Automatic power generation control performance evaluation method, device, equipment, medium and computer program product

Abstract

The invention relates to an automatic power generation control performance evaluation method and device, equipment, a medium and a computer program product. The method comprises the following steps: constructing an evaluation system for automatic power generation control of a target power system; performing dimension reduction processing on the multi-dimensional indexes in the evaluation system through a principal component analysis algorithm, and extracting a plurality of principal component factors; and establishing an automatic power generation control performance evaluation model according to the principal component factors through a random forest algorithm, and outputting a performance evaluation result of the target power system according to an input data set through the automatic power generation control performance evaluation model. By adopting the method, the accuracy and timeliness of automatic power generation control performance evaluation can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automatic generation control, and in particular to an automatic generation control performance evaluation method, device, equipment, medium and computer program product. BACKGROUND

[0002] The penetration rate of new energy such as wind power and photovoltaic power in the power system continues to rise. However, the strong volatility and low inertia characteristics of new energy result in new problems such as a decline in frequency regulation capability and a lag in dynamic response of the automatic generation control (AGC) system.

[0003] The traditional AGC performance evaluation method based on steady-state indicators cannot accurately quantify the influence of new energy access on the dynamic characteristics of the system, and new energy does not have the rotating inertia of synchronous generators when connected to the grid through power electronic devices, so the equivalent inertia of the system is significantly reduced. A single indicator system cannot effectively evaluate the AGC performance. At the same time, the existing evaluation methods also have obvious limitations. The fuzzy comprehensive evaluation method needs to rely on expert setting of the weight matrix, and when the penetration rate of new energy is more than 40%, the misjudgment rate is high, and the model parameters cannot be corrected in real time to cope with the minute-level fluctuations of wind and light output. These technical defects make it difficult for traditional methods to balance evaluation accuracy, real-time performance and model explainability in the scenario of high-penetration new energy. SUMMARY

[0004] Therefore, it is necessary to provide an automatic generation control performance evaluation method, device, equipment, medium and computer program product capable of improving the accuracy and timeliness of automatic generation control performance evaluation.

[0005] In a first aspect, the present application provides an automatic generation control performance evaluation method, which comprises:

[0006] constructing an evaluation system of automatic generation control of a target power system;

[0007] performing dimensionality reduction processing on multi-dimensional indicators in the evaluation system through a principal component analysis algorithm, and extracting a plurality of principal component factors;

[0008] establishing an automatic generation control performance evaluation model according to the principal component factors through a random forest algorithm, and outputting a performance evaluation result of the target power system according to an input data set through the automatic generation control performance evaluation model.

[0009] In some embodiments of the method, the dimensionality reduction processing on the multi-dimensional indicators in the evaluation system through the principal component analysis algorithm and the extraction of the plurality of principal component factors comprise:

[0010] Perform data standardization processing on the multi-dimensional indexes in the evaluation system to obtain a standardized data matrix;

[0011] Determine a covariance matrix according to the standardized data matrix;

[0012] Perform eigenvalue decomposition on the covariance matrix to determine a plurality of eigenvectors;

[0013] Select a first preset number of the eigenvectors in a sequential order to construct a principal component loading matrix;

[0014] Determine a principal component score matrix according to the standardized data matrix and the principal component loading matrix.

[0015] In some embodiments of the method, the automatic generation control performance evaluation model is established by the random forest algorithm according to the principal component factors, and the performance evaluation result of the target power system is output by the automatic generation control performance evaluation model according to the input data set, which includes:

[0016] The performance level of the automatic generation control is formed into a label vector, and the principal component score matrix and the label vector are used to construct the input data set;

[0017] A second preset number of training subsets are generated by random sampling with replacement on the data set, and a decision tree is constructed for each training subset, and the construction is repeated until a second preset number of decision trees are generated;

[0018] A predicted performance evaluation sub-result is output by each decision tree according to the corresponding internal splitting rule, and the final performance evaluation result of the target power system is determined by a majority voting mechanism according to the performance evaluation sub-results output by all decision trees.

[0019] In some embodiments of the method, the method further includes:

[0020] The total amount of Gini coefficient reduction of the principal component score matrix in all node splits is counted for each decision tree to determine the feature importance score of the corresponding principal component factor;

[0021] The feature importance scores of all principal component factors are normalized to output the normalized feature importance scores of all principal component factors.

[0022] In some embodiments of the method, the multi-dimensional indexes include at least two of the following: dynamic performance indexes, steady-state performance indexes, economic indexes, and new energy characteristic indexes.

[0023] In some embodiments of the method, the dynamic performance indicators include at least one of frequency deviation rate, regulation time constant, and mean regional control error; the steady-state performance indicators include at least one of tie-line power error, frequency qualification rate, and renewable energy frequency fluctuation rate; the economic indicators include at least one of regulation cost, unit wear rate, and energy storage cycle loss rate; and the renewable energy characteristic indicators include at least one of renewable energy penetration rate, wind and solar power output prediction error, and inertial support coefficient.

[0024] According to a second aspect of the present disclosure, an automatic power generation control performance evaluation apparatus is provided, the apparatus comprising:

[0025] The evaluation system construction module is used to construct an evaluation system for the automatic generation control of the target power system.

