Short-term power load prediction method and system based on improved GWO algorithm

Abstract

The invention discloses a short-term power load prediction method and system based on an improved GWO algorithm, and relates to the technical field of power systems, and the method comprises the steps: obtaining the historical load data of a power system, determining a power load influence factor, and carrying out the preprocessing; constructing a power load prediction SVM model based on the preprocessed data; and on the basis of the constructed power load prediction SVM model, an improved GWO algorithm is combined for optimization, and a power load prediction result is obtained. The method not only can adapt to load change and still keep high-precision prediction under the condition of large load fluctuation, but also effectively avoids the problem of falling into local optimum through the improved algorithm, and enhances the robustness of the model. The method has high application value in an actual power system, can provide more accurate and reliable decision support for power dispatching, power grid operation optimization, demand side management and the like, is beneficial to improving the economical efficiency and reliability of the power system and reducing the operation cost, and provides powerful support for fine management of power enterprises.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and in particular to a short-term power load forecasting method and system based on an improved GWO algorithm. Background Technology

[0002] Electricity load forecasting is a crucial component, prerequisite, and foundation for power system planning and grid operation. Its forecasting methods primarily start with known economic and social development and electricity demand conditions. Through the analysis and research of historical data, and based on future economic and social development scenarios, it makes advance estimates of electricity demand. Therefore, electricity load forecasting is of paramount importance.

[0003] Because power systems are complex and large-scale systems with uncertainty, nonlinearity, and dynamic openness, traditional power load forecasting methods (time series method, regression analysis method, grey model method, expert system method, etc.) are difficult to accurately describe the complex nonlinear characteristics of power systems, and therefore cannot perform more accurate load forecasting.

[0004] Currently, some progress has been made in electricity load forecasting, primarily through machine learning algorithms such as Support Vector Machines (SVM). However, while SVM possesses good local search capabilities, it suffers from poor stability and low accuracy when searching the entire market. Therefore, a novel electricity load forecasting method is urgently needed to improve the accuracy of electricity load forecasting. Summary of the Invention

[0005] In view of the problems existing in the short-term power load forecasting methods based on the improved GWO algorithm, this invention is proposed.

[0006] Therefore, the problem that this invention aims to solve is that traditional power load forecasting methods are difficult to accurately describe the complex nonlinear characteristics of power systems, and thus cannot perform more accurate load forecasting.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0008] In a first aspect, embodiments of the present invention provide a short-term power load forecasting method based on an improved GWO algorithm, comprising: acquiring historical load data of the power system, determining the factors influencing the power load, and performing preprocessing; constructing a power load forecasting SVM model based on the preprocessed data; and optimizing the constructed power load forecasting SVM model in conjunction with the improved GWO algorithm to obtain the power load forecasting result.

[0009] As a preferred embodiment of the short-term power load forecasting method based on the improved GWO algorithm described in this invention, the power load influencing factors include meteorological indicators, demographic indicators, economic indicators, holiday indicators, government control measures, and electricity price-economic indicators; the meteorological indicators are daily meteorological data concurrent with the power system load data; the demographic indicators are year-end resident population data from official statistical yearbooks concurrent with the power system load data; the economic indicators are GDP data from the National Bureau of Statistics concurrent with the power system load data; the holiday indicators are statutory holiday data concurrent with the power system load data; the government control measures are government electricity price control measures concurrent with the power system load data; and the electricity price-economic composite indicator is the ratio of electricity price to per capita GDP concurrent with the power system load data.

[0010] As a preferred embodiment of the short-term power load forecasting method based on the improved GWO algorithm described in this invention, the preprocessing includes equalizing the data and removing outliers.

[0011] As a preferred embodiment of the short-term power load forecasting method based on the improved GWO algorithm described in this invention, the improved GWO algorithm includes: constructing an initial wolf pack; processing the initial wolf pack based on the differential evolution algorithm to obtain an optimized initial wolf pack; improving the convergence factor based on the cosine law; updating the positions of the four types of wolf packs based on the step size Euclidean distance; optimizing the GWO algorithm to escape local optima through a lens imaging back learning strategy; repeating the update and optimization until the maximum number of iterations is reached to obtain the optimization result.

