Ultra-short-term power load prediction method and system

Abstract

The invention provides an ultra-short-term power load prediction method and system, and the method comprises the steps: obtaining historical power load data, dividing a training set, and carrying out the normalization processing of the training set; setting the input of the prediction model KELM as load data of seven days before the prediction date; setting the population number and the number of iterations of the IWOA algorithm, and upper and lower limits of a KELM parameter C and gamma; performing random initialization on the population to generate a parameter initial solution of the KELM; judging whether the current number of iterations t of the IWOA algorithm reaches the maximum number of iterations Tmax or not; when tlt is satisfied; on the premise of Tmax, calculating a target function value of an individual, and recording a current optimal target function value; updating the positions of the whale individuals according to the parameters, and recording the updated positions; outputting an IWOA algorithm operation result to obtain optimization parameters C and gamma of the KELM; and taking the optimization parameters C and gamma as configuration parameters of the KELM, and obtaining a power load prediction result according to the input data. According to the method, the parameters of the power load prediction model KELM can be optimized, so that the model can perform more accurate prediction under different transformer substation load data sets.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to an ultra-short-term power load forecasting method and system. Background Art

[0002] As a high-energy-consuming industry, how to reduce power loss has always been a hot and difficult issue in the development process of shipbuilding enterprises. At present, the production tasks of shipyards are relatively saturated, and the power system is often in a high-level operation state. In addition, the continuous increase in new loads connected to the shipyard has directly led to a continuous increase in the power load of the enterprise, so that the original distribution network has been in a high-loss operation state. In addition, due to the long construction years of some substations, the planned and designed transformer capacity can no longer meet the current load access situation, and power overcapacity often occurs, which further increases power loss and causes equipment overheating, seriously affecting the safe operation of the distribution system. Therefore, it is necessary to accurately predict the power load of shipbuilding enterprises and predict the time period of the maximum power load in advance, so as to adjust the power consumption plan in time and avoid power overcapacity. This is of great significance for ensuring the safe and stable operation of the power system, reducing the energy consumption of enterprises, and saving production costs for enterprises.

[0003] For the daily operation and maintenance and production scheduling needs of shipyards, ultra-short-term power load forecasting should be carried out in units of hours or minutes. At present, the power load forecasting model can include a combined forecasting model that integrates the advantages of multiple models. This model has the problem that parameter selection cannot be applied to a variety of complex power data. Therefore, it is very necessary to design a method for adaptive optimization of parameters to enhance the adaptability of the power load forecasting model to predict under different substation load data sets. Summary of the invention

[0004] In order to overcome the problem of difficulty in optimizing the parameters of the power load forecasting model, the present application provides an ultra-short-term power load forecasting method and system.

[0005] To achieve the above objectives, some embodiments of the present application provide an ultra-short-term power load forecasting method, comprising the following steps:

[0006] Step 1. Obtain historical power load data, select historical active load data of a certain period of time as a training set, and normalize the training set;

[0007] Step 2. Set the input of the prediction model KELM to the data of the seven days before the prediction date, and the output to the data of the 24 hours on the prediction day;

[0008] Step 3. Set the population size and iteration number of the IWOA algorithm, and set the upper and lower limits of the parameters C and γ of the prediction model KELM, where C is the regularization coefficient and γ represents the generalization ability of the prediction model KELM and the degree of adaptability to the data;

[0009] Step 4. Randomly initialize the population to generate an initial solution of the parameters of the prediction model KELM;

[0010] Step 5. Determine whether the current number of iterations t of the IWOA algorithm reaches the maximum number of iterations T max ;

[0011] Step 6. When the t is satisfied <T max Under the premise of , calculate the individual objective function value and record the current optimal objective function value;

[0012] Step 7. Update the individual position of the whale according to the probability p and distance adjustment parameter |A| in the shrinking encirclement and spiral advance mechanism in the IWOA algorithm, and record the updated position; wherein the expression of the probability p is:

[0013] Step 8. Repeat steps 5 to 7;

[0014] Step 9. Output the calculation result of the IWOA algorithm to obtain the optimization parameters C and γ of the KELM;

[0015] Step 10. Use the optimization parameters C and γ as configuration parameters of the prediction model KELM, input the historical power load data of seven days before the date to be predicted into the prediction model KELM, and obtain the power load prediction result on the prediction date.