[0026] The principal component factor extraction module is used to perform dimensionality reduction processing on the multidimensional indicators in the evaluation system through principal component analysis algorithm and extract multiple principal component factors.

[0027] The evaluation result output module establishes an automatic generation control performance evaluation model based on the principal component factors using a random forest algorithm, and outputs the performance evaluation results of the target power system based on the input dataset using the automatic generation control performance evaluation model.

[0028] According to a third aspect of the present disclosure, a computer device is provided. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program performing the following steps:

[0029] Construct an evaluation system for the automatic generation control of the target power system;

[0030] The multidimensional indicators in the evaluation system are reduced in dimensionality using principal component analysis algorithm to extract multiple principal component factors.

[0031] An automatic generation control performance evaluation model is established based on the principal component factors using the random forest algorithm. The automatic generation control performance evaluation model then outputs the performance evaluation results of the target power system based on the input dataset.

[0032] According to a fourth aspect of the present disclosure, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, performs the following steps:

[0033] Construct an evaluation system for the automatic generation control of the target power system;

[0034] The multidimensional indicators in the evaluation system are reduced in dimensionality using principal component analysis algorithm to extract multiple principal component factors.

[0035] An automatic generation control performance evaluation model is established based on the principal component factors using the random forest algorithm. The automatic generation control performance evaluation model then outputs the performance evaluation results of the target power system based on the input dataset.

[0036] According to a fifth aspect of the present disclosure, a computer program product is provided. The computer program product includes a computer program that, when executed by a processor, performs the following steps:

[0037] Construct an evaluation system for the automatic generation control of the target power system;

[0038] The multidimensional indicators in the evaluation system are reduced in dimensionality using principal component analysis algorithm to extract multiple principal component factors.

[0039] An automatic generation control performance evaluation model is established based on the principal component factors using the random forest algorithm. The automatic generation control performance evaluation model then outputs the performance evaluation results of the target power system based on the input dataset.

[0040] The automatic generation control performance evaluation scheme provided in this application breaks through the limitations of traditional single-dimensional evaluation system construction, constructing a multi-level system containing multiple dimensional indicators. It can use principal component analysis algorithm to reduce the dimensionality of the multidimensional indicators in the evaluation system, extract principal component factors, eliminate multicollinearity among indicators, balance the dimensionality reduction effect and information preservation, and solve the model instability problem caused by indicator redundancy in traditional methods. At the same time, it adopts random forest algorithm to capture the nonlinear relationship between principal component factors and AGC performance level by integrating multiple decision trees, providing accurate guidance for optimization and overcoming the defects of insufficient accuracy and lack of dynamic interpretation capability of traditional linear models.

[0041] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0042] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure, and are not intended to unduly limit this disclosure.

[0043] Figure 1 This is a flowchart illustrating an automatic power generation control performance evaluation method according to an exemplary embodiment;

[0044] Figure 2 This is a schematic diagram illustrating the specific process of extracting principal component factors according to an exemplary embodiment;

[0045] Figure 3 Here is a flowchart illustrating the random forest algorithm according to an exemplary embodiment;

[0046] Figure 4 This is a schematic diagram illustrating a specific process of an automatic power generation control performance evaluation method according to an exemplary embodiment.

[0047] Figure 5 This is a schematic diagram of the IEEE 33-node simulation model structure according to an exemplary embodiment;

[0048] Figure 6 A confusion matrix heatmap illustrated according to an exemplary embodiment;

[0049] Figure 7 This is a principal component importance ranking graph illustrated according to an exemplary embodiment;

[0050] Figure 8 This is a schematic diagram illustrating the accuracy variation under different penetration rates according to an exemplary embodiment;

[0051] Figure 9 This is a structural block diagram of an automatic power generation control performance evaluation device according to an exemplary embodiment;

[0052] Figure 10 This is a diagram illustrating the internal structure of a computer device according to an exemplary embodiment. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0054] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure. The terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, product, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, product, or apparatus. Without further limitation, the presence of other identical or equivalent elements in a process, method, product, or apparatus that includes said elements is not excluded. For example, the use of terms such as "first," "second," etc., is to denote names and does not indicate any specific order.

[0055] In some embodiments provided in this disclosure, the execution of the automatic power generation control performance evaluation method can be controlled by a unified controller or by multiple controllers. These controllers may include controllers on local terminals or controllers on remote servers. In some embodiments, the controllers on local terminals and the controllers on servers may work together to implement the automatic power generation control performance evaluation method. The local terminal mentioned in this disclosure may include, but is not limited to, various robotic devices, vehicle-mounted devices, personal computers, laptops, smartphones, tablets, wearable devices, medical devices, VR (Virtual Reality) devices, etc. The server may also be a server, server cluster, distributed subsystem, cloud processing platform, server containing blockchain nodes, or a combination thereof. The controllers described in this disclosure may include various control units capable of implementing logic processing functions, including but not limited to CPU (Central Processing Unit), PLC (Programmable Logic Controller), ECU (Electronic Control Unit), MCU (Microcontroller Unit), FPGA (Field Programmable Gate Array), and CPLD (Complex Programmable Logic Device), as well as controllers composed of one or more logic function units, chips, etc.