[0012] As a preferred embodiment of the short-term power load forecasting method based on the improved GWO algorithm described in this invention, the following steps are taken: The initial wolf pack is processed using the differential evolution algorithm to obtain an optimized initial wolf pack. Specifically, this includes: constructing a parent population based on the initial wolf pack; inputting both the parent population and the preprocessed dataset into an SVM model to calculate the population objective function value of the parent population; performing crossover and mutation processing on the parent population using the differential evolution algorithm to obtain a offspring population; inputting both the offspring population and the preprocessed dataset into an SVM model to calculate the population objective function value of the offspring population; determining whether the population objective function value of the offspring population is greater than the population objective function value of the parent population, obtaining a first determination result; if the first determination result is yes, updating the parent population to the offspring population; if the first determination result is no, retaining the parent population; incrementing the value of the second iteration number by 1, and returning to the differential evolution algorithm to perform crossover and mutation processing on the parent population until the second iteration number reaches the second iteration number threshold, and determining the parent population as the optimized initial wolf pack.

[0013] As a preferred embodiment of the short-term power load forecasting method based on the improved GWO algorithm described in this invention, the convergence factor is improved based on the cosine law, and the calculation formula is expressed as follows:

[0014]

[0015] Among them, a finiaial and a final Let be the initial and final values ​​of the convergence factor 'a', and t be the current iteration number. max The maximum number of iterations is n, where n is the decreasing exponent, and 0 is the maximum number of iterations. <n≤1。

[0016] As a preferred embodiment of the short-term power load forecasting method based on the improved GWO algorithm described in this invention, the location of the four types of wolf packs is updated based on the step-size Euclidean distance, and the calculation formula is expressed as follows:

[0017]

[0018] Where W1, W2, and W3 represent the learning rates of wolf ω to wolves α, β, and δ, respectively, and X1, X2, and X3 represent the position vectors of the gray wolves.

[0019] Secondly, embodiments of the present invention provide a short-term power load forecasting system based on an improved GWO algorithm, comprising: a collection module for acquiring historical load data of the power system, determining the factors influencing power load, and performing preprocessing; a construction module for constructing a power load forecasting SVM model based on the preprocessed data; and an optimization module for optimizing the constructed power load forecasting SVM model in conjunction with the improved GWO algorithm to obtain power load forecasting results.

[0020] Thirdly, embodiments of the present invention provide a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement any of the steps of the above-described short-term power load forecasting method based on the improved GWO algorithm.

[0021] Fourthly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the above-described short-term power load forecasting method based on the improved GWO algorithm.

[0022] The beneficial effects of this invention are that it not only adapts to load changes and maintains high-accuracy prediction even under conditions of large load fluctuations, but also effectively avoids getting trapped in local optima through improved algorithms, thus enhancing the robustness of the model. It has high application value in practical power systems, providing more accurate and reliable decision support for power dispatching, grid operation optimization, and demand-side management, helping to improve the economy and reliability of power systems, reduce operating costs, and provide strong support for the refined management of power enterprises. Attached Figure Description

[0023] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0024] Figure 1 The flowchart shows a short-term power load forecasting method based on the improved GWO algorithm.

[0025] Figure 2 A flowchart for improving the GWO algorithm. Detailed Implementation

[0026] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0027] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0028] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0029] Example 1

[0030] Reference Figure 1 and Figure 2This is the first embodiment of the present invention, which provides a short-term power load forecasting method based on an improved GWO algorithm, including:

[0031] S1: Obtain historical load data of the power system, determine the factors affecting power load, and perform preprocessing.

[0032] Factors affecting electricity load include meteorological indicators, demographic indicators, economic indicators, holiday indicators, government regulation measures, and electricity price-economic indicators.

[0033] Meteorological indicators are daily meteorological data concurrent with power system load data; demographic indicators are year-end resident population data from official statistical yearbooks concurrent with power system load data; economic indicators are GDP data from the National Bureau of Statistics concurrent with power system load data; holiday indicators are statutory holiday data concurrent with power system load data; government control measures are government electricity price control measures concurrent with power system load data; and the electricity price-economic composite indicator is the ratio of electricity price to per capita GDP concurrent with power system load data.

[0034] Preprocessing the acquired data includes equalizing the intervals and removing outliers.

[0035] For data at different time intervals, interpolation or sampling is performed to make it into time series data with equal intervals, including:

[0036] For data with large intervals, such as monthly data, linear interpolation is used to insert missing daily data points. For data with small intervals, such as minute-level data, the mean or median method is used for downsampling to obtain hourly or daily data.

[0037] After processing, all data will be converted to the same time interval, which facilitates subsequent modeling and analysis.

[0038] The moving window method is used for further data processing. First, an appropriate window size is determined so that the current data point is placed in the center of the window. For daily load data, the initial window size is set to 7 days.

[0039] The window slides across the time series, moving one time unit at a time.

