System for optimizing distribution rule of frequency modulation performance parameters of thermal power generating unit

Abstract

The invention discloses a thermal power generating unit frequency modulation performance parameter distribution rule optimization system, and belongs to the technical field of electric power. The system comprises a parameter rule analysis module, a sorting index making module, an algorithm research module, a clearing model construction module and a resource optimization configuration module, and determines an optimal parameter distribution rule by collecting multi-dimensional historical frequency modulation operation data, extracting statistical characteristics and combining a preset distribution type; a thermal power generating unit frequency modulation performance distribution rule is accurately grasped, a multi-dimension core parameter system is constructed, weights are distributed in combination with an analytic hierarchy process, comprehensive sorting indexes are calculated through standardization processing and weighted summation, and a multi-constraint and multi-target clearing model is constructed. The transaction details are determined based on the clearing result, and a thermal power generating unit flexibility transformation incentive mechanism is set, so that the low-efficiency mode of traditional administrative resource allocation is broken, the decisive effect of the market in resource allocation is fully played, the overall frequency modulation capability of a power system is enhanced, and the consumption demand of new energy power is better adapted.

Description

Technical Field

[0001] This invention relates to the field of power technology, and in particular to a system for optimizing the distribution of frequency regulation performance parameters of thermal power units. Background Technology

[0002] As the core force of frequency regulation in the power system, the frequency regulation performance of thermal power units directly affects the safe and stable operation of the power system and the efficiency of renewable energy consumption.

[0003] Currently, the frequency regulation performance evaluation of thermal power units largely relies on single parameters or empirical indicators, lacking in-depth research on the distribution patterns of frequency regulation performance parameters. This results in incomplete and inaccurate performance evaluations, failing to objectively reflect the actual frequency regulation capabilities of the units. In frequency regulation market transactions, existing bidding ranking and clearing mechanisms often prioritize price factors, neglecting the frequency regulation performance of the units and failing to fully consider the differences in characteristics between thermal power units and other types of regulation resources such as energy storage. This can easily lead to unfair market competition and hinder the optimal allocation of frequency regulation service resources. Furthermore, the lack of effective market incentive mechanisms to guide thermal power units in carrying out flexibility upgrades makes it difficult for some units to improve their frequency regulation performance, failing to meet the growing demand for frequency regulation services from the power system and restricting the large-scale consumption of renewable energy. Summary of the Invention

[0004] The purpose of this invention is to provide an optimization system for the distribution law of frequency regulation performance parameters of thermal power units, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a system for optimizing the distribution law of frequency regulation performance parameters of thermal power units, comprising: The parameter pattern analysis module is configured to collect historical frequency regulation operation data of thermal power units, determine the statistical characteristics of each frequency regulation performance parameter in the historical frequency regulation operation data, and analyze the distribution pattern of the frequency regulation performance parameters of thermal power units. The ranking index formulation module is configured to formulate a ranking index for the comprehensive performance of frequency modulation based on the distribution pattern obtained by the parameter pattern analysis module. The algorithm research module is configured to study the performance parameters of energy storage frequency regulation and the algorithm for price ranking. The clearing model construction module is configured to formulate a clearing model for the comprehensive frequency regulation performance index based on the distribution law of the frequency regulation performance parameters of the thermal power unit obtained by the parameter law analysis module. The resource optimization and allocation module is configured to optimize resource allocation through market-based means.

[0006] Furthermore, the parameter pattern analysis module analyzes the distribution pattern of the frequency regulation performance parameters of thermal power units, specifically including: Collect historical frequency regulation operation data of thermal power units. The historical frequency regulation operation data includes frequency regulation response speed, frequency regulation accuracy, frequency regulation capacity, continuous frequency regulation duration, frequency regulation deviation rate, and frequency regulation dead zone. Feature extraction is performed on the historical frequency modulation operation data to determine the statistical characteristics of each frequency modulation performance parameter. The statistical characteristics include mean, variance, standard deviation, maximum value, minimum value, median, and distribution pattern. Based on the statistical characteristics, a parameter distribution model is constructed. The distribution type of each frequency regulation performance parameter is determined by hypothesis testing. The hypothesis testing adopts the KS test method to calculate the distance between the sample data and the cumulative distribution function of the preset distribution type. The preset distribution type includes normal distribution, Weibull distribution, log-normal distribution and gamma distribution. The distribution law of the frequency regulation performance parameters of thermal power units is determined based on the test results.

[0007] Furthermore, the process by which the parameter pattern analysis module constructs a parameter distribution model based on statistical characteristics and determines the distribution type through the KS test includes: Outlier removal is performed on the statistical feature data of each frequency modulation performance parameter obtained by the feature extraction unit, and the data after outlier removal is used as the sample data for the KS test. Preset distribution parameter estimation: For four preset distribution types, namely normal distribution, Weibull distribution, log-normal distribution and gamma distribution, the maximum likelihood estimation method is used to estimate the distribution parameters of the sample data respectively; For the normal distribution, the mean and variance estimates are calculated; for the Weibull distribution, the shape and scale parameters are estimated by iteratively solving the likelihood equation; for the log-normal distribution, the sample data x is converted to y=lnx, and the mean and variance estimates of y are obtained using the normal distribution parameter estimation method, which serve as the parameters of the log-normal distribution; for the gamma distribution, the shape and scale parameters are estimated using the method of moments and the sample mean and variance. The KS test statistic is calculated by constructing a theoretical cumulative distribution function based on the estimated distribution parameters for each preset distribution type, and simultaneously calculating the empirical cumulative distribution function of the sample data. The KS test statistic Dn is then calculated based on the theoretical and empirical cumulative distribution functions.

[0008] Furthermore, after obtaining the KS test statistic Dn, the significance level is set, and the critical value table of the KS test is consulted to obtain the critical value corresponding to the sample size and significance level. If the calculated critical value for the corresponding sample size and significance level is greater than or equal to Dn, the null hypothesis is accepted and the sample data is determined to conform to the preset distribution type; if the calculated critical value for the corresponding sample size and significance level is less than Dn, the null hypothesis is rejected and the sample data is determined to not conform to the preset distribution type. After performing KS tests on the four preset distribution types in sequence, all distribution types that pass the tests are selected. If multiple distribution types pass the tests, the Dn value corresponding to each passing distribution type is calculated, and the distribution type with the smallest Dn value is selected as the optimal distribution type for the frequency modulation performance parameter. Based on the optimal distribution type of each frequency regulation performance parameter, the distribution law of the frequency regulation performance parameters of thermal power units is determined.

