TOPSIS-based peak regulation and frequency modulation performance evaluation method

Abstract

The invention discloses a peak regulation and frequency modulation performance evaluation method based on TOPSIS. The method comprises the following steps: obtaining real-time frequency modulation parameter sub-indexes for data preprocessing and synchronization; determining a baseline load through a long-short term neural network; sorting the comprehensive frequency modulation performance of the frequency modulation parameter sub-indexes of the different frequency modulation resources, wherein the comprehensive frequency modulation performance is determined by calculating the distances between the optimal values and the worst values of the frequency modulation resources and the frequency modulation parameter sub-indexes; the information of original data can be fully utilized, and the difference between evaluation schemes can be accurately reflected; factors with large influence such as errors and disturbance in the frequency modulation process can be described, related parameters of different conditions before and after adjustment in the frequency modulation process can be described, and the frequency modulation performance of each index can be evaluated from different angles; according to the characteristics of load fluctuation and uncertainty, a baseline load obtained through LSTM prediction is used as a reference value for evaluating the power regulation quantity, and frequency modulation evaluation under fluctuation is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of peak shaving and frequency modulation performance evaluation, and in particular to a peak shaving and frequency modulation performance evaluation method based on TOPSIS. Background Art

[0002] Different from the evaluation of the quality of ancillary services such as peak shaving and frequency modulation on the power generation side, the number of charging piles of EVA is huge. Whether it is ancillary services or demand-side response, there are minimum requirements for service quality. Currently, the main evaluation indicators mainly use the primary frequency modulation contribution power as the main performance evaluation indicator, and a method for assessing and compensating the primary frequency modulation ability of units is proposed. This contribution power belongs to an integral quantity and has strong robustness to signal noise and the error of the integral time starting point. The classical primary frequency modulation indicators use the frequency modulation dead zone and speed regulation inequality rate as the monitoring objects, but the corresponding calculation methods are easily restricted by frequency and power signals with relatively low signal-to-noise ratios. The above-mentioned parameter indicators are basically static indicators based on the steady-state operation of the unit, and cannot reflect the rapid support ability of the unit after power grid disturbances or accidents, and this ability is more critical to the dynamic stability of the power grid. The above methods cannot comprehensively evaluate the performance evaluation and verification of EVA's participation in ancillary services from multiple dimensions such as actual adjustment amount, adjustment rate, response time, and duration, it is difficult to reflect the emphasis on a certain performance parameter when the system changes, and cannot accurately reflect the gap between each evaluation scheme.

[0003] For example, the "Comprehensive Evaluation Method and System for Energy Storage Power Station for Power Grid Peak Shaving and Frequency Modulation" disclosed in the Chinese patent literature, with the publication number: CN112330092A, discloses calculating the values of each index in the evaluation index system based on the operation data of the energy storage power station and a pre-constructed multi-level evaluation index system for the operation performance of the energy storage power station; calculating the weights of each index of the multi-level evaluation index system for the operation performance of the energy storage power station according to the values of each index; obtaining the regulation performance of the energy storage power station based on the weights and index values corresponding to each index in the hierarchical evaluation index system; the multi-level evaluation index system for the operation performance of the energy storage power station is constructed based on the characteristics of peak shaving and frequency modulation for the energy storage power station, but the simple weighting of multiple indicators of the frequency modulation comprehensive parameters in this scheme is a simple and crude processing method, and it is difficult to reflect the emphasis on a certain performance parameter when the system changes. Summary of the Invention

[0004] In order to solve the problem that it is difficult to reflect the emphasis on a certain performance parameter when the system changes in the prior art, the present invention provides a peak shaving and frequency modulation performance evaluation method based on TOPSIS, which can accurately reflect the gap between each evaluation scheme and rank the comprehensive frequency modulation performance of each frequency modulation resource under different frequency modulation indicators in different aspects.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] A peaking and frequency modulation performance evaluation method based on TOPSIS, comprising the following steps:

[0007] Obtain sub-indices of real-time frequency modulation parameters for data preprocessing and synchronization;

[0008] Determine the baseline load through a long short-term neural network;

[0009] Rank the comprehensive frequency modulation performance of sub-indices of frequency modulation parameters of different frequency modulation resources, including determining the comprehensive frequency modulation performance by calculating the distances between the frequency modulation resources and the optimal and worst values of the sub-indices of frequency modulation parameters. It can synthesize the demands of frequency modulation in different time periods, rank the frequency modulation performance of parameters focusing on different aspects, make full use of the information of the original data, and accurately reflect the gaps between different evaluation schemes.

[0010] Preferably, the ranking includes establishing a positive matrix of frequency modulation sub-indices and standardizing it; scoring and normalizing according to the standardized matrix of frequency modulation sub-indices; ranking the normalized scores, and taking the increasing trend of the scores as the direction approaching the optimal solution. In the positive matrix X, the objects to be evaluated are denoted as n, and the standardized matrix is denoted as Z, and each element in it is equal to the value of the corresponding element in matrix X divided by the square root of the sum of the squares of the elements in the column. By normalization and standardization, the influence of dimensions is eliminated, making the results easier to compare.

[0011] Preferably, in the scoring and normalizing, the set of the maximum values of each column element in the standardized matrix is taken as the optimal value, and the set of the minimum values of each column element in the standardized matrix is taken as the worst value. It can determine the positions of the optimal and worst values in the data.

[0012] Preferably, determining the comprehensive frequency modulation performance includes determining performance indicators. The performance indicators include real-time evaluation indicators and long-term statistical indicators. The performance indicators are weighted and fused to obtain a comprehensive evaluation indicator, and then the frequency modulation performance is evaluated according to the interval where the comprehensive evaluation indicator is located. It can determine the frequency modulation characteristics in different time periods through performance indicators, evaluate the frequency modulation performance from the perspective of single-time performance through real-time evaluation indicators, and evaluate the frequency modulation performance from the perspective of long-term performance through long-term statistical indicators, so as to realize the comprehensive evaluation of frequency modulation performance.

[0013] Preferably, it includes setting output decision values for the sub-indices of frequency modulation parameters, and determining the output judgment values through the real-time sub-indices of frequency modulation parameters and the corresponding parameters of the sub-indices of frequency modulation parameters in the baseline load. The output decision values include 0 and 1; the sub-indices of frequency modulation parameters are used to describe the relevant parameters of different frequency modulation effects in different stages of frequency modulation, can describe the factors with greater influence on errors and disturbances in the frequency modulation process, and can describe the relevant parameters of different conditions before and after regulation in the frequency modulation process. It realizes the evaluation of the frequency modulation performance of each index from different angles.

[0014] Preferably, each performance index includes the sorting results of multiple different frequency modulation parameter sub-indices; among them, the frequency modulation parameter sub-indices in the long-term statistical index adopt the average value within a time period. The effect evaluation of frequency modulation for each frequency modulation parameter sub-index in different time periods is realized.

[0015] Preferably, the data preprocessing and synchronization include: after obtaining signals from different sources, summarizing the power with the same timestamp to obtain the total power, and then verifying the total output data through the data quality code. All signals are obtained through the power-frequency curve under online disturbance to realize the acquisition of relevant data that cannot be directly measured; all data is aligned through the timestamp to ensure data synchronization; the total output data is verified through the data quality code to filter out abnormal data, and the suspicious points or missing points are repaired by fitting or interpolating with adjacent valid points.

[0016] Preferably, determining the baseline load includes training a load prediction model with historical data, obtaining data at different sampling frequencies and making predictions according to the load prediction model, and using the predicted load value as the original power reference value during the evaluation of frequency modulation performance. The rated power with large output fluctuations is obtained, so as to obtain an accurate reference load, and further ensure the accuracy of frequency modulation performance evaluation.