[0040] For the data point at the center of the window, compare it with other data points within the window to determine if it is an outlier. Specifically:

[0041] Calculate the statistics within the window: including the mean (μ) and standard deviation (σ) of the data within the window;

[0042] Set a threshold. If the difference between the data point at the window center and the window average exceeds the set threshold, mark it as a potential outlier. For the data points marked as outliers, choose to replace or adjust them.

[0043] S2: Based on the preprocessed data, construct an SVM model for power load forecasting.

[0044] Based on the preprocessed data, divide the data set, including a training set, a validation set, and a test set;

[0045] The training set is represented as: {x1, x2,..., xn};

[0046] The validation set is represented as: {x(n + 1), x(n + 2),..., xm};

[0047] The test set is represented as: {x(m + 1), x(m + 2),..., xT}

[0048] Among them, 1 < n < m < T. By moving the window multiple times, each time sliding forward a fixed time step, multiple training-validation-test sets are created.

[0049] The basic principle of the support vector machine (SVM) is to find an optimal hyperplane in the feature space to maximize the margin between different classes of samples. In the case of using the RBF kernel, the decision function is expressed as:

[0050]

[0051] Among them, f(x) is the prediction function, α i and α i * are Lagrange multipliers, K(x i , x) is the kernel function, x i is the support vector, x is the input feature vector, b is the bias term, and N is the number of support vectors.

[0052] Furthermore, the kernel function K(x [[ID=四十]] i , x) is expressed as:

[0053] K(x i , x) = exp(-γ||x <一〇〇〇〇一一>- x|| 2 )

[0054] Among them, γ is the parameter of the RBF kernel, controlling the width of the Gaussian function.

[0055] Train the model and construct the objective function, expressed as:

[0056]

[0057] It should be noted that there are some incorrect or unclear tag references in the original text, such as the inconsistent format of some tags like i , i , etc. I have translated them as they are while maintaining the integrity of the text. If there are specific requirements for these tags, please provide more accurate information for a more appropriate translation.Constraints must be met:

[0058]

[0059] ξ i ξ i * ≥0

[0060] Where w represents the weight vector, C is the penalty parameter, and ξ represents the tolerance for control error. i ξ i * Let represent the slack variable, and ε represent the allowable error range. This represents a function that maps the input space to the feature space.

[0061] The model performance is evaluated using a validation set, with metrics including root mean square error (RMSE) and mean absolute error (MAE). Based on the validation set results, further parameter tuning or feature selection is performed, repeated multiple times until satisfactory performance is achieved. The model is then evaluated using a test set to verify its generalization ability.

[0062] S3: Based on the constructed power load forecasting SVM model combined with the improved GWO algorithm, the power load forecasting results are obtained.

[0063] The steps to improve the GWO algorithm include:

[0064] S3.1 Construct the initial wolf pack;

[0065] S3.2 The initial wolf pack is processed based on the differential evolution algorithm to obtain an optimized initial wolf pack;

[0066] S3.3 Improvement of the convergence factor based on the cosine law;

[0067] S3.4 updates the positions of the four types of wolf packs based on the step size Euclidean distance;

[0068] S3.5 optimizes the GWO algorithm to escape local optima through a lens imaging reverse learning strategy;

[0069] Repeat steps S3.4 and S3.5 until the maximum number of iterations is reached to obtain the optimization result.

[0070] Preferably, in step S3.2, the initial wolf pack is processed based on the differential evolution algorithm to obtain an optimized initial wolf pack, specifically including:

[0071] Construct the parent population based on the initial wolf pack;

[0072] Both the parent population and the preprocessed dataset are input into the SVM model to calculate the population objective function value of the parent population.

[0073] The differential evolution algorithm is used to perform crossover and mutation on the parent population to obtain the offspring population;

[0074] Both the offspring population and the preprocessed dataset are input into the SVM model to calculate the population objective function value of the offspring population.

[0075] Determine whether the population objective function value of the offspring population is greater than the population objective function value of the parent population to obtain the first determination result;

[0076] If the first judgment result is yes, then update the parent population to the offspring population;

[0077] If the first judgment result is negative, then the parent population is retained;

[0078] Increment the number of the second iteration by 1, and return to the differential evolution algorithm to perform crossover and mutation on the parent population to obtain the offspring population again, until the number of the second iteration reaches the threshold of the second iteration, and then determine the parent population as the optimized initial wolf pack.