[0009] Furthermore, the ranking index formulation module formulates ranking indexes for the overall performance of frequency modulation, specifically including: The core parameter dimensions affecting frequency modulation performance are determined. These core parameter dimensions include response efficiency, adjustment accuracy, capacity support, stability, and economy. The response efficiency dimension corresponds to the frequency modulation response speed, the adjustment accuracy dimension corresponds to the frequency modulation accuracy and frequency modulation deviation rate, the capacity support dimension corresponds to the frequency modulation capacity, the stability dimension corresponds to the continuous frequency modulation duration, and the economy dimension corresponds to the unit cost of frequency modulation. The ranking index value of frequency modulation comprehensive performance is calculated based on the core parameter dimensions.

[0010] Furthermore, the algorithm research module studies the energy storage frequency regulation performance parameters and the price ranking processing algorithm, specifically including: Construct a pricing dataset that includes thermal power units and energy storage units. The pricing dataset includes unit type, unit number, frequency regulation response speed, frequency regulation accuracy, frequency regulation capacity, continuous frequency regulation duration, frequency regulation unit price, and historical frequency regulation service records. The reasonableness assessment threshold for quotations is determined based on the frequency regulation cost model for different types of generating units. The frequency regulation cost model is constructed based on the equipment depreciation cost, operation and maintenance cost, energy consumption cost and labor cost of the generating units. The unit frequency regulation cost of each type of generating unit is calculated through the cost model, and 1.2-1.5 times the unit frequency regulation cost is set as the reasonableness assessment threshold for quotations. Design a multi-objective price ranking algorithm. The algorithm uses the deviation between the price and the reasonableness assessment threshold, the frequency regulation comprehensive performance ranking index value, and the unit type balance coefficient as optimization objectives, and constructs an objective function. The objective function is solved using the particle swarm optimization algorithm to obtain the ranking results of the unit bids.

[0011] Furthermore, the clearing model construction module formulates a clearing model for the comprehensive performance index of frequency modulation, specifically including: The constraints of the clearing model are determined, including system frequency regulation demand constraints, unit frequency regulation capacity constraints, bidding constraints, grid security constraints, and unit operation constraints. With the goal of minimizing the total cost of frequency regulation services, a clearing model is constructed by combining the distribution pattern of frequency regulation performance parameters of thermal power units and the ranking index of comprehensive frequency regulation performance, while the value of the ranking index of comprehensive frequency regulation performance is used as one of the constraints. The clearing model is solved using a linear programming algorithm, and the winning frequency regulation capacity and winning bid price of each unit are calculated using the simplex method to form the frequency regulation clearing result.

[0012] Furthermore, the resource optimization and allocation module employs market-based methods for resource optimization and allocation, specifically including: Establish a trading mechanism for frequency modulation services, and clarify the trading entities, trading products, and trading procedures; Based on the clearing results of the clearing model, the trading volume and price of frequency regulation services for each unit are determined; a flexibility upgrade incentive mechanism is set up, and thermal power units that improve their frequency regulation performance to the preset standard are given a higher trading price and priority trading incentives.

[0013] Furthermore, the ranking index value of FM comprehensive performance is calculated based on the core parameter dimensions, including: The parameters corresponding to the core parameter dimensions of each thermal power unit are classified to determine the indicator types; the indicator types include positive indicators and negative indicators. Calculate the standardized index value for each indicator; Construct a standardized decision matrix based on the standardized indicator values; The entropy value of each indicator is calculated based on the entropy weight method; The weight value of each indicator is calculated based on the entropy value of each indicator; A weighted decision matrix is ​​constructed based on the standardized decision matrix and the weight value of each indicator; Extract the maximum value from each column of the weighted decision matrix to determine the positive ideal solution; extract the minimum value from each column of the weighted decision matrix to determine the negative ideal solution. The Euclidean distances between the thermal power unit and the positive ideal solution and the negative ideal solution are calculated respectively to obtain the first distance and the second distance. The ranking index value of the comprehensive frequency regulation performance of thermal power units is determined based on the first distance and the second distance.

[0014] Furthermore, with the goal of minimizing the total cost of frequency regulation services, a clearing model is constructed by combining the distribution patterns of the frequency regulation performance parameters of the thermal power units and the ranking indicators of comprehensive frequency regulation performance, including: Obtain the training dataset for the clearing model; Predictive features are constructed based on the training dataset, including time-series features, meteorological features, time-period features, and historical frequency regulation demand features. Based on the aforementioned timing characteristics and historical frequency regulation demand characteristics, an ARIMA model is constructed, the optimal order is determined according to the AIC / BIC criterion, and the initial frequency regulation demand prediction value is output. Dynamic weights are assigned to each type of feature in the prediction features. The dynamic weights are optimized by gradient descent. The optimization objective is to minimize the mean square error between the predicted value and the historical actual frequency regulation demand value. The final frequency regulation demand prediction value is output by combining the dynamic weights and the predicted values ​​corresponding to each type of feature. The range of frequency regulation demand fluctuations is determined by fitting a normal distribution to the predicted residuals. Based on the predicted final frequency regulation demand and its fluctuation range, system frequency regulation demand constraints are constructed. Based on the available frequency regulation capacity, a frequency regulation capacity constraint for thermal power units is constructed. A pricing constraint is constructed based on the market's maximum price limit and the marginal cost of thermal power units; Based on node voltage, line power data, and sensitivity coefficients calculated from static power flow, power grid security constraints are constructed. Based on the ranking index of safety regulation rate and frequency regulation comprehensive performance, operational constraints of thermal power units are constructed. A constraint system is constructed based on system frequency regulation demand constraints, thermal power unit frequency regulation capacity constraints, pricing constraints, power grid security constraints, and thermal power unit operation constraints. With the goal of minimizing the total cost of frequency regulation services, an objective function for the clearing model is constructed based on the frequency regulation demand forecast results and the distribution pattern of frequency regulation performance parameters of thermal power units. ; in, This indicates the total cost of FM service. This indicates optimization of the total time period. This represents the scalar value of the frequency regulation capacity of the g-th thermal power unit during time period t. This represents the actual frequency regulation rate of the g-th thermal power unit during time period t. Indicates the duration of adjustment. This represents the ranking index of the comprehensive frequency regulation performance of the g-th thermal power unit; This represents the marginal cost of frequency regulation for the g-th thermal power unit; Indicates the performance excitation factor; This indicates the total number of thermal power units; The training dataset is input into the clearing model sequentially, linearized, and then solved to output the clearing results. When the clearing results meet the requirements, the initial clearing model is obtained. Obtain the test dataset for the clearing model; The initial clearing model is tested using a test dataset based on the clearing model. When the test results meet the requirements, the target clearing model is obtained.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. The parameter pattern analysis module of this invention collects multi-dimensional historical frequency regulation operation data, extracts statistical features, and uses the KS test combined with multiple preset distribution types to determine the optimal distribution pattern of parameters. This breaks through the limitations of traditional parameter analysis, which has a single data dimension and a crude test method. Based on the analysis results, the subsequent formulation of ranking indicators and the construction of clearing models effectively avoid decision-making errors caused by cognitive biases of parameter characteristics, improve the pertinence and rationality of decisions related to the optimization of frequency regulation performance of thermal power units, and accurately grasp the inherent pattern of frequency regulation performance of thermal power units.