[0056] In some embodiments of this disclosure, an automatic power generation control performance evaluation method is provided, such as... Figure 1 As shown, it includes the following steps:

[0057] S20. Construct an evaluation system for the automatic generation control of the target power system.

[0058] In some embodiments of this disclosure, the target power system typically refers to a modern power system with grid-connected renewable energy sources. The target power system may include a modern power system with a high proportion of renewable energy sources (e.g., wind power, photovoltaic power, etc.) connected to the grid. Automatic generation control typically refers to a closed-loop control mode or control system in a power system that maintains system frequency stability and ensures planned power transmission along tie lines by adjusting generator output in real time. The evaluation system typically refers to a multi-level, multi-dimensional indicator evaluation system that quantifies the performance of automatic generation control.

[0059] S22. The multidimensional indicators in the evaluation system are reduced in dimensionality using principal component analysis algorithm to extract multiple principal component factors.

[0060] In some embodiments of this disclosure, Principal Component Analysis (PCA) typically refers to an unsupervised statistical method that projects original high-dimensional data onto a new set of orthogonal variables (principal components) through orthogonal transformations, maximizing the variance of the data and thus achieving data compression and feature extraction. Dimensionality reduction typically refers to converting high-dimensional data into a low-dimensional representation through mathematical transformations while retaining the main information of the data. The aim is to reduce data redundancy, lower computational complexity, and eliminate multicollinearity among features. Principal component factors typically refer to new variables obtained through PCA, which are linear combinations of the original features. Each principal component represents a latent dimension in the data, and the principal components are uncorrelated with each other.

[0061] S24. An automatic generation control performance evaluation model is established based on the principal component factors using the random forest algorithm. The performance evaluation results of the target power system are then output based on the input dataset using the automatic generation control performance evaluation model.

[0062] In some embodiments of this disclosure, the random forest algorithm typically refers to an ensemble learning algorithm composed of multiple decision trees. The random forest algorithm can improve the generalization ability and robustness of the model through the dual randomness of random sample sampling and random feature selection. The automatic generation control performance evaluation model typically refers to a principal component factor mapping model constructed based on the random forest algorithm, which maps the automatic generation control performance level to the principal component factor. The automatic generation control performance evaluation model can be used to quantitatively evaluate the comprehensive performance of AGC in a new energy power system. The performance evaluation result typically refers to the quantitative judgment result of the AGC performance of the target power system output by the automatic generation control performance evaluation model. The performance evaluation result may include the evaluated performance level of the target power system, and may also include the importance results of key factors affecting AGC performance.

[0063] The automatic generation control performance evaluation method provided in this disclosure breaks through the limitations of traditional single-dimensional evaluation system construction, constructing a multi-level system containing multiple dimensional indicators. It utilizes principal component analysis (PCA) to reduce the dimensionality of the multidimensional indicators in the evaluation system, extracting principal component factors, eliminating multicollinearity among indicators, balancing dimensionality reduction effects with information retention, and solving the model instability problem caused by indicator redundancy in traditional methods. Simultaneously, it employs a random forest algorithm, integrating multiple decision trees to capture the nonlinear relationship between principal component factors and AGC performance levels, providing precise guidance for optimization and overcoming the shortcomings of insufficient accuracy and lack of dynamic interpretability in traditional linear models.

[0064] In some embodiments of this disclosure, the multidimensional indicators include at least two of the following: dynamic performance indicators, steady-state performance indicators, economic indicators, and new energy characteristic indicators.

[0065] In some examples, an evaluation system can be constructed that includes multiple dimensions, such as dynamic performance indicators, steady-state performance indicators, economic indicators, and renewable energy characteristic indicators. Dynamic performance indicators reflect the system's ability to recover stability after being disturbed. Steady-state performance indicators measure the stability of the system's long-term operating parameters. Economic indicators assess the cost-effectiveness of AGC control. Renewable energy characteristic indicators are specific indicators designed for the grid connection characteristics of renewable energy sources. By adopting this approach, dynamic performance indicators, steady-state performance indicators, economic indicators, and renewable energy characteristic indicators can be analyzed as multiple dimensions within the evaluation system, enabling precise control of the system.

[0066] In some embodiments of this disclosure, the dynamic performance indicators include at least one of frequency deviation rate, regulation time constant, and mean regional control error; the steady-state performance indicators include at least one of tie-line power error, frequency qualification rate, and renewable energy frequency fluctuation rate; the economic indicators include at least one of regulation cost, unit wear rate, and energy storage cycle loss rate; and the renewable energy characteristic indicators include at least one of renewable energy penetration rate, wind and solar power output prediction error, and inertial support coefficient.