[0017] Preferably, after obtaining the original power reference value, the sorting results are evaluated according to the original power reference value and the actually measured power, and the sorting results are used as the reference value of the correction coefficient. By comparing with the predicted reference load, the difference between the effect of frequency modulation and the ideal value is determined, so as to determine the comprehensive evaluation of different indexes.

[0018] The present invention has the following advantages:

[0019] (1) It can make full use of the information of the original data and accurately reflect the gap between different evaluation schemes; (2) It can describe the factors with greater influence such as errors and disturbances during the frequency modulation process, and can describe the relevant parameters in different situations before and after the adjustment during the frequency modulation process, realizing the evaluation of the frequency modulation performance of each index from different angles; (3) In view of the characteristics of load fluctuation and uncertainty, the baseline load obtained by using the LSTM prediction method is used as the reference value for evaluating the power adjustment amount, improving the frequency modulation evaluation under fluctuations. Description of the Drawings

[0020] The drawings in the following description are only exemplary. For those of ordinary skill in the art, other implementation drawings can be obtained by extending according to the provided drawings without creative efforts.

[0021] Figure 1 It is a schematic diagram of the method steps in the embodiment.

[0022] Figure 2 It is a flowchart for performance evaluation in another embodiment. Detailed implementation manners

[0023] The following specific embodiments illustrate the implementation manners of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present invention.

[0024] As Figure 1 shown, in a preferred embodiment, the present invention discloses a peaking and frequency modulation performance evaluation method based on TOPSIS, including the following steps:

[0025] Obtain sub-indicators of real-time frequency modulation parameters for data preprocessing and synchronization;

[0026] Determine the baseline load through a long short-term neural network;

[0027] Rank the comprehensive frequency modulation performance of sub-indicators of frequency modulation parameters of different frequency modulation resources, including determining the comprehensive frequency modulation performance by calculating the distances between the frequency modulation resources and the optimal value and the worst value of the sub-indicators of frequency modulation parameters. It can synthesize the requirements of frequency modulation at different time periods, rank the frequency modulation performance of parameters focusing on different aspects, make full use of the information of the original data, and accurately reflect the gap between each evaluation scheme.

[0028] When in use, determine the standard through the neural network prediction method to obtain the fluctuation prediction value of the time series, then determine the sub-indicators of frequency modulation of different frequency modulation resources, and determine the influence of each sub-indicator of frequency modulation parameter on the comprehensive frequency modulation performance according to the distances between the sub-indicators of frequency modulation parameters of each frequency modulation resource and the optimal value and the worst value.

[0029] In other embodiments, the sub-indicators of frequency modulation parameters include a response speed indicator, an adjustment amplitude indicator, an adjustment deviation indicator, and a power contribution indicator. And the above indicators are all extremely large indicators where the larger the value, the better the performance. Therefore, the step of normalizing the indicators is omitted.

[0030] Among them, after the grid frequency crosses the frequency modulation dead zone set by the unit, ideally, the output of EVA should change immediately without delay. However, due to factors such as signal transmission, the power change of EVA will lag for a certain time. This lag time is called the primary frequency modulation load response lag time, and this time is used to measure the response speed of EVA frequency modulation to the change of grid frequency. The smaller the load response lag time of frequency modulation, the faster the response speed.

[0031] Among them, the regulation amplitude index is used to identify the effectiveness of the output regulation amplitude in the initial stage of frequency modulation. The regulation amplitude is measured by the output response index. In an ideal situation, when the grid frequency crosses the dead zone, precise adjustment should be made according to the change of EVA active power with frequency to make the theoretical adjustment amount ΔP(t)′ consistent with the actual adjustment amount.

[0032]

[0033] In the formula, K d is the droop coefficient promised by EVA; f t is the grid frequency (Hz) corresponding to the t moment; f N is the system rated frequency; f 死区 is the EVA primary frequency modulation dead zone; P base (t) is the baseline load corresponding to the t moment; P(t) is the actual power of EVA at the t moment; P base (t) is the baseline power at the t moment.