[0016] 2. This invention constructs a multi-dimensional core parameter system, combines the analytic hierarchy process (AHP) to allocate weights, and calculates comprehensive ranking indicators through standardization and weighted summation. It designs a multi-objective bidding ranking algorithm, constructs a multi-constraint, multi-objective clearing model, and solves it using a linear programming algorithm. This achieves a scientific process from quantitative evaluation of unit frequency regulation performance to bidding ranking and market clearing, comprehensively covering key influencing factors of frequency regulation performance. It balances price rationality, performance quality, and unit type equilibrium, meeting multiple objectives of system requirements, safe operation, and cost control. This effectively improves the efficiency and fairness of unit selection, ranking, and clearing in the frequency regulation market.

[0017] 3. This invention establishes a sound frequency regulation service market trading mechanism, determines transaction details based on clearing results, and sets up an incentive mechanism for the flexible transformation of thermal power units. This breaks away from the inefficient traditional administrative resource allocation model, fully leveraging the decisive role of the market in resource allocation. By clarifying transaction rules, it ensures fair and orderly transactions, optimizes the allocation of frequency regulation service resources, and effectively stimulates the enthusiasm of thermal power unit operators to carry out flexible transformations through targeted incentive measures. This promotes the continuous improvement of the frequency regulation performance of thermal power units, injects market-driven momentum into the transformation and upgrading of thermal power units, enhances the overall frequency regulation capability of the power system, and better adapts to the consumption needs of new energy power. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the parameter distribution law analysis process of the present invention; Figure 2 A schematic diagram illustrating the process for establishing the comprehensive performance ranking index of this invention; Figure 3 This is a schematic diagram of the bidding sorting and clearing process of the present invention; Figure 4 This is a schematic diagram of the market-based resource optimization allocation process of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1-4 The present invention provides the following technical solutions: A system for optimizing the distribution of frequency regulation performance parameters of thermal power units, including: The parameter pattern analysis module is configured to collect historical frequency regulation operation data of thermal power units, determine the statistical characteristics of each frequency regulation performance parameter in the historical frequency regulation operation data, and analyze the distribution pattern of the frequency regulation performance parameters of thermal power units. The ranking index formulation module is configured to formulate a ranking index for the comprehensive performance of frequency modulation based on the distribution pattern obtained by the parameter pattern analysis module. The algorithm research module is configured to study the performance parameters of energy storage frequency regulation and the algorithm for price ranking. The clearing model construction module is configured to formulate a clearing model for the comprehensive frequency regulation performance index based on the distribution law of the frequency regulation performance parameters of the thermal power unit obtained by the parameter law analysis module. The resource optimization and allocation module is configured to optimize resource allocation through market-based means.

[0021] The parameter pattern analysis module analyzes the distribution pattern of frequency regulation performance parameters of thermal power units, specifically including: Collect historical frequency regulation operation data of thermal power units. The historical frequency regulation operation data includes frequency regulation response speed, frequency regulation accuracy, frequency regulation capacity, continuous frequency regulation duration, frequency regulation deviation rate, and frequency regulation dead zone. Feature extraction is performed on the historical frequency modulation operation data to determine the statistical characteristics of each frequency modulation performance parameter. The statistical characteristics include mean, variance, standard deviation, maximum value, minimum value, median, and distribution pattern. Based on the statistical characteristics, a parameter distribution model is constructed. The distribution type of each frequency regulation performance parameter is determined by hypothesis testing. The hypothesis testing adopts the KS test method to calculate the distance between the sample data and the cumulative distribution function of the preset distribution type. The preset distribution type includes normal distribution, Weibull distribution, log-normal distribution and gamma distribution. The distribution law of the frequency regulation performance parameters of thermal power units is determined based on the test results.

[0022] Outlier removal is performed on the statistical feature data of each frequency modulation performance parameter obtained by the feature extraction unit, and the data after outlier removal is used as the sample data for the KS test. Preset distribution parameter estimation: For four preset distribution types, namely normal distribution, Weibull distribution, log-normal distribution and gamma distribution, the maximum likelihood estimation method is used to estimate the distribution parameters of the sample data respectively; For the normal distribution, the mean and variance estimates are calculated; for the Weibull distribution, the shape and scale parameters are estimated by iteratively solving the likelihood equation; for the log-normal distribution, the sample data x is converted to y=lnx, and the mean and variance estimates of y are obtained using the normal distribution parameter estimation method, which serve as the parameters of the log-normal distribution; for the gamma distribution, the shape and scale parameters are estimated using the method of moments and the sample mean and variance. The KS test statistic is calculated by constructing a theoretical cumulative distribution function based on the estimated distribution parameters for each preset distribution type, and simultaneously calculating the empirical cumulative distribution function of the sample data. The KS test statistic Dn is then calculated based on the theoretical cumulative distribution function and the empirical cumulative distribution function. Set the significance level, consult the KS test critical value table, and obtain the critical values ​​for the corresponding sample size and significance level; If the calculated critical value for the corresponding sample size and significance level is greater than or equal to Dn, the null hypothesis is accepted and the sample data is determined to conform to the preset distribution type; if the calculated critical value for the corresponding sample size and significance level is less than Dn, the null hypothesis is rejected and the sample data is determined to not conform to the preset distribution type. After performing KS tests on the four preset distribution types in sequence, all distribution types that pass the tests are selected. If multiple distribution types pass the tests, the Dn value corresponding to each passing distribution type is calculated, and the distribution type with the smallest Dn value is selected as the optimal distribution type for the frequency modulation performance parameter. Based on the optimal distribution type of each frequency regulation performance parameter, the distribution law of the frequency regulation performance parameters of thermal power units is determined.