[0067] Specifically, dynamic performance indicators essentially refer to the dynamic response characteristics of a system maintaining frequency stability under the dual effects of renewable energy output fluctuations and load disturbances. Dynamic performance indicators can include at least one of the following: frequency deviation rate, regulation time constant, and mean regional control error. The frequency deviation rate characterizes the degree to which the actual system frequency deviates from its rated value; the regulation time constant characterizes the dynamic process rate at which the system responds to power disturbances and recovers stability, reflecting the inertial characteristics and dynamic response efficiency of the AGC regulation loop; and the mean regional control error reflects the long-term balance between generation and load of the target power system during steady-state operation. Steady-state performance indicators essentially refer to the ability of automatic generation control to ensure that the system recovers and maintains a preset balance state after disturbance elimination. Steady-state performance indicators include at least one of the following: tie-line power error, frequency compliance rate, and renewable energy frequency volatility. The tie-line power error characterizes the degree to which the actual exchanged power between regions deviates from the planned value. The frequency compliance rate is used to statistically analyze the percentage of time the system frequency operates within the allowable deviation range and is a core indicator for measuring the long-term stability of the power system. The renewable energy frequency volatility rate is used to quantify the intensity and frequency of system frequency fluctuations caused by the grid connection of renewable energy sources such as wind and solar power. Economic indicators measure the comprehensive optimization capability of resource allocation and cost-effectiveness in the automatic generation control of a renewable energy power system during the process of achieving power balance and stable operation. Its core lies in balancing regulation costs and renewable energy consumption benefits, ensuring that the system achieves economical operation while meeting technical performance requirements. Economic indicators include at least one of the following: regulation cost, unit wear rate, and energy storage cycle loss rate. Regulation cost is the total economic resources invested in maintaining power balance and frequency stability during the automatic generation control process of a renewable energy power system; unit wear rate is a key indicator quantifying the degree of mechanical wear caused by frequent output adjustments of power generation equipment during automatic generation control, directly affecting unit lifespan and maintenance costs; energy storage cycle loss rate is a key indicator quantifying the capacity decay and performance degradation of energy storage systems due to charge-discharge cycles during automatic generation control, directly affecting the economic efficiency and regulation reliability of energy storage throughout its entire lifecycle. Renewable energy characteristic indicators are key indicators for evaluating the adaptability and regulation capability of automatic generation control in renewable energy power systems. Renewable energy characteristic indicators include at least one of the following: renewable energy penetration rate, wind and solar power output prediction error, and inertial support coefficient. The penetration rate of new energy sources is a core indicator for measuring the proportion of renewable energy in the installed capacity or power generation of the power system, and directly reflects the degree of decarbonization of the energy structure; the wind and solar power output prediction error refers to the difference between the predicted and actual values ​​of wind and solar power generation; the inertial support coefficient is a key parameter for evaluating the inertial support capability of power electronic equipment for the power grid.Given the inherent randomness and intermittency of wind and solar power, the AGC (Automatic Generation Control) of renewable energy sources connected to the grid requires more frequent adjustments to the output of traditional generating units to compensate for fluctuations in renewable energy output. This exacerbates the pressure on generating unit regulation. Furthermore, renewable energy sources connected to the grid via power electronic devices lack the rotational inertia of synchronous generators, resulting in a significant reduction in the system's equivalent inertia. A single indicator system is insufficient for effectively evaluating the AGC of the power system. This disclosure proposes an evaluation system comprising four primary indicators: dynamic performance indicators, steady-state performance indicators, economic indicators, and renewable energy characteristic indicators. Secondary indicators are established under these primary indicators, creating a multi-level, multi-dimensional indicator system to improve the accuracy of the evaluation.

[0068] In some embodiments of this disclosure, such as Figure 2 As shown, S22 includes:

[0069] S220. Perform data standardization processing on the multi-dimensional indicators in the evaluation system to obtain a standardized data matrix;

[0070] S222. Determine the covariance matrix based on the standardized data matrix;

[0071] S224. Perform eigenvalue decomposition on the covariance matrix to determine multiple eigenvectors;

[0072] S226. Construct the principal component load matrix by selecting a first preset number of the feature vectors in sequence;

[0073] S228. Determine the principal component score matrix based on the standardized data matrix and the principal component loading matrix.

[0074] In some embodiments of this disclosure, dimensional differences can be eliminated through data standardization (Z-score standardization) to ensure that all indicators participate in the calculation on the same scale. First, the original data matrix... Calculate the mean of each indicator. and standard deviation Its calculation formula is as follows (1):

[0075] (1)

[0076] In equation (1), i represents the current sample, j represents the current indicator, m represents the total number of samples, and n represents the total number of indicators.

[0077] The covariance matrix can be calculated to quantify the linear correlation between indicators, thus providing a basis for eigenvalue decomposition. The covariance matrix is ​​a symmetric matrix, with diagonal elements representing the variances of each indicator and off-diagonal elements representing the covariances between indicators. When the covariance of two indicators is high, it indicates a strong correlation between them, which can be eliminated by principal component analysis. The standardized data matrix can be calculated using the following formula (2). covariance matrix :

[0078] (2)

[0079] In equation (2), m represents the total number of samples. Represents a standardized data matrix.

[0080] Then, eigenvalue decomposition and eigenvector extraction can be used to determine the main direction of the data and its interpretability. First, the covariance matrix... Perform eigenvalue decomposition and solve for the eigenvalues. and the corresponding unit eigenvector You can refer to the following formula (3):

[0081] (3)

[0082] In equation (3), Represents eigenvalues. This represents the corresponding unit eigenvector. In some examples, the eigenvalues ​​can be sorted by size: .