[0034] In the initial stage of frequency modulation, factors such as errors and disturbances have a greater impact on the control system. The controllability of EVA is not strong, and it is difficult to adjust according to the theoretical adjustment amount. In this case, in order to effectively adjust the frequency into the dead zone, it is required that the power adjustment of EVA quickly reaches a certain amplitude.

[0035] The frequency modulation adjustment amplitude index is used to measure the EVA power adjustment amplitude in the initial stage of frequency modulation. It is stipulated that the load adjustment amplitude of EVA within T2 (unit: s) should reach the theoretical adjustment load amount corresponding to the frequency extreme point within this time period by y1 (the value of T2 and y1 can be set according to the specific situation of the power grid). If the adjustment amplitude index β 2 is:

[0036]

[0037] In the formula: P max and ΔP′ max are respectively the maximum actual power and the maximum theoretical adjustment of EVA within T2.

[0038] Among them, if EVA performs non-offset adjustment according to the theoretical adjustment amount, over-regulation or under-regulation problems may occur. To prevent over-regulation or under-regulation from causing secondary disturbances to the grid frequency and ensure that the grid frequency is steadily adjusted into the dead zone, it is required that the EVA power be adjusted according to the change trend of the theoretical adjustment amount, and the adjustment deviation should be kept within a certain range, that is, it is required that the EVA primary frequency modulation has no bias.

[0039] The frequency modulation adjustment deviation index is used to measure the unbiasedness of the unit's frequency modulation. It is stipulated that starting from the time when the grid frequency change exceeds the unit's frequency modulation dead zone until 60 s or until the frequency change returns to the frequency modulation dead zone, the deviation between the actual EVA power and the theoretical target during this period should be within K of the theoretical adjustment load e (K e value can be set according to the specific situation of the power grid). The calculation formula of the frequency modulation adjustment deviation index is as follows:

[0040]

[0041]

[0042] In the formula: P(t i ), P base (t i ) are respectively the maximum values of the actual active power, baseline power and theoretical adjustment amount of EVA at the ti moment during the period from the time when the frequency crosses the frequency modulation dead zone until 60 s or until the frequency returns to the dead zone; n is the number of sampling points, and ε i is the unbiased index within each sampling period. If the absolute value of the deviation exceeds the threshold, the unbiased coefficient is taken as 0, otherwise it is taken as 1.

[0043] Among them, the frequency modulation response index measures the overall effect of EVA frequency modulation by the size of the integrated power. It is stipulated that during the period from the time when the grid frequency change exceeds the unit's primary frequency modulation dead zone for 3 - 60 s or until the frequency change returns to the primary frequency modulation dead zone, the actual weighted integrated power of the unit's primary frequency modulation should reach Q of the theoretically calculated integrated power within the response time e or more (Q e value can be set according to the specific situation of the power grid). If Q rate > Q e , then the primary frequency modulation response

[0044]

[0045] Among them: Fs is the sampling frequency; C P (t) is the primary frequency modulation action switch signal, which outputs 1 when the frequency deviation exceeds the frequency modulation dead zone, and outputs 0 otherwise. is the weighting function, which weights the periods that are more concerned during the primary frequency modulation process, highlighting the frequency modulation effect of the unit during this period. In the formula, the numerator represents the actual weighted adjustment power, and the denominator represents the theoretical weighted adjustment power.

[0046] The frequency modulation contribution power belongs to an integral quantity, has relatively good robustness, and does not require high precision for algorithms and signals. It is often used to measure the contribution of each unit in responding to load disturbances during primary frequency modulation, and can be used as an assessment and compensation for units' participation in the grid's primary frequency modulation after the event to encourage units to actively participate in primary frequency modulation and maintain the stability of the power grid. The frequency modulation contribution power mainly reflects the cumulative contribution of units during the process of participating in grid frequency modulation, and its main role is to maintain the stability of the power grid.

