable inertia estimation for renewable energy units, fully demonstrating the good applicability of the proposed method across different power source types.

[0052] Perturbation test based on scenario 2: To verify the effectiveness of the proposed disturbance timing identification method, the following analysis was conducted for scenario 2: Based on the simulated frequency data of unit G4, the sliding window length was set to 10 sampling points, and the root mean square value of the detrended signal corresponding to each window was calculated to obtain the fluctuation index sequence as follows: Figure 5 As shown, the threshold value is set to 0.05. Analysis reveals that within the window number range (less than 10), the fluctuation index for each window remains at a low level; however, at window number 10, the fluctuation index shows a significant jump and exceeds the set threshold. Based on this, the disturbance occurs around t≈2s, a result consistent with the simulation settings, thus verifying the accuracy and effectiveness of the disturbance timing determination method proposed in this embodiment.

[0053] The update process of the inertia assessment results for units G2, G3, and G4 in scenario 2 is as follows: Figure 6 As shown, before the disturbance occurs (t<2s), the inertia assessment values ​​of each unit remain at their initial preset values; once the system enters the disturbance response phase (t≥2s), the assessment results begin to converge rapidly towards the actual inertia reference values. Around t≈2.5s, the assessment values ​​of each unit have stabilized and approached the reference values, and remain in a quasi-steady state in the subsequent period. The steady-state inertia assessment values ​​H of each unit obtained in scenario 2 based on this method are shown below. est The evaluation error (err) is as follows: Unit number N1: H gen =3.00s, H est =3.014s, err=0.467%; Unit number N2: H gen =3.50s, H est =3.522s, err=0.629%; Unit number N3: H gen =4.00s, H est =4.026s, err=0.650%; Unit number G2: H gen =3.00s, H est =3.006s, err=0.200%; Unit number G3: H gen =4.36s, H est =4.364s, err=0.092%; Unit number G4: H gen =3.58s, H est =3.615s, err=0.978%; Unit number G6: H gen =3.80s, H est =3.818s, err=0.474%; Unit number G8: H gen =4.98s, H est =5.011s, err=0.622%; Unit number G9:H gen =2.94s, H est =2.955s, err=0.510%; Unit number G10: H gen =4.15s, H est =4.169s, err=0.458%.

[0054] As can be seen, the unit inertia assessment results obtained using this method in Scenario 2 still exhibit good consistency with the theoretical inertia setpoint. The relative errors of all assessed units, including synchronous units and new energy units, are strictly controlled within 1%. Specifically, the maximum assessment error is 0.978% (corresponding to unit G4), and the minimum assessment error is as low as 0.092% (corresponding to unit G3). This result not only verifies the accuracy of the method under different disturbance timings and spatial locations, but also further demonstrates that this method has strong robustness and wide applicability to the spatiotemporal distribution of disturbances.

[0055] The present invention also relates to a power system inertia distribution assessment system, comprising: The data acquisition module is used to collect measurement data of active power and frequency at the grid-connected nodes on the power supply side of the power system, and to filter and clean the measurement data, remove bad data and reconstruct it by interpolation, and then perform normalization processing to form a standard data sequence of active power and a standard data sequence of frequency. The identification module is used to identify the initial moment of system disturbance by combining sliding window technology and detrended fluctuation analysis of active power standard data sequence and frequency standard data sequence. The modeling module is used to construct an analytical model for inertia assessment from an energy perspective based on the power frequency response characteristics of generator sets. The final evaluation module is used to dynamically update the inertia evaluation results through continuous input of real-time measurement data and iterative calculation of the model, and to determine the final inertia evaluation value based on the quasi-steady-state results.

[0056] The present invention also relates to a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the power system inertia distribution assessment method described above.

[0057] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of the application as defined herein, and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Thus, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A method for evaluating the inertia distribution of a power system, characterized in that, include: The active power and frequency measurement data of the grid-connected nodes on the power supply side of the power system are collected, and the measurement data are filtered, cleaned, bad data is removed and interpolated and reconstructed, and normalized to form a standard data sequence of active power and a standard data sequence of frequency. By combining sliding window technology and employing detrended fluctuation analysis of active power standard data sequences and frequency standard data sequences, the initial moment of system disturbance is identified; Based on the power frequency response characteristics of generator sets, an analytical model for inertia assessment is constructed from an energy perspective. By continuously inputting real-time measurement data and iteratively calculating the model, the inertia assessment results are dynamically updated, and the final inertia assessment value is determined based on the quasi-steady-state results.

2. The method for evaluating the inertia distribution of a power system according to claim 1, characterized in that, The specific steps for filtering and cleaning measurement data, removing bad data, interpolating and reconstructing, and normalizing the data include: The active power measurement data are integrated into an active power data sequence in chronological order, and the frequency measurement data are integrated into a frequency data sequence in chronological order. For each data point in the active power data sequence and frequency data sequence, determine its corresponding sliding time window, and calculate the median and absolute deviation of the median of all data within the window. Based on the median and the absolute deviation of the median, it is determined whether each data point in the frequency data sequence and the active power data sequence is an outlier. If the difference between the data point and the corresponding median exceeds a set threshold, it is marked as an outlier data point; otherwise, it is marked as a normal data point. For data points marked as anomalous, linear interpolation is performed using the values ​​of the two nearest normal data points to reconstruct effective alternative data. For all normal data points and reconstructed valid data points in the frequency data sequence and active power data sequence, the difference is calculated and normalized based on the rated frequency and rated power of the unit to complete the data normalization.