[0083] The variance contribution rate of the j-th principal component is calculated as follows (4):

[0084] (4)

[0085] The cumulative variance contribution rate is the sum of the contribution rates of the first k principal components, calculated as follows (5):

[0086] (5)

[0087] In equations (4) and (5), the eigenvalues This represents the variance of the corresponding principal component, reflecting its information retention. In some embodiments, the smallest k value with a cumulative contribution rate ≥ 85% is typically selected to balance the dimensionality reduction effect with information loss.

[0088] Principal component loading matrices can be constructed to interpret the correlation between principal components and the original indices, thus giving the principal components physical meaning. The first k eigenvectors are selected. , forming the load matrix The j-th principal component The expression is as follows (6):

[0089] (6)

[0090] In equation (6), the larger the absolute value of the loading coefficient, the more significant the influence of the principal component on the corresponding original index.

[0091] The original data can be projected onto the principal component space to obtain a low-dimensional representation of the principal component score matrix. The calculation formula is as follows (7):

[0092] (7)

[0093] In equation (7), To standardize the data matrix, It is the load matrix composed of the first k eigenvectors.

[0094] In this embodiment, principal component analysis can be used to eliminate multicollinearity and avoid coefficient distortion in the regression model through orthogonal transformation, and high-dimensional indicators can be mapped to multi-dimensional space to facilitate intuitive analysis of AGC performance distribution patterns. Furthermore, load analysis can reveal the core indicator combinations that affect AGC performance, thereby guiding system optimization and improving the accuracy of performance evaluation.

[0095] In some embodiments of this disclosure, S24 includes:

[0096] S240. Form a label vector from the performance level of the automatic power generation control, and construct the input dataset by combining the principal component score matrix and the label vector;

[0097] S242. Generate a second preset number of training subsets from the dataset by random sampling with replacement, and construct a decision tree for each training subset. Repeat the construction until the second preset number of decision trees are generated.

[0098] S244. Each decision tree outputs a predicted performance evaluation sub-result according to its corresponding internal splitting rule. Based on the performance evaluation sub-results output by all decision trees, the final performance evaluation result of the target power system is determined through a majority voting mechanism.

[0099] In some embodiments of this disclosure, such as Figure 3 As shown, the Random Forest algorithm is an ensemble learning algorithm composed of multiple decision trees, which can be used to train and predict sample data. The algorithm combines the bagging algorithm based on training samples and the random subspace method based on feature sets, combining multiple decision trees. It randomly selects samples with replacement and uses some feature values ​​as output. The prediction result is generated by voting from the results of each tree, and the average of these results is taken as the final result. This significantly improves the model's generalization ability and robustness.

[0100] In some implementations, such as Figure 4 As shown, the dimensionality-reduced principal component factors can be used as input to construct a "Principal Component Factor-AGC Performance Level" dataset. The input data specifically includes the principal component score matrix and label vectors. The principal component score matrix is... , where m is the number of samples and k is the number of principal components extracted by PCA. The label vector is... This represents the AGC performance level. In some examples, the AGC performance level can be discretized into four categories: Excellent, Good, Average, and Poor.

[0101] In some examples, bootstrap sampling can be used to generate training subsets, and randomness can be introduced through sampling with replacement to enhance model diversity. Specifically, m samples (with duplicates allowed) are randomly selected from the original dataset D to generate a training subset. The samples that were not selected constitute the out-of-bag (OOB) dataset. , used to evaluate the performance of a single tree. For a random forest containing B trees, the training set of the t-th tree satisfies the following equation (8):

[0102] (8)

[0103] In equation (8), It indicates a uniform distribution.

[0104] The core of constructing a single decision tree is to maximize the class purity by recursively splitting the nodes of each tree. This can be achieved by minimizing the Gini index, the specific expression of which is given in equation (9):

[0105] (9)

[0106] In equation (9), The percentage of samples in category c is denoted as c, where C is the number of performance level categories.

[0107] In the splitting process, k principal components are first randomly selected. For the quantitative features, after sorting each feature by value, try all possible segmentation thresholds, calculate the decrease in Gini coefficient after splitting, and select the splitting method with the largest decrease, referring to the following formula (10):

[0108] (10)

[0109] In equation (10), and These are the sample sets of the left and right child nodes after the split, respectively.

[0110] The splitting process stops when the number of node samples is less than a preset threshold, all samples belong to the same category, or the maximum tree depth is reached.

[0111] Repeat the process until a second predetermined number of decision trees, B decision trees (usually B is between 500 and 1000), is generated, constructing a random forest. Each tree is trained independently, without interfering with others, forming a "forest". For new samples... Each tree outputs a class prediction. The final result is the mode, as shown in the following formula (11):

[0112] (11)

[0113] If the performance level is a continuous value, then the average value is output, as shown in the following formula (12):

[0114] (12)

[0115] Through the above steps, random forest can effectively establish a "principal component factor-performance level" mapping model. Compared with traditional methods, random forest captures the complex relationship between principal component factors and performance levels through multi-tree ensemble, reduces variance and improves generalization performance through bootstrap sampling and random feature selection, and quantifies the contribution of each principal component through Gini importance to guide the optimization direction of AGC system, providing high-precision and interpretable decision support for AGC performance evaluation of new energy power systems.