[0023] In the above embodiments, the parameter pattern analysis module accurately collects multi-dimensional historical frequency regulation operation data of thermal power units, comprehensively extracts the core statistical characteristics of each parameter, and uses the KS test method combined with multiple preset distribution types to analyze the distribution patterns. The reliability of the sample data is ensured by removing outliers, and then appropriate parameter estimation methods are used for different distribution types. Through layers of screening of statistical calculation and significance test, the optimal distribution type of each parameter is finally determined. This overcomes the limitations of traditional parameter distribution analysis, which has a single data dimension and a coarse test method. It can accurately capture the inherent distribution characteristics of the frequency regulation performance parameters of thermal power units, making the subsequent decision-making of the entire optimization system more scientific and targeted, and effectively improving the accuracy of frequency regulation performance optimization of thermal power units.

[0024] The ranking index setting module sets the ranking indexes for the overall performance of frequency modulation, specifically including: The core parameter dimensions affecting frequency modulation performance are determined. These core parameter dimensions include response efficiency, adjustment accuracy, capacity support, stability, and economy. The response efficiency dimension corresponds to the frequency modulation response speed, the adjustment accuracy dimension corresponds to the frequency modulation accuracy and frequency modulation deviation rate, the capacity support dimension corresponds to the frequency modulation capacity, the stability dimension corresponds to the continuous frequency modulation duration, and the economy dimension corresponds to the unit cost of frequency modulation. The ranking index value of frequency modulation comprehensive performance is calculated based on the core parameter dimensions.

[0025] In the above embodiments, the ranking index formulation module constructs a comprehensive frequency regulation performance ranking index system from five core dimensions: response efficiency, regulation accuracy, capacity support, stability, and economy. It clarifies the specific parameters corresponding to each dimension to ensure that the index covers the key influencing factors of frequency regulation performance. Through hierarchical analysis, weight coefficients are allocated, and through standardization, the dimensional differences of different parameters are eliminated. Finally, the index value is calculated by weighted summation, avoiding the one-sidedness of evaluation by a single parameter. The weight allocation highlights the importance of the core performance dimensions, while the standardization process ensures the fairness and comparability of the index calculation. Compared with traditional ranking indexes, it is more comprehensive and scientific, and can objectively reflect the comprehensive frequency regulation capability of the unit, providing a unified and accurate evaluation standard for subsequent bidding ranking, clearing calculation, etc.

[0026] The algorithm research module studies the energy storage frequency regulation performance parameters and price ranking algorithms, specifically including: Construct a pricing dataset that includes thermal power units and energy storage units. The pricing dataset includes unit type, unit number, frequency regulation response speed, frequency regulation accuracy, frequency regulation capacity, continuous frequency regulation duration, frequency regulation unit price, and historical frequency regulation service records. The reasonableness assessment threshold for quotations is determined based on the frequency regulation cost model for different types of generating units. The frequency regulation cost model is constructed based on the equipment depreciation cost, operation and maintenance cost, energy consumption cost and labor cost of the generating units. The unit frequency regulation cost of each type of generating unit is calculated through the cost model, and 1.2-1.5 times the unit frequency regulation cost is set as the reasonableness assessment threshold for quotations. Design a multi-objective price ranking algorithm. The algorithm uses the deviation between the price and the reasonableness assessment threshold, the frequency regulation comprehensive performance ranking index value, and the unit type balance coefficient as optimization objectives, and constructs an objective function. The objective function is solved using the particle swarm optimization algorithm to obtain the ranking results of the unit bids.

[0027] In the above embodiments, the algorithm research module constructs a bidding dataset containing key information of multiple types of generating units, builds a frequency regulation cost model based on the full cost composition of the generating units, accurately determines the threshold for evaluating the reasonableness of bidding, and provides an objective basis for bidding selection. At the same time, it designs a multi-objective bidding ranking algorithm with bidding deviation, comprehensive performance index value and generating unit type balance coefficient as objectives, and solves it efficiently through particle swarm optimization algorithm. It breaks through the limitations of traditional bidding ranking that only focuses on price or a single performance index, and realizes multi-objective collaborative optimization of price reasonableness, performance quality and generating unit type balance. The ranking result can not only select generating units with reasonable prices and excellent performance, but also take into account the market participation of different types of generating units, providing a scientific ranking basis for the operation of the subsequent clearing model.

[0028] The clearing model construction module formulates a clearing model for the overall performance indicators of frequency regulation, specifically including: The constraints of the clearing model are determined, including system frequency regulation demand constraints, unit frequency regulation capacity constraints, bidding constraints, grid security constraints, and unit operation constraints. With the goal of minimizing the total cost of frequency regulation services, a clearing model is constructed by combining the distribution pattern of frequency regulation performance parameters of thermal power units and the ranking index of comprehensive frequency regulation performance, while the value of the ranking index of comprehensive frequency regulation performance is used as one of the constraints. The clearing model is solved using a linear programming algorithm, and the winning frequency regulation capacity and winning bid price of each unit are calculated using the simplex method to form the frequency regulation clearing result.

[0029] In the above embodiments, the clearing model construction module comprehensively considers various factors such as system frequency regulation requirements, unit self-constraints, market rule constraints, and grid security constraints to construct a frequency regulation clearing model under multiple constraints. The core objective is to minimize the total cost of frequency regulation services. Simultaneously, the frequency regulation comprehensive performance ranking index is incorporated into the constraints to achieve the dual objectives of cost control and performance assurance. The simplex method in linear programming is used to solve the model, ensuring the efficiency of the solution process and the accuracy of the results. Compared with traditional models, this clearing model has more comprehensive constraints and more diverse objectives, meeting the frequency regulation requirements and safe operation requirements of the power grid. It reduces the total cost of frequency regulation services through cost optimization and ensures the quality of frequency regulation services through performance constraints, providing scientific model support for the efficient clearing of the frequency regulation market and promoting the standardized operation of the frequency regulation market.