3. The method for evaluating the inertia distribution of a power system according to claim 2, characterized in that, The specific algorithm for calculating the median and absolute deviation of the median of all data within this window is as follows: ； ； In the formula, The median of the active power measurement data sequence obtained by the generator set within the nth sliding window; P i This refers to the active power data acquired by the generator set at the i-th sampling time. MED stands for Median of a data sequence. h represents the number of sampling points contained within a single sliding window; P i,res For data P i The absolute deviation relative to the middle value within the nth sliding window; The median of the frequency measurement data sequence obtained by the generator set within the nth sliding window; f i This refers to the frequency data acquired by the generator set at the i-th sampling time. f i,res For data f i The absolute deviation relative to the middle value within the nth sliding window.

4. The method for evaluating the inertia distribution of a power system according to claim 3, characterized in that, The specific algorithm for distinguishing between outlier and normal data points is as follows: ； ； In the formula, and These are respectively the abnormal power data points and the abnormal frequency data points of the generator set; and These are normal power data points and normal frequency data points for the generator set. c is the threshold for judging abnormal data.

5. The method for evaluating the inertia distribution of a power system according to claim 4, characterized in that, The specific algorithm for reconstructing effective replacement data is as follows: ； In the formula, and These are the valid data points obtained after linear interpolation reconstruction of generator power and frequency, respectively. and These are the power anomaly data points. The two nearest preceding and following normal power data points in the time series; and These are the frequency anomaly data points. The two nearest preceding and following normal frequency data points in the time series.

6. The method for evaluating the inertia distribution of a power system according to claim 5, characterized in that, The specific algorithm for per-unit processing is as follows: ； ； In the formula, and These are the dimensionless values ​​corresponding to the changes in generator power and frequency after being normalized to per-unit. S gen This refers to the rated power of the generator set. f s This is the rated frequency.

7. The method for evaluating the inertia distribution of a power system according to claim 1, characterized in that, The specific steps for identifying the initial moment of a system disturbance include: For each data point in the active power standard data sequence and the frequency standard data sequence, based on its own corresponding sliding time window, calculate the median and the absolute deviation of the median of all data in the window, and use the time series composed of the absolute deviations of the median corresponding to each window as a first-time processing signal. Based on the first-processed signals obtained under different windows, a second-order polynomial fitting is performed, and the fitted signal is used as the second-order processed signal. The difference between the primary and secondary processed signals under different windows is calculated to obtain the detrended processed signal; Calculate the root mean square value of the detrended signal corresponding to different sliding windows and use it as the criterion for the timing of system disturbance. When the criterion exceeds a preset threshold, it is determined that a system disturbance has occurred within the sliding window, and the starting time corresponding to the window is determined as the initial time of the disturbance.

8. The method for evaluating the inertia distribution of a power system according to claim 7, characterized in that, The specific algorithm for the power frequency response characteristics of generator sets is as follows: ； In the formula, H gen The inertial time constant of the generator set; This represents the per-unit value of the frequency variation of the generator set; This represents the per-unit value of the change in the mechanical power of the generator set. This represents the per-unit value of the change in active power of the generator set; D gen This is the damping coefficient of the generator set.

9. The method for evaluating the inertia distribution of a power system according to claim 8, characterized in that, The specific algorithm for constructing an analytical model for inertia assessment from an energy perspective is as follows: ； In the formula, The inertia assessment result of the generator set at time t; t0 is the initial time of data sampling; and Let τ be the per-unit value of the change in mechanical power and the change in active power of the generator set at time τ; The frequency per unit value of the generator set at time t; This represents the per-unit frequency value of the generator set at time t0.

10. The method for evaluating the inertia distribution of a power system according to claim 8, characterized in that, The specific algorithm for dynamically updating the inertia assessment results is as follows: ； In the formula, The inertia assessment result of the generator set at time t-1; η is the threshold for judging frequency changes; This is the inertia assessment result of the generator set at the initial moment, which is a preset initial value; The trapezoidal numerical integration method is used to solve the problem, and the specific algorithm is as follows: ； In the formula, M is the number of active power samples collected by the generator set during the time period from the initial time t0 to time t. This refers to the Mth active power change data collected by the generator set during the time period from time t0 to t.

11. A power system inertia distribution assessment system, characterized in that, include: The data acquisition module is used to collect measurement data of active power and frequency at the grid-connected nodes on the power supply side of the power system, and to filter and clean the measurement data, remove bad data and reconstruct it by interpolation, and then perform normalization processing to form a standard data sequence of active power and a standard data sequence of frequency. The identification module is used to identify the initial moment of system disturbance by combining sliding window technology and detrended fluctuation analysis of active power standard data sequence and frequency standard data sequence. The modeling module is used to construct an analytical model for inertia assessment from an energy perspective based on the power frequency response characteristics of generator sets. The final evaluation module is used to dynamically update the inertia evaluation results through continuous input of real-time measurement data and iterative calculation of the model, and to determine the final inertia evaluation value based on the quasi-steady-state results.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the power system inertia distribution assessment method as described in any one of claims 1 to 10.

Images