[0116] In some embodiments of this disclosure, the method further includes:

[0117] For each decision tree, the total decrease in the Gini coefficient of the principal component score matrix during all node splits is calculated to determine the feature importance score of the corresponding principal component factor;

[0118] Normalize the feature importance scores of all principal component factors and output the normalized feature importance scores of all principal component factors.

[0119] In some implementations, feature importance can be quantified by assessing the contribution of principal component factors to classification based on the decrease in the Gini coefficient, and statistical features can be evaluated for each tree. Total decrease in Gini coefficient across all node splits The feature importance score is obtained by averaging all trees. The calculation process can be referred to the following formula (13):

[0120] (13)

[0121] Then, normalization is performed so that the sum of the importance of all features is 1, as shown in equation (14):

[0122] (14)

[0123] In this embodiment, the contribution of each principal component can be quantified by Gini importance, which can guide the optimization direction of the AGC system and provide high-precision and interpretable decision support for the performance evaluation of AGC in new energy power systems.

[0124] In some examples, such as Figure 5 As shown, a simulation model of a new energy power system in a certain region is established based on the IEEE 33-node model, in which the wind power penetration rate is 35% and the photovoltaic penetration rate is 20%. The Matlab / Simulink simulation platform is used to simulate the AGC dynamic response process under different operating scenarios.

[0125] The entire simulation system includes 4 thermal power units, 2 hydropower units, 3 wind farms, 2 photovoltaic power stations, and 1 energy storage system traditional unit, with specific sub-parameters shown in Table 1.

[0126] Table 1 Unit performance parameters

[0127]

[0128] In the simulation environment, a random fluctuation of ±10% was superimposed on the base load curve (daily peak-to-valley difference 15%). Wind speed variations were simulated using a Weibull distribution (shape parameter 2.5, scale parameter 8 m / s). Photovoltaic power fluctuations caused by cloud cover were simulated using a Beta distribution (α=0.8, β=0.6). The energy storage system was configured with lithium-ion batteries, with a response time ≤200ms and a charge / discharge efficiency of 95%. A tie-line N-1 fault (lasting 5 minutes) was set to trigger AGC emergency control. A 24-hour load curve was simulated using Matlab / Simulink. System status data was collected every 5 seconds, generating 300 sets of samples. Each set contained 12 indicators, and the specific definitions and ranges of the evaluation indicators are shown in Table 2.

[0129] Table 2. Definitions and Data Scope of Evaluation Indicators

[0130]

[0131] Principal component extraction first calculates the mean and standard deviation from the original data matrix, and then uses the covariance matrix to quantify the linear relationship between the various indicators. The standardized 12 indicators are then subjected to PCA dimensionality reduction to obtain the principal component variance contribution rate shown in Table 3. The cumulative contribution rate of the first 3 principal components reaches 85.6%.

[0132] Table 3 Principal Component Variance Contribution Rate

[0133]

[0134] PC1 is the dynamic response factor, with the dominant indicators being frequency deviation rate, settling time constant, and mean ACE, used to reflect the system's dynamic adjustment capability. PC2 is the renewable energy correlation factor, heavily loaded by renewable energy penetration rate, inertia support coefficient, and wind and solar power output prediction error, characterizing the impact of renewable energy grid connection on system stability. PC3 is the economic factor, mainly driven by adjustment cost, unit wear rate, and energy storage cycle loss rate, reflecting the economic cost of the control process. The PCA load matrix is ​​shown in Table 4, mainly reflecting the linear combination relationship between the original indicators and principal components, obtained after eigenvector normalization.

[0135] Table 4 Load Matrix Analysis

[0136]

[0137] A random forest model was constructed with the following parameters: 500 trees, maximum depth 10, and 3 randomly selected features. The dataset was divided into a training set (240 groups) and a test set (60 groups) in an 8:2 ratio. The classification accuracy is shown in Table 5.

[0138] Table 5 Comparison of model classification accuracy

[0139]

[0140] Through confusion matrix analysis, the test set prediction results are as follows: Figure 6 As shown, the model has high accuracy in recognizing the "excellent" and "poor" categories, with accuracy rates of 0.965 and 0.889 respectively. However, there are a few misclassifications for "good" and "medium", mainly due to the overlap of features between the two categories.

[0141] The significance of the characteristic of the decrease in the Gini coefficient, such as Figure 7 As shown, the importance scores of PC1, PC2, and PC3 are 0.52, 0.31, and 0.17, respectively, with PC1 having the highest importance score, indicating that dynamic response capability is the core factor affecting AGC performance.

[0142] Figure 8 The test shows that the model's performance is tested by adjusting the permeability of different landscapes. When the permeability is less than 40%, the accuracy can reach more than 85%. When the permeability is greater than 40%, the system will deteriorate in dynamic characteristics due to insufficient inertia, which will reduce the accuracy.