[0047] In other embodiments, the sorting includes establishing a positive matrix of frequency modulation sub-indicators and standardizing it; scoring and normalizing according to the standardized matrix of frequency modulation sub-indicators; sorting the normalized scores, and taking the increasing trend of the scores as the direction approaching the optimal solution. In the positive matrix X, for the objects to be evaluated, denoted as n, and the standardized matrix denoted as Z, each element in it is equal to the value of the corresponding element in matrix X divided by the square root of the sum of the squares of the elements in the corresponding column. By normalization and standardization, the influence of dimensions is eliminated, making the results easier to compare in size.

[0048] When in use, assuming there are n objects to be evaluated and 4 positive evaluation indicators, a positive matrix can be constructed.

[0049]

[0050] Denote the standardized matrix as Z, then each element in it is equal to the value of the corresponding element in matrix X divided by the square root of the sum of the squares of the elements in the corresponding column, that is

[0051] In other embodiments, in the scoring and normalizing, the set of the maximum values of each column of elements in the standardized matrix is taken as the optimal value, and the set of the minimum values of each column of elements in the standardized matrix is taken as the worst value. It can determine the positions of the optimal and worst values in the data.

[0052] When in use, the standardized matrix of n evaluation objects and 4 evaluation indicators is as follows:

[0053]

[0054] Define the maximum value as the set of the maximum values of each column of elements:

[0055] Z + =(max(z 11 ,z 21 ,...,z n1 ),max(z 12 ,z 22 ,...,z n2 ),max(z 12 ,z 22 ,...,z n2), max(z 13 , z 23 ,..., z n3 ), max(z 14 , z 24 ,..., z n4 ))

[0056] Define the minimum value as the set of minimum values of each column element:

[0057] Z - = (min(z 11 , z 21 ,..., z n1 ), max(z 12 , z 22 ,..., z n2 ), max(z 12 , z 22 ,..., z n2 ), max(z 13 , z 23 ,..., z n3 ), max(z 14 , z 24 ,..., z n4 ))

[0058] Then the distance between the i-th evaluation object and the maximum value is the sum after calculating the distances between the j indicators and the maximum value respectively:

[0059]

[0060] Similarly, the distance between the i-th evaluation object and the minimum value is the sum after calculating the distances between the j indicators and the minimum value respectively:

[0061]

[0062] Then, the unnormalized score of the i-th evaluation object is That is, the distance between z and the minimum value divided by the sum of the distance between z and the maximum value and the distance between z and the minimum value. Since the distances are all non-negative, it is obvious that S i takes values between 0 and 1, and D i + The larger it is, the larger S i is, that is, the closer it is to the optimal solution.

[0063] Among them, the normalized score is It should satisfy here

[0064] In other embodiments, the determination of the comprehensive frequency modulation performance includes determining performance indicators, where the performance indicators include real-time evaluation indicators and long-term statistical indicators. The performance indicators are weighted and fused to obtain a comprehensive evaluation indicator, and then the frequency modulation performance is evaluated according to the interval where the comprehensive evaluation indicator is located. The frequency modulation characteristics in different time periods can be determined through the performance indicators, the frequency modulation performance can be evaluated from the perspective of single-time performance through the real-time evaluation indicators, and the frequency modulation performance can be evaluated from the perspective of long-term performance through the long-term statistical indicators, thereby realizing the comprehensive evaluation of the frequency modulation performance.

[0065] During use, the sub-indicators of the frequency modulation parameters under the real-time evaluation indicators and the long-term statistical indicators are sorted respectively to achieve a comprehensive evaluation combining multiple perspectives.