[0016] Furthermore, the IWOA algorithm searches for the optimal solution process including:

[0017] Each whale in the population represents an individual, and a position of the individual in the search space represents a solution to the optimization problem; the position of the target prey represents the target solution. If a random whale individual in the current population is represented by The position of the optimal solution in the population is expressed as S represents the dimension of the solution;

[0018] In addition, in order to enhance the optimization ability in the KELM process, an adaptive inertia weight method is used to update the individual position. The expression of the adaptive inertia weight w is shown in the following formula (1):

[0019]

[0020] In the formula, w maxand w min The values ​​are set to 1 and 0.3 respectively;

[0021] After the introduction of the adaptive inertia weight, all individuals in the population will surround the optimal solution position according to the model constructed according to the following formulas (2)(5) to (6)(9):

[0022]

[0023] A=2a·rand-a (4)

[0024] B=2·rand (5)

[0025] a=2×(1-t / T max ) (6)

[0026] In the formula, t represents the current iteration number, T max Indicates the maximum number of iterations; represents the distance between the position of the i-th whale and the position of an individual in the population; A and B are distance adjustment parameters; a represents the convergence factor, which decreases linearly from 2 to 0 as the number of iterations increases.

[0027] Furthermore, the IWOA algorithm searching for the optimal solution process also includes:

[0028] During the predation process, the whale changes its position through two mechanisms: contraction and spiral advance. The contraction and spiral mechanism performs linear decreasing encirclement by reducing the convergence factor a. The mechanism expression is shown in equations (7), (8) and (10), where represents the distance between the position of the i-th whale and the optimal individual position in the population;

[0029]

[0030] In the formula, represents the optimal individual position in the current population, Indicates the position of the new generation of individuals in the population;

[0031] The spiral forward mechanism is that the whale moves towards the target prey in a spiral motion trajectory to achieve the search for the optimal solution. The expression of the search mechanism is shown in equations (9) and (12):

[0032]

[0033] In the formula, b is a constant used to define the shape of the spiral, l is a random number in (-1,1), and e bl ×cos(2πl) represents the mathematical model of the whale's spiral motion.

[0034] Furthermore, the IWOA algorithm searching for the optimal solution process also includes:

[0035] The individual whales randomly search for target prey during the shrinking and encircling process by controlling the distance adjustment parameter |A|. When |A|≥1, the whales perform random encirclement according to formula (2)(5) to formula (6)(5). When |A|<1, the whales perform encirclement toward the optimal solution according to formula (7), (8) and formula (9)(10).

[0036] Furthermore, the objective function of the IWOA algorithm is expressed as shown in equations (10) and (17):

[0037]

[0038] In the formula, represents the predicted power load, y i represents the actual power load, N represents the number of points in the data set, and w j represents the output weight of KELM, and λ represents the L1 regularization coefficient.

[0039] Furthermore, the expression of the prediction model KELM algorithm is shown in formula (11)(1):

[0040]

[0041] Where I represents the identity matrix of the same dimension as the kernel function matrix; the regularization coefficient C is set by the user to ensure the balance between the output weight and the training error; S represents the expected output; K(x,x1) represents the kernel function; Ω ELM Represents the ELM kernel function matrix. The KELM algorithm maps the input space to a high-dimensional feature space through the kernel function, where:

[0042] Ω ELMi,j =K(x i ,x j ) (12).

[0043] Furthermore, the kernel function of the KELM algorithm adopts the radial basis RBF kernel function, and its expression is shown in formula (13)(4):

[0044] K(x,z)=exp(-γ||xz|| 2 ) (13)

[0045] In the formula, ||xz|| 2 It represents the squared Euclidean distance between two feature vectors. The parameter γ affects the generalization ability of the model and its adaptability to the data.

[0046] On the other hand, the present application provides an ultra-short-term electricity load forecasting system, which includes a memory for storing computer program instructions and a processor for executing program instructions, wherein when the computer program instructions are executed by the processor, the system is triggered to execute the above-mentioned ultra-short-term electricity load forecasting method. DETAILED DESCRIPTION

[0047] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0048] The following is a detailed description of the embodiments of the present invention.