[0030] The resource optimization and allocation module uses market-based methods to optimize resource allocation, specifically including: Establish a trading mechanism for frequency modulation services, and clarify the trading entities, trading products, and trading procedures; Based on the clearing results of the aforementioned clearing model, the trading volume and price of frequency regulation services for each generating unit are determined; a flexibility upgrade incentive mechanism is established, providing trading price increases and priority trading incentives to thermal power units whose frequency regulation performance improvements reach preset standards. In the above embodiments, the resource optimization and allocation module establishes a sound frequency regulation service market trading mechanism, clarifies the trading entities, types, and processes, and provides a clear operational framework for market-based resource allocation. Based on the results of the clearing model, it determines the trading volume and price of generating units, ensuring the fairness and rationality of the transactions. At the same time, it sets up a flexibility transformation incentive mechanism, guiding thermal power units to carry out flexibility transformation through incentive measures such as price increases and priority trading. This breaks the inefficient model of traditional administrative resource allocation, fully leverages the decisive role of the market in resource allocation, and achieves optimized allocation of frequency regulation service resources. The incentive mechanism stimulates the enthusiasm of thermal power units to improve frequency regulation performance, provides market-based impetus for the transformation and upgrading of thermal power units, and helps improve the overall frequency regulation capability of the power system.

[0031] The ranking index value of frequency modulation comprehensive performance is calculated based on core parameter dimensions, including: The parameters corresponding to the core parameter dimensions of each thermal power unit are classified to determine the indicator types; the indicator types include positive indicators and negative indicators. Calculate the standardized index value for each indicator; Construct a standardized decision matrix based on the standardized indicator values; The entropy value of each indicator is calculated based on the entropy weight method; The weight value of each indicator is calculated based on the entropy value of each indicator; A weighted decision matrix is ​​constructed based on the standardized decision matrix and the weight value of each indicator; Extract the maximum value from each column of the weighted decision matrix to determine the positive ideal solution; extract the minimum value from each column of the weighted decision matrix to determine the negative ideal solution. The Euclidean distances between the thermal power unit and the positive ideal solution and the negative ideal solution are calculated respectively to obtain the first distance and the second distance. The ranking index value of the comprehensive frequency regulation performance of thermal power units is determined based on the first distance and the second distance.

[0032] In this embodiment, positive indicators include frequency modulation response speed, frequency modulation capacity, and continuous frequency modulation duration; negative indicators include frequency modulation accuracy, frequency modulation deviation rate, and frequency modulation unit cost.

[0033] In this embodiment, the standardized index value corresponding to each index is calculated, including: Positive standardized index value: ;in, Indicates the first The first thermal power unit The standardized values ​​of each indicator; Indicates the first The first thermal power unit The original values ​​of each indicator; Indicates the first The global minimum value of each indicator among all thermal power units participating in the evaluation; Indicates the first The global maximum value of each indicator among all thermal power units participating in the evaluation; Negative standardized index value: .

[0034] In this embodiment, the entropy value of each index is calculated based on the entropy weight method, including: ; in, Indicates the first The entropy value of each indicator; This indicates the total number of thermal power units participating in the evaluation.

[0035] In this embodiment, the weight value of each indicator is calculated based on the entropy value of each indicator, including: ; in, Indicates the first The weight of each indicator; This indicates the total number of indicators.

[0036] In this embodiment, a weighted decision matrix is ​​constructed based on the standardized decision matrix and the weight value of each indicator. That is, the contribution of each indicator to the overall performance of frequency modulation is quantified by weight × standardized value to obtain the weighted decision matrix.

[0037] In this embodiment, the first distance includes: ; in, This represents the j-th positive ideal solution; This represents the product of the standardized value of the j-th indicator of the i-th thermal power unit and the weight of that indicator; Indicates the first distance.

[0038] In this embodiment, the second distance includes: ; in, Indicates the second distance; Let j represent j negative ideal solutions.

[0039] In this embodiment, the ranking index values ​​for overall frequency modulation performance include: ; in, Indicates the first Ranking index values ​​of the comprehensive frequency regulation performance of each thermal power unit.

[0040] The working principle and beneficial effects of the above technical solution are as follows: The core parameters of thermal power units are divided into positive and negative indicators; a larger positive indicator value indicates better performance, and a smaller negative indicator value indicates better performance. This classification method can more accurately reflect the influence direction of different parameters on the frequency regulation performance of thermal power units, making subsequent analysis and evaluation more consistent with reality. The standardized indicator value corresponding to each indicator is calculated and a standardized decision matrix is ​​constructed, eliminating differences in dimensions and magnitudes between different indicators. For example, different indicators may have different units (such as power, time, etc.) and numerical ranges; standardization allows these indicators to be compared and analyzed on the same scale. This method improves the accuracy and fairness of the evaluation; it calculates the entropy and weight values ​​of each indicator based on the entropy weight method; it determines the positive and negative ideal solutions by extracting the maximum and minimum values ​​of each column in the weighted decision matrix; the positive ideal solution represents the optimal state of all indicators, while the negative ideal solution represents the worst state of all indicators; this method provides a clear benchmark for each thermal power unit, enabling a comprehensive measurement of the frequency regulation performance of the thermal power unit; it determines the ranking index value of the frequency regulation performance of the thermal power unit based on the first and second distances; this index value can quantitatively evaluate the frequency regulation performance of the thermal power unit and provide a clear basis for ranking the thermal power unit.