[0066] In other embodiments, it includes setting output determination values for the sub-indicators of the frequency modulation parameters, and determining the output judgment value through the sub-indicators of the real-time frequency modulation parameters and the corresponding sub-indicators of the baseline load. The output determination values include 0 and 1; the sub-indicators of the frequency modulation parameters are used to describe the parameters related to different frequency modulation effects in different stages of frequency modulation, can describe the factors with greater influence such as errors and disturbances during the frequency modulation process, and can describe the relevant parameters in different situations before and after adjustment during the frequency modulation process. It realizes the evaluation of the frequency modulation performance of each indicator from different angles.

[0067] During use, the actual value of the sub-indicator of the frequency modulation parameter is compared with the standard value, and the output determination value is used as the output of the sub-indicator of the frequency modulation parameter.

[0068] In other embodiments, each performance indicator includes the sorting results of multiple different sub-indicators of the frequency modulation parameters; among them, the sub-indicators of the frequency modulation parameters in the long-term statistical indicators adopt the average value within a time period. It realizes the evaluation of the frequency modulation effect of each sub-indicator of the frequency modulation parameter in different time periods.

[0069] During use, the average value of one month is used to replace the sub-indicator of the frequency modulation parameter to achieve offline long-term statistics.

[0070] In other embodiments, the data preprocessing and synchronization include, after obtaining signals from different sources, summarizing the power with the same timestamp to obtain the total power, and then verifying the total output data through the data quality code. All signals are obtained through the power-frequency curve under online disturbances to realize the acquisition of relevant data that cannot be directly measured; all data is aligned through timestamps to ensure data synchronization; abnormal data filtering is realized by verifying the total output data through the data quality code, and suspicious points or missing points are repaired by fitting or interpolating with adjacent valid points.

[0071] In other embodiments, obtaining signals from different sources includes obtaining online monitoring data of relevant data of sub-indices of frequency modulation parameters, and performing discrimination analysis and parameter estimation on the online monitoring data to obtain raw data.

[0072] Since the operation data comes from the signals of each charging pile under the EVA platform collected in real time during the operation process, the sensors used for these signals have different accuracies, and noise and delays are inevitably introduced during the measurement and transmission processes. Before summarizing the power at the same timestamp, the power signals of each charging pile are synchronized to a unified timestamp and uploaded at the same sampling frequency.

[0073] In other embodiments, setting an effective power grid frequency disturbance means that the frequency deviation exceeds the dead zone of primary frequency modulation, the duration is greater than the set time threshold, and the maximum frequency deviation during the disturbance period exceeds the set frequency deviation threshold.

[0074] In other embodiments, the determination of the baseline load includes training a load prediction model using historical data, obtaining data at different sampling frequencies and making predictions according to the load prediction model, and using the predicted load value as the raw power reference value during the evaluation of frequency modulation performance. This realizes the acquisition of the rated power with large output fluctuations, thereby obtaining an accurate reference load, and further ensuring the accuracy of the frequency modulation performance evaluation.

[0075] In other embodiments, since EVA is different from conventional units and its output is usually smooth. When evaluating the frequency modulation performance of conventional units, the rated output of the unit is generally used as a reference. However, the concept of rated power is no longer applicable to EVA with a large number of charging piles and large output fluctuations. Compared with the rated power, the baseline load is more the raw power reference value for evaluating the frequency modulation performance of EVA. Therefore, baseline load predictions are made for the day-ahead, hour-ahead, and 15-minute-ahead.

[0076] Using the LSTM prediction algorithm, the historical data of EVA in the previous month is used to train the LSTM load prediction model of EVA, and the data in the most recent 10 days is used for day-ahead load prediction, and the predicted load value is used as the baseline load.

[0077] In other embodiments, after obtaining the raw power reference value, the sorting result is evaluated according to the raw power reference value and the actually measured power, and the sorting result is used as the correction coefficient reference value. By comparing with the predicted reference load, the difference between the frequency modulation effect and the ideal value is determined, so as to determine the comprehensive evaluation of different indicators.