[0143] Table 6 compares the performance of the model by different principal component numbers k. When k=3, the accuracy and computational efficiency can reach the optimal balance. When k=4, the cumulative contribution rate increases to 92.1%, but the model complexity increases and the training time increases by 23%. Therefore, this paper selects 3 principal components as the optimal state of the model.

[0144] Table 6. Impact of Principal Component Count on Model Performance

[0145]

[0146] The automatic generation control performance evaluation method provided in this disclosure breaks through the limitations of traditional single-dimensional evaluation system construction, constructing a multi-level system containing multiple dimensional indicators. It utilizes principal component analysis (PCA) to reduce the dimensionality of the multidimensional indicators in the evaluation system, extracting principal component factors, eliminating multicollinearity among indicators, balancing dimensionality reduction effects with information retention, and solving the model instability problem caused by indicator redundancy in traditional methods. Simultaneously, it employs a random forest algorithm, integrating multiple decision trees to capture the nonlinear relationship between principal component factors and AGC performance levels, providing precise guidance for optimization and overcoming the shortcomings of insufficient accuracy and lack of dynamic interpretability in traditional linear models.

[0147] It is understood that the various embodiments of the methods described in this specification are presented in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on its differences from other embodiments. Relevant details can be found in the descriptions of other method embodiments.

[0148] It should be understood that although the steps in the flowcharts shown in the accompanying drawings are displayed sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise expressly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some of the steps in the accompanying drawings may include multiple steps or stages, which are not necessarily completed at the same time, but may be executed at different times, and the execution order of these steps or stages is not necessarily sequential, but may be performed alternately or in turn with other steps or at least a portion of the steps or stages of other steps.

[0149] Based on the description of the above-described embodiments of the automatic power generation control performance evaluation method, this disclosure also provides an automatic power generation control performance evaluation device for implementing the above-described automatic power generation control performance evaluation method. The device may include a system (including a distributed system), software (application), module, component, controller, server, terminal, etc., using the method described in the embodiments of this specification, combined with necessary implementation hardware. Based on the same innovative concept, the devices in one or more embodiments provided in this disclosure are as described in the following embodiments. Since the implementation schemes and methods for solving the problem by the devices are similar, the implementation of specific devices in the embodiments of this specification can refer to the implementation of the aforementioned method, and repeated details will not be repeated. As used below, the terms "unit" or "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0150] The device can be the aforementioned terminal, a server, or a module, component, device, control unit, etc., integrated into the terminal. For details, please refer to... Figure 9 The device may include: an evaluation system construction module for constructing an evaluation system for automatic generation control of a target power system; a principal component factor extraction module for performing dimensionality reduction on multidimensional indicators in the evaluation system using a principal component analysis algorithm to extract multiple principal component factors; and an evaluation result output module for establishing an automatic generation control performance evaluation model based on the principal component factors using a random forest algorithm, and outputting the performance evaluation results of the target power system based on the input dataset using the automatic generation control performance evaluation model.

[0151] In some embodiments of the device, the principal component factor extraction module is further configured to perform data standardization processing on the multidimensional indicators in the evaluation system to obtain a standardized data matrix; and to determine the covariance matrix based on the standardized data matrix; and to perform eigenvalue decomposition on the covariance matrix to determine multiple eigenvectors; and to select a first preset number of eigenvectors in sequence to construct a principal component loading matrix; and to determine the principal component score matrix based on the standardized data matrix and the principal component loading matrix.

[0152] In some embodiments of the device, the evaluation result output module is further configured to form a label vector from the performance level of the automatic generation control, construct an input dataset from the principal component score matrix and the label vector; generate a second preset number of training subsets from the dataset by random sampling with replacement, construct a decision tree for each training subset, and repeat the construction until a second preset number of decision trees are generated; and output a predicted performance evaluation sub-result from each decision tree according to the corresponding internal splitting rule, and determine the final performance evaluation result of the target power system by a majority voting mechanism based on the performance evaluation sub-results output by all decision trees.

[0153] In some embodiments of the device, the evaluation result output module is further used to calculate the total decrease in the Gini coefficient of the principal component score matrix in all node splits for each decision tree, and determine the feature importance score of the corresponding principal component factor; and to perform normalization processing on the feature importance scores of all principal component factors, and output the normalized feature importance scores of all principal component factors.

[0154] In some embodiments of the device, the multidimensional indicators include at least two of the following: dynamic performance indicators, steady-state performance indicators, economic indicators, and new energy characteristic indicators.

[0155] In some embodiments of the device, the dynamic performance indicators include at least one of frequency deviation rate, regulation time constant, and mean regional control error; the steady-state performance indicators include at least one of tie-line power error, frequency qualification rate, and renewable energy frequency fluctuation rate; the economic indicators include at least one of regulation cost, unit wear rate, and energy storage cycle loss rate; and the renewable energy characteristic indicators include at least one of renewable energy penetration rate, wind and solar power output prediction error, and inertial support coefficient.

[0156] Each module in the aforementioned automatic power generation control performance evaluation device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0157] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 10As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements an automatic power generation control performance evaluation method.