[0041] With the goal of minimizing the total cost of frequency regulation services, and combining the distribution patterns of frequency regulation performance parameters of thermal power units with the ranking index of comprehensive frequency regulation performance, a clearing model is constructed, including: Obtain the training dataset for the clearing model; Predictive features are constructed based on the training dataset, including time-series features, meteorological features, time-period features, and historical frequency regulation demand features. Based on the aforementioned timing characteristics and historical frequency regulation demand characteristics, an ARIMA model is constructed, the optimal order is determined according to the AIC / BIC criterion, and the initial frequency regulation demand prediction value is output. Dynamic weights are assigned to each type of feature in the prediction features. The dynamic weights are optimized by gradient descent. The optimization objective is to minimize the mean square error between the predicted value and the historical actual frequency regulation demand value. The final frequency regulation demand prediction value is output by combining the dynamic weights and the predicted values ​​corresponding to each type of feature. The range of frequency regulation demand fluctuations is determined by fitting a normal distribution to the predicted residuals. Based on the predicted final frequency regulation demand and its fluctuation range, system frequency regulation demand constraints are constructed. Based on the available frequency regulation capacity, a frequency regulation capacity constraint for thermal power units is constructed. A pricing constraint is constructed based on the market's maximum price limit and the marginal cost of thermal power units; Based on node voltage, line power data, and sensitivity coefficients calculated from static power flow, power grid security constraints are constructed. Based on the ranking index of safety regulation rate and frequency regulation comprehensive performance, operational constraints of thermal power units are constructed. A constraint system is constructed based on system frequency regulation demand constraints, thermal power unit frequency regulation capacity constraints, pricing constraints, power grid security constraints, and thermal power unit operation constraints. With the goal of minimizing the total cost of frequency regulation services, an objective function for the clearing model is constructed based on the frequency regulation demand forecast results and the distribution pattern of frequency regulation performance parameters of thermal power units. ; in, This indicates the total cost of FM service. This indicates optimization of the total time period. This represents the scalar value of the frequency regulation capacity of the g-th thermal power unit during time period t. This represents the actual frequency regulation rate of the g-th thermal power unit during time period t. Indicates the duration of adjustment. This represents the ranking index of the comprehensive frequency regulation performance of the g-th thermal power unit; This represents the marginal cost of frequency regulation for the g-th thermal power unit; Indicates the performance excitation factor; This indicates the total number of thermal power units; The training dataset is input into the clearing model sequentially, linearized, and then solved to output the clearing results. When the clearing results meet the requirements, the initial clearing model is obtained. Obtain the test dataset for the clearing model; The initial clearing model is tested using a test dataset based on the clearing model. When the test results meet the requirements, the target clearing model is obtained.

[0042] In this embodiment, an ARIMA model is constructed based on the time-series characteristics and historical frequency regulation demand characteristics. The optimal order is determined through the AIC / BIC criterion, and the initial frequency regulation demand prediction value is output, including: To address the timing-dependent nature of frequency modulation (FM) demand, an ARIMA(p,d,q) model is constructed. The optimal order (p=2, d=1, q=2, verified by residual white noise) is determined using the AIC / BIC criteria, and the initial predicted FM demand value is output.

[0043] In this embodiment, the final frequency modulation demand prediction value is output by combining dynamic weights with the predicted values ​​corresponding to various features, including: ; in, The weight represents the predicted sub-result corresponding to the i-th feature class; Optimization is performed using gradient descent; the objective is to minimize the mean squared error between the predicted value and the historical actual value. This represents the final forecast value for frequency regulation demand.

[0044] In this embodiment, determining the frequency regulation demand fluctuation range based on a normal distribution fitted to the predicted residuals includes: based on the predicted residuals... Fitting a normal distribution The 95% confidence interval is taken as the range of frequency regulation demand fluctuation; This represents the actual frequency regulation demand value during time period t; This represents the mean of the frequency modulation demand forecast residuals; This represents the variance of the predicted residuals.

[0045] In this embodiment, based on the final predicted frequency regulation demand and its fluctuation range, system frequency regulation demand constraints are constructed, including: the sum of the winning bid capacities of all qualified generating units ≥ the maximum frequency regulation demand under a 95% confidence level; the sum of the adjustment rates of all qualified generating units ≥ the product of the frequency regulation demand change rate and the rate safety factor, wherein the rate safety factor is taken as 0.5.

[0046] In this embodiment, based on the available frequency regulation capacity, a frequency regulation capacity constraint for thermal power units is constructed, namely: the winning bid capacity of thermal power units ≤ available frequency regulation capacity, the fluctuation of the winning bid capacity in adjacent time periods ≤ 20% of the available frequency regulation capacity, and the winning bid capacity of thermal power units ≤ the product of the regulation rate and the maximum regulation duration, wherein the maximum regulation duration is 30 minutes.

[0047] In this embodiment, a pricing constraint is constructed based on the market maximum price limit and the marginal cost of thermal power units, namely, the price limit for frequency regulation services of thermal power units is ≤ the market maximum price limit, the price limit for capacity of thermal power units is ≥ 0.8 times the product of the marginal cost of frequency regulation and the capacity lock-in time, and the price limit for adjustment of thermal power units is ≥ 0.8 times the marginal cost of frequency regulation, wherein the capacity lock-in time is 24 hours.

[0048] In this embodiment, based on node voltage, line power data, and sensitivity coefficients calculated from static power flow, power grid safety constraints are constructed, namely: node voltage after frequency regulation output changes is within the range of 0.95-1.05 times the rated voltage; line power flow after frequency regulation output changes is ≤0.9 times the maximum transmission capacity of the line; frequency regulation output changes do not change the direction of line power flow; system power is balanced and frequency deviation is ≤0.2Hz; and node voltage stability margin is ≥0.15.

[0049] In this embodiment, based on the safety regulation rate and the frequency regulation comprehensive performance ranking index, the operating constraints of thermal power units are constructed, namely: the actual regulation rate of thermal power units ≤ the safety regulation rate, the fluctuation of regulation rate between adjacent time periods ≤ 30% of the safety regulation rate, the frequency regulation response delay of thermal power units ≤ 2s, the deviation rate between actual regulation amount and command ≤ 5%, the frequency regulation comprehensive performance ranking index of thermal power units ≥ 0.6, the frequency regulation output of thermal power units ≤ 10% of the main power generation output, and the minimum continuous frequency regulation time of thermal power units ≥ 4 time periods.