[0079] Preferably, in step S3.3, the convergence factor is improved based on the cosine law, and the calculation formula is expressed as follows:

[0080]

[0081] Among them, a finiaial and a final Let be the initial and final values ​​of the convergence factor 'a', and t be the current iteration number. max The maximum number of iterations is n, where n is the decreasing exponent, and 0 is the maximum number of iterations. <n≤1。

[0082] Preferably, in step S3.4, the positions of the four types of wolf packs are updated based on the step-size Euclidean distance, and the calculation formula is as follows:

[0083]

[0084] Where W1, W2, and W3 represent the learning rates of wolf ω to wolves α, β, and δ, respectively, X1, X2, and X3 represent the position vectors of gray wolves, and t is the number of iterations.

[0085] The improved GWO algorithm can effectively avoid the problem of getting trapped in local optima, and has stronger global search capabilities, resulting in higher accuracy and stability. Furthermore, the improved GWO algorithm, combined with SVM, can achieve accurate prediction of power load, allowing staff to provide guidance on power load optimization in advance.

[0086] In summary, the improved GWO algorithm used in this invention can effectively avoid the problem of the GWO algorithm getting trapped in local optima, and has stronger global search capabilities, resulting in higher accuracy and stability. The improved GWO algorithm combined with SVM can achieve accurate prediction of power load, allowing staff to be arranged in advance to guide the optimization of power load.

[0087] Example 2

[0088] Reference Figure 1 Based on the first embodiment, this embodiment further provides a short-term power load forecasting system based on an improved GWO algorithm, including:

[0089] The data collection module is used to acquire historical load data of the power system, identify factors affecting power load, and perform preprocessing.

[0090] The building module is used to construct an SVM model for power load forecasting based on preprocessed data;

[0091] The optimization module is used to optimize the power load forecasting SVM model based on the constructed power load forecasting model and the improved GWO algorithm to obtain the power load forecasting results.

[0092] This embodiment also provides a computer device applicable to the short-term power load forecasting method based on the improved GWO algorithm, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the short-term power load forecasting method based on the improved GWO algorithm proposed in the above embodiment.

[0093] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices 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 and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0094] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the short-term power load forecasting method based on the improved GWO algorithm as proposed in the above embodiments.

[0095] The storage medium proposed in this embodiment and the data storage method proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0096] Example 3

[0097] Based on the previous two embodiments, this embodiment provides a short-term power load forecasting method based on the improved GWO algorithm. To verify the beneficial effects of the present invention, a simulation experiment is conducted for scientific demonstration.

[0098] To verify the effectiveness of the short-term power load forecasting method based on the improved GWO algorithm proposed in this invention, this embodiment compares the method with the traditional Support Vector Machine (SVM) method and the method combining the basic Grey Wolf Optimization (GWO) algorithm with SVM. The experimental data uses daily power load data and related influencing factor data from a certain province from January 1, 2018 to December 31, 2019.

[0099] The comparison methods include:

[0100] Method 1: Use the traditional SVM method for modeling and prediction;

[0101] Method 2: Optimize SVM parameters using the basic GWO algorithm for modeling and prediction;

[0102] Method 3: Use the improved GWO algorithm proposed in this invention to optimize SVM parameters and perform modeling and prediction.

[0103] The comparison results are shown in Table 1:

[0104] Table 1 Comparison of Experimental Results

[0105] method RMSE (MW) MAE(MW) MAPE (%) R2 PLFE (%) CT(s) MPE (%) SD(MW) Method 1 156.32 124.75 3.85 0.9124 4.62 15.3 8.75 168.43 Method 2 132.18 105.63 3.26 0.9386 3.95 42.7 7.43 142.56 Method 3 98.45 78.92 2.43 0.9672 2.81 56.2 5.68 105.29

[0106] The table above shows that the method of this invention demonstrates comprehensive advantages in short-term power load forecasting. In terms of forecast accuracy, the three key indicators—RMSE, MAE, and MAPE—are significantly better than the traditional SVM and basic GWO-SVM methods, with reductions exceeding 36% compared to the traditional SVM method and exceeding 25% compared to the basic GWO-SVM method. The goodness of fit is also significantly improved, with the coefficient of determination R0... 2 The value reached 0.9672, indicating that the predicted results fit the actual load better.

[0107] In peak load forecasting, a key indicator, the method of this invention has an error of only 2.81%, which is 39.18% and 28.86% lower than that of traditional SVM and basic GWO-SVM, respectively, highlighting its advantages in forecasting critical load points. The stability of the forecast results is also significantly improved, with the lowest maximum percentage error and standard deviation at 5.68% and 105.29MW, respectively, meaning more stable forecast results and fewer extreme errors. Although the computation time is slightly increased (56.2 seconds), considering the significant improvement in forecast accuracy, this computation time is still within an acceptable range and does not affect practical applications.