[0078] In other embodiments, such as Figure 2As shown, it includes the following steps. First, determine the benchmark load according to the LSTM prediction method, and at this time, use the historical data within the previous month; then perform real-time data collection; after data collection, preprocess the data, and perform frequency regulation based on the preprocessed data. The data collection includes the data during frequency disturbance before and after the frequency regulation process and the data at the end of the frequency disturbance; after frequency regulation, use the TOPSIS method to comprehensively rank the multiple frequency modulation parameter sub-indicators of different frequency modulation resources for frequency modulation performance.

[0079] Although the present invention has been described in detail with general descriptions and specific embodiments above, based on the present invention, some modifications or improvements can be made, which are obvious to those skilled in the art. Therefore, these modifications or improvements made without departing from the spirit of the present invention all fall within the scope of protection required by the present invention.

Claims

1. A peak-shaving and frequency-modulation performance evaluation method based on TOPSIS, characterized in that: The steps include: Obtain real-time frequency modulation parameter sub-indicators for data preprocessing and synchronization; Determination of baseline load through long-term and short-term neural networks; The comprehensive frequency modulation performance of frequency modulation parameter sub-indicators of different frequency modulation resources are ranked, including determining the comprehensive frequency modulation performance by calculating the distance between the frequency modulation resources and the optimal value and the worst value of the frequency modulation parameter sub-indicators.

2. The peak-shaving and frequency-modulation performance evaluation method based on TOPSIS according to claim 1, characterized in that: The sorting includes establishing a forward matrix of the frequency modulation index and standardizing it; scoring and normalizing according to the standardized matrix of the frequency modulation index; sorting the normalized scores, and taking the trend of increasing scores as the direction of approaching the optimal solution.

3. The peak-shaving and frequency-modulation performance evaluation method based on TOPSIS according to claim 2 is characterized in that: In the scoring and normalization, the set of maximum values ​​of the elements in each column of the standardized matrix is ​​taken as the optimal value, and the set of minimum values ​​of the elements in each column of the standardized matrix is ​​taken as the worst value.

4. A peak-shaving and frequency-modulation performance evaluation method based on TOPSIS according to claim 1, 2 or 3, characterized in that: The determination of the comprehensive frequency modulation performance includes determining performance indicators, and the performance indicators include real-time evaluation indicators and long-term statistical indicators. The performance indicators are weighted and integrated to obtain comprehensive evaluation indicators, and then the frequency modulation performance is evaluated according to the interval where the comprehensive evaluation indicators are located.

5. A peak-shaving and frequency-modulation performance evaluation method based on TOPSIS according to claim 1, 2 or 3, characterized in that: The method includes setting an output judgment value for the frequency modulation parameter sub-indicator, and determining the output judgment value through the parameters of the real-time frequency modulation parameter sub-indicator and the corresponding frequency modulation parameter sub-indicator in the baseline load.

6. The peak-shaving and frequency-modulation performance evaluation method based on TOPSIS according to claim 4 is characterized in that: Each performance indicator includes the ranking results of multiple different frequency modulation parameter sub-indicators; the frequency modulation parameter sub-indicator in the long-term statistical indicator adopts the average value within a time period.

7. A peak-shaving and frequency-modulation performance evaluation method based on TOPSIS according to claim 1, 2 or 3, characterized in that: The data preprocessing and synchronization include, after acquiring signals from different sources, aggregating the powers with the same timestamp to obtain the total power, and then verifying the total output data through the data quality code.

8. The peak-shaving and frequency-modulation performance evaluation method based on TOPSIS according to claim 7, characterized in that: The determination of the baseline load includes using historical data to train a load prediction model, using different sampling frequencies to acquire data and make predictions based on the load prediction model, and using the load prediction value as the original power reference value for frequency regulation performance evaluation.

9. The peak-shaving and frequency-modulation performance evaluation method based on TOPSIS according to claim 8, characterized in that: It also includes evaluating the sorting result according to the original power reference value and the actually measured power after obtaining the original power reference value, and using the sorting result as a correction coefficient reference value.

Images