[0050] The working principle and beneficial effects of the above technical solution are as follows: By combining time-series features, meteorological features, time-period features, and historical frequency regulation demand features to construct predictive features, it is possible to capture factors affecting frequency regulation demand from multiple dimensions; for example, meteorological conditions affect power load, and power demand varies at different times, while historical data reflects the changing patterns of demand; this comprehensive feature consideration makes the prediction more accurate and reliable; dynamic weights are assigned to each type of feature, and gradient descent optimization minimizes the mean square error between the predicted value and the historical actual frequency regulation demand value; this means that the model can automatically adjust the weights according to the actual contribution of different features to the prediction results, further improving the accuracy of the prediction; based on time-series features and ARIMA model was constructed based on historical frequency regulation demand characteristics, and the optimal order was determined according to the AIC / BIC criterion. A constraint system was built, including system frequency regulation demand constraints, thermal power unit frequency regulation capacity constraints, bidding constraints, grid security constraints, and thermal power unit operation constraints. These constraints cover all aspects of power system operation, ensuring the feasibility and security of the clearing model results in the actual system. The objective function aims to minimize the total cost of frequency regulation services, while also considering the marginal cost of thermal power unit frequency regulation and the ranking index of comprehensive frequency regulation performance. By introducing performance incentive coefficients, better-performing thermal power units are encouraged to participate in frequency regulation services, thereby improving the overall utilization efficiency of frequency regulation resources and reducing the system's frequency regulation cost.

[0051] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A system for optimizing the distribution of frequency regulation performance parameters of thermal power units, characterized in that, include: The parameter pattern analysis module is configured to collect historical frequency regulation operation data of thermal power units, determine the statistical characteristics of each frequency regulation performance parameter in the historical frequency regulation operation data, and analyze the distribution pattern of the frequency regulation performance parameters of thermal power units. The ranking index formulation module is configured to formulate a ranking index for the comprehensive performance of frequency modulation based on the distribution pattern obtained by the parameter pattern analysis module. The algorithm research module is configured to study the performance parameters of energy storage frequency regulation and the algorithm for price ranking. The clearing model construction module is configured to formulate a clearing model for the comprehensive frequency regulation performance index based on the distribution law of the frequency regulation performance parameters of the thermal power unit obtained by the parameter law analysis module. The resource optimization and allocation module is configured to optimize resource allocation through market-based means.

2. The system for optimizing the distribution of frequency regulation performance parameters of thermal power units as described in claim 1, characterized in that, The parameter pattern analysis module analyzes the distribution pattern of frequency regulation performance parameters of thermal power units, specifically including: Collect historical frequency regulation operation data of thermal power units. The historical frequency regulation operation data includes frequency regulation response speed, frequency regulation accuracy, frequency regulation capacity, continuous frequency regulation duration, frequency regulation deviation rate, and frequency regulation dead zone. Feature extraction is performed on the historical frequency modulation operation data to determine the statistical characteristics of each frequency modulation performance parameter. The statistical characteristics include mean, variance, standard deviation, maximum value, minimum value, median, and distribution pattern. Based on the statistical characteristics, a parameter distribution model is constructed. The distribution type of each frequency regulation performance parameter is determined by hypothesis testing. The hypothesis testing adopts the KS test method to calculate the distance between the sample data and the cumulative distribution function of the preset distribution type. The preset distribution type includes normal distribution, Weibull distribution, log-normal distribution and gamma distribution. The distribution law of the frequency regulation performance parameters of thermal power units is determined based on the test results.

3. The system for optimizing the distribution of frequency regulation performance parameters of thermal power units as described in claim 2, characterized in that, The process by which the parameter regularity analysis module constructs a parameter distribution model based on statistical characteristics and determines the distribution type through the KS test includes: Outlier removal is performed on the statistical feature data of each frequency modulation performance parameter obtained by the feature extraction unit, and the data after outlier removal is used as the sample data for the KS test. Preset distribution parameter estimation: For four preset distribution types, namely normal distribution, Weibull distribution, log-normal distribution and gamma distribution, the maximum likelihood estimation method is used to estimate the distribution parameters of the sample data respectively; For the normal distribution, the mean and variance estimates are calculated; for the Weibull distribution, the shape and scale parameters are estimated by iteratively solving the likelihood equation; for the log-normal distribution, the sample data x is converted to y=lnx, and the mean and variance estimates of y are obtained using the normal distribution parameter estimation method, which serve as the parameters of the log-normal distribution; for the gamma distribution, the shape and scale parameters are estimated using the method of moments and the sample mean and variance. The KS test statistic is calculated by constructing a theoretical cumulative distribution function based on the estimated distribution parameters for each preset distribution type, and simultaneously calculating the empirical cumulative distribution function of the sample data. The KS test statistic Dn is then calculated based on the theoretical and empirical cumulative distribution functions.

4. The system for optimizing the distribution law of frequency regulation performance parameters of thermal power units as described in claim 3, characterized in that, After obtaining the KS test statistic Dn, set the significance level, and look up the KS test critical value table to obtain the critical value corresponding to the sample size and significance level. If the calculated critical value for the corresponding sample size and significance level is greater than or equal to Dn, the null hypothesis is accepted and the sample data is determined to conform to the preset distribution type; if the calculated critical value for the corresponding sample size and significance level is less than Dn, the null hypothesis is rejected and the sample data is determined to not conform to the preset distribution type. After performing KS tests on the four preset distribution types in sequence, all distribution types that pass the tests are selected. If multiple distribution types pass the tests, the Dn value corresponding to each passing distribution type is calculated, and the distribution type with the smallest Dn value is selected as the optimal distribution type for the frequency modulation performance parameter. Based on the optimal distribution type of each frequency regulation performance parameter, the distribution law of the frequency regulation performance parameters of thermal power units is determined.

5. The system for optimizing the distribution of frequency regulation performance parameters of thermal power units as described in claim 1, characterized in that, The ranking index setting module sets the overall performance ranking index for frequency modulation, specifically including: The core parameter dimensions affecting frequency modulation performance are determined. These core parameter dimensions include response efficiency, adjustment accuracy, capacity support, stability, and economy. The response efficiency dimension corresponds to the frequency modulation response speed, the adjustment accuracy dimension corresponds to the frequency modulation accuracy and frequency modulation deviation rate, the capacity support dimension corresponds to the frequency modulation capacity, the stability dimension corresponds to the continuous frequency modulation duration, and the economy dimension corresponds to the unit cost of frequency modulation. The ranking index value of frequency modulation comprehensive performance is calculated based on the core parameter dimensions.