[0108] In summary, except for computation time, the method of this invention performs best in all evaluation indicators, fully demonstrating its comprehensive advantages in short-term power load forecasting and providing a more reliable and accurate forecasting tool for power system planning and operation.

[0109] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A short-term power load forecasting method based on an improved GWO algorithm, characterized in that: include, Obtain historical load data of the power system, determine the factors affecting the power load, and perform preprocessing; Based on the preprocessed data, a power load forecasting SVM model is constructed; Based on the constructed power load forecasting SVM model combined with the improved GWO algorithm, the power load forecasting results are obtained.

2. The short-term power load forecasting method based on the improved GWO algorithm according to claim 1, characterized in that: The power load influencing factors include meteorological indicators, demographic indicators, economic indicators, holiday indicators, government control measures and electricity price-economic indicators; The meteorological index is the daily meteorological data of the same period as the power system load data; The demographic indicators are the year-end resident population data from the official statistical yearbook for the same period as the power system load data; The economic indicator is the GDP data of the Statistics Bureau for the same period as the power system load data; The holiday index is the statutory holiday data for the same period as the power system load data; The government control measures are government electricity price control measures in the same period as the power system load data; The electricity price-economic comprehensive index is the ratio of electricity price to per capita GDP in the same period as the power system load data.

3. The short-term power load forecasting method based on the improved GWO algorithm according to claim 2, characterized in that: The preprocessing includes equalizing the data and removing outliers.

4. The short-term power load forecasting method based on the improved GWO algorithm according to claim 3 is characterized in that: The improved GWO algorithm comprises: Build the initial wolf pack; The initial wolf pack is processed based on the differential evolution algorithm to obtain an optimized initial wolf pack; Improve the convergence factor based on the cosine law; The positions of four types of wolf packs are updated based on the step-length Euclidean distance; Optimize the GWO algorithm to escape from the local optimum through the reverse learning strategy of lens imaging; Repeat the update and optimization until the maximum number of iterations is reached and the optimal result is obtained.

5. The short-term power load forecasting method based on the improved GWO algorithm according to claim 4, characterized in that: The initial wolf pack is processed based on the differential evolution algorithm to obtain an optimized initial wolf pack, specifically including: Construct the parent population based on the initial wolf pack; Input the parent population and the preprocessed data set into the SVM model and calculate the population objective function value of the parent population; Using differential evolution algorithm, crossover and mutation processing are performed on the parent population to obtain the offspring population; Input the offspring population and the preprocessed data set into the SVM model and calculate the population objective function value of the offspring population; Determine whether the population objective function value of the offspring population is greater than the population objective function value of the parent population, and obtain a first determination result; If the first judgment result is yes, the parent population is updated to the child population; If the first judgment result is no, the parent population is retained; The value of the second iteration number is increased by 1, and the differential evolution algorithm is returned to perform crossover and mutation processing on the parent population until the second iteration number reaches the second iteration number threshold, and the parent population is determined as the optimized initial wolf population.

6. The short-term power load forecasting method based on the improved GWO algorithm according to claim 5, characterized in that: The convergence factor is improved based on the cosine law, and the calculation formula is expressed as: Among them, a finiaial and a final is the initial value and final value of the convergence factor a, t is the current iteration number, t max is the maximum number of iterations, n is the decreasing index, 0 <n≤1。 7. The short-term power load forecasting method based on the improved GWO algorithm according to claim 6, characterized in that: Based on the step length Euclidean distance, the positions of the four types of wolf packs are updated, and the calculation formula is expressed as: Among them, W1, W2, W3 represent the learning rates of ω wolf to α, β, δ wolf respectively, and X1, X2, X3 represent the position vector of gray wolf.

8. A short-term power load forecasting system based on an improved GWO algorithm, based on the short-term power load forecasting method based on an improved GWO algorithm according to any one of claims 1 to 7, characterized in that: include, The collection module is used to obtain the historical load data of the power system, determine the factors affecting the power load, and perform preprocessing; A construction module is used to construct a power load prediction SVM model based on the preprocessed data; The optimization module is used to obtain the power load forecasting result by optimizing the constructed power load forecasting SVM model combined with the improved GWO algorithm.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the short-term power load forecasting method based on the improved GWO algorithm according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the short-term power load forecasting method based on the improved GWO algorithm described in any one of claims 1 to 7 are implemented.

Images