[0158] Those skilled in the art will understand that Figure 10 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0159] Based on the foregoing description of the relevant methods and apparatus embodiments, this disclosure also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, it implements the automatic power generation control performance evaluation method described in any embodiment of this specification.

[0160] Based on the foregoing description of the relevant methods and apparatus embodiments, this disclosure also provides a computer-readable storage medium that, when the instructions in the computer-readable storage medium are executed by the processor of a computer device, enables the computer device to implement the automatic power generation control performance evaluation method as described in any embodiment of this disclosure.

[0161] Based on the foregoing description of the relevant methods and apparatus embodiments, this disclosure also provides a computer program product, including a computer program that, when executed by a processor, implements the automatic power generation control performance evaluation method described in any embodiment of this specification.

[0162] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on its differences from other embodiments. In particular, for hardware + program embodiments, since they are basically similar to the method embodiments, the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0163] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0164] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0165] It should be noted that the apparatus, computer equipment, storage medium, and computer program products described above may also include other implementation methods according to the description of the method embodiments. Specific implementation methods can be found in the description of the relevant method embodiments. Furthermore, new embodiments formed by combinations of features from various methods, apparatuses, devices, and server embodiments still fall within the scope of this disclosure and will not be elaborated upon here.

[0166] For ease of description, the above devices are described in terms of function, divided into various modules. Of course, when implementing one or more of these specifications, the functions of each module can be implemented in the same or different software and / or hardware, or a module that performs the same function can be implemented by a combination of multiple sub-modules or sub-units. The device embodiments described above are merely illustrative. For example, the division of modules or units is only a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling and communication connections between the devices or units shown or described can be implemented through direct and / or indirect coupling / connection, through standard or custom interfaces or protocols, and can be implemented electrically, mechanically, or in other forms.

[0167] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the following claims.

[0168] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope.

Claims

1. A method for evaluating the performance of automatic power generation control, characterized in that, The method includes: Construct an evaluation system for the automatic generation control of the target power system; The multidimensional indicators in the evaluation system are reduced in dimensionality using principal component analysis algorithm to extract multiple principal component factors. An automatic generation control performance evaluation model is established based on the principal component factors using the random forest algorithm. The automatic generation control performance evaluation model then outputs the performance evaluation results of the target power system based on the input dataset.

2. The method according to claim 1, characterized in that, The step involves using principal component analysis (PCA) to reduce the dimensionality of the multidimensional indicators in the evaluation system and extracting multiple principal component factors, including: The multidimensional indicators in the evaluation system are subjected to data standardization processing to obtain a standardized data matrix; Determine the covariance matrix based on the standardized data matrix; The covariance matrix is ​​subjected to eigenvalue decomposition to determine multiple eigenvectors; The principal component loading matrix is ​​constructed by selecting a first preset number of feature vectors in sequence. The principal component score matrix is ​​determined based on the standardized data matrix and the principal component loading matrix.

3. The method according to claim 2, characterized in that, The step of establishing an automatic generation control performance evaluation model based on the principal component factors using a random forest algorithm, and then outputting the performance evaluation results of the target power system based on the input dataset using the automatic generation control performance evaluation model, includes: The performance levels of automatic power generation control are used to form label vectors, and the principal component score matrix and the label vectors are used to construct the input dataset. A second preset number of training subsets are generated from the dataset by random sampling with replacement. A decision tree is constructed for each training subset. The construction is repeated until a second preset number of decision trees are generated. Each decision tree outputs a predicted performance evaluation sub-result based on its corresponding internal splitting rules. The final performance evaluation result of the target power system is determined by a majority voting mechanism based on the performance evaluation sub-results output by all decision trees.

4. The method according to claim 3, characterized in that, The method further includes: For each decision tree, the total decrease in the Gini coefficient of the principal component score matrix during all node splits is calculated to determine the feature importance score of the corresponding principal component factor; Normalize the feature importance scores of all principal component factors and output the normalized feature importance scores of all principal component factors.

5. The method according to claim 1, characterized in that, The multidimensional indicators include at least two of the following: dynamic performance indicators, steady-state performance indicators, economic indicators, and new energy characteristic indicators.

6. The method according to claim 5, characterized in that, The dynamic performance indicators include at least one of frequency deviation rate, regulation time constant, and mean regional control error; the steady-state performance indicators include at least one of tie-line power error, frequency qualification rate, and renewable energy frequency fluctuation rate; the economic indicators include at least one of regulation cost, unit wear rate, and energy storage cycle loss rate; and the renewable energy characteristic indicators include at least one of renewable energy penetration rate, wind and solar power output prediction error, and inertial support coefficient.

7. An automatic power generation control performance evaluation device, characterized in that, The device includes: The evaluation system construction module is used to construct an evaluation system for the automatic generation control of the target power system. The principal component factor extraction module is used to perform dimensionality reduction processing on the multidimensional indicators in the evaluation system through principal component analysis algorithm and extract multiple principal component factors. The evaluation result output module establishes an automatic generation control performance evaluation model based on the principal component factors using a random forest algorithm, and outputs the performance evaluation results of the target power system based on the input dataset using the automatic generation control performance evaluation model.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images