6. The system for optimizing the distribution law of frequency regulation performance parameters of thermal power units as described in claim 1, characterized in that, The algorithm research module studies the energy storage frequency regulation performance parameters and price ranking processing algorithm, specifically including: Construct a pricing dataset that includes thermal power units and energy storage units. The pricing dataset includes unit type, unit number, frequency regulation response speed, frequency regulation accuracy, frequency regulation capacity, continuous frequency regulation duration, frequency regulation unit price, and historical frequency regulation service records. The reasonableness assessment threshold for quotations is determined based on the frequency regulation cost model for different types of generating units. The frequency regulation cost model is constructed based on the equipment depreciation cost, operation and maintenance cost, energy consumption cost and labor cost of the generating units. The unit frequency regulation cost of each type of generating unit is calculated through the cost model, and 1.2-1.5 times the unit frequency regulation cost is set as the reasonableness assessment threshold for quotations. Design a multi-objective price ranking algorithm. The algorithm uses the deviation between the price and the reasonableness assessment threshold, the frequency regulation comprehensive performance ranking index value, and the unit type balance coefficient as optimization objectives, and constructs an objective function. The objective function is solved using the particle swarm optimization algorithm to obtain the ranking results of the unit bids.

7. The system for optimizing the distribution of frequency regulation performance parameters of thermal power units as described in claim 1, characterized in that, The clearing model construction module formulates a clearing model for the comprehensive performance indicators of frequency modulation, specifically including: The constraints of the clearing model are determined, including system frequency regulation demand constraints, unit frequency regulation capacity constraints, bidding constraints, grid security constraints, and unit operation constraints. With the goal of minimizing the total cost of frequency regulation services, a clearing model is constructed by combining the distribution pattern of frequency regulation performance parameters of thermal power units and the ranking index of comprehensive frequency regulation performance, while the value of the ranking index of comprehensive frequency regulation performance is used as one of the constraints. The clearing model is solved using a linear programming algorithm, and the winning frequency regulation capacity and winning bid price of each unit are calculated using the simplex method to form the frequency regulation clearing result.

8. The system for optimizing the distribution of frequency regulation performance parameters of thermal power units as described in claim 1, characterized in that, The resource optimization and allocation module employs market-based methods for resource optimization and allocation, specifically including: Establish a trading mechanism for frequency modulation services, and clarify the trading entities, trading products, and trading procedures; Based on the clearing results of the clearing model, the trading volume and price of frequency regulation services for each unit are determined; a flexibility upgrade incentive mechanism is set up, and thermal power units that improve frequency regulation performance to the preset standard are given a higher trading price and priority trading incentives.

9. The system for optimizing the distribution law of frequency regulation performance parameters of thermal power units as described in claim 5, characterized in that, The ranking index value of frequency modulation comprehensive performance is calculated based on core parameter dimensions, including: The parameters corresponding to the core parameter dimensions of each thermal power unit are classified to determine the indicator types; the indicator types include positive indicators and negative indicators. Calculate the standardized index value for each indicator; Construct a standardized decision matrix based on the standardized indicator values; The entropy value of each indicator is calculated based on the entropy weight method; The weight value of each indicator is calculated based on the entropy value of each indicator; A weighted decision matrix is ​​constructed based on the standardized decision matrix and the weight value of each indicator; Extract the maximum value from each column of the weighted decision matrix to determine the positive ideal solution; extract the minimum value from each column of the weighted decision matrix to determine the negative ideal solution. The Euclidean distances between the thermal power unit and the positive ideal solution and the negative ideal solution are calculated respectively to obtain the first distance and the second distance. The ranking index value of the comprehensive frequency regulation performance of thermal power units is determined based on the first distance and the second distance.

10. The system for optimizing the distribution law of frequency regulation performance parameters of thermal power units as described in claim 7, characterized in that, With the goal of minimizing the total cost of frequency regulation services, and combining the distribution patterns of frequency regulation performance parameters of thermal power units with the ranking index of comprehensive frequency regulation performance, a clearing model is constructed, including: Obtain the training dataset for the clearing model; Predictive features are constructed based on the training dataset, including time-series features, meteorological features, time-period features, and historical frequency regulation demand features. Based on the aforementioned timing characteristics and historical frequency regulation demand characteristics, an ARIMA model is constructed, the optimal order is determined according to the AIC / BIC criterion, and the initial frequency regulation demand prediction value is output. Dynamic weights are assigned to each type of feature in the prediction features. The dynamic weights are optimized by gradient descent. The optimization objective is to minimize the mean square error between the predicted value and the historical actual frequency regulation demand value. The final frequency regulation demand prediction value is output by combining the dynamic weights and the predicted values ​​corresponding to each type of feature. The range of frequency regulation demand fluctuations is determined by fitting a normal distribution to the predicted residuals. Based on the predicted final frequency regulation demand and its fluctuation range, system frequency regulation demand constraints are constructed. Based on the available frequency regulation capacity, a frequency regulation capacity constraint for thermal power units is constructed. A pricing constraint is constructed based on the market's maximum price limit and the marginal cost of thermal power units; Based on node voltage, line power data, and sensitivity coefficients calculated from static power flow, power grid security constraints are constructed. Based on the ranking index of safety regulation rate and frequency regulation comprehensive performance, operational constraints of thermal power units are constructed. A constraint system is constructed based on system frequency regulation demand constraints, thermal power unit frequency regulation capacity constraints, pricing constraints, power grid security constraints, and thermal power unit operation constraints. With the goal of minimizing the total cost of frequency regulation services, an objective function for the clearing model is constructed based on the frequency regulation demand forecast results and the distribution pattern of frequency regulation performance parameters of thermal power units. ； in, This indicates the total cost of FM service. This indicates optimization of the total time period. This represents the scalar value of the frequency regulation capacity of the g-th thermal power unit during time period t. This represents the actual frequency regulation rate of the g-th thermal power unit during time period t. Indicates the duration of adjustment. This represents the ranking index of the comprehensive frequency regulation performance of the g-th thermal power unit; This represents the marginal cost of frequency regulation for the g-th thermal power unit; Indicates the performance excitation coefficient; This indicates the total number of thermal power units; The training dataset is input into the clearing model sequentially, linearized, and then solved to output the clearing results. When the clearing results meet the requirements, the initial clearing model is obtained. Obtain the test dataset for the clearing model; The initial clearing model is tested using a test dataset based on the clearing model. When the test results meet the requirements, the target clearing model is obtained.

Images