[0049] It can be understood that the extreme learning machine (ELM) algorithm is a single-layer feedforward neural network algorithm. The algorithm uses a random mechanism to reduce parameter settings and selections. Compared with traditional neural network algorithms, the ELM algorithm has the advantages of fast training speed and good generalization performance. The kernelized extreme learning machine (KELM) algorithm combines the fast learning ability of ELM and the nonlinear mapping ability of the kernel function to achieve effective processing of nonlinear data while maintaining a high training speed and generalization performance. The expression of the KELM algorithm is shown in formula (1)(1):

[0050]

[0051] Where I represents the unit matrix of the same dimension as the kernel function matrix, and C represents the regularization coefficient, which is set by the user to ensure the balance between the output weight and the training error. Generally speaking, C is set to a small value to ensure that the model has good generalization ability.

[0052] Where S represents the expected output; Ω ELM Represents the ELM kernel function matrix. The KELM algorithm maps the input space to a high-dimensional feature space through the kernel function, where:

[0053] Ω ELMi,j =K(x i ,x j ) (2)

[0054] K(x, x1) represents the kernel function. In practical applications, the more common kernel functions include radial basis RBF (Radial Basis Function) kernel function, polynomial (poly) kernel function, linear (line) kernel function, etc. RBF kernel function has excellent local approximation properties and global continuity. The present invention selects RBF as the kernel function of KELM, and its expression is shown in formula (3) (4).

[0055] K(x,z)=exp(-γ||xz|| 2 ) (3)

[0056] In the formula, ||xz|| 2 It represents the squared Euclidean distance between two feature vectors. The parameter γ affects the generalization ability of the model and its adaptability to the data.

[0057] The selection of parameters C and γ has a crucial impact on the prediction performance of the KELM model and needs to be selected according to the specific data set and prediction problem. This application uses the improved Whale Optimization Algorithm (WOA) to adaptively optimize the parameters C and γ to enhance the adaptability of the KELM model to predict under different substation load data sets.

[0058] It can be understood that the Whale Optimization Algorithm (WOA) is a meta-heuristic swarm optimization algorithm proposed based on the hunting behavior of humpback whale groups in nature. Compared with other swarm optimization algorithms, this algorithm has the advantages of simple operation and few adjustment parameters, and can meet the needs of shipyard operation and maintenance personnel. The algorithm searches for the optimal solution through three processes: surrounding prey, bubbling predation, and random search.

[0059] (1) Surrounding the prey

[0060] Each whale in the population represents an individual, and a position of an individual in the search space represents a solution to the optimization problem. The position of the target prey represents the target solution. If a random whale individual in the current population is represented by The position of the optimal solution in the population is expressed as S represents the dimension of the solution. All individuals in the population will surround the model constructed according to equations (4), (5) to (8), (9) to the position of the optimal solution.

[0061]

[0062] A=2a·rand-a (6)

[0063] B=2·rand (7)

[0064] a=2×(1-t / T max) (8)

[0065] In the formula, t represents the current iteration number, T max Indicates the maximum number of iterations; represents the distance between the position of the i-th whale and the position of an individual in the population; A and B are distance adjustment parameters; a represents the convergence factor, which decreases linearly from 2 to 0 as the number of iterations increases.

[0066] (2) Foaming predation

[0067] During the hunting process, whales change their positions through two mechanisms: contraction and spiral advancement. In the WOA algorithm, the probability of both methods of movement is 50%.

[0068] ① Contraction and encirclement mechanism: This mechanism performs linear decreasing encirclement by reducing parameter a. The expression of this mechanism is shown in formula (10), where Represents the distance between the position of the i-th whale and the best individual position in the population.

[0069]

[0070] In the formula, represents the optimal individual position in the current population, Represents the position of the new generation of individuals in the population.

[0071] ② Spiral advancement mechanism: Whales move toward the target prey in a spiral trajectory to search for the optimal solution.

[0072] The search mechanism expression is shown in formula (12).

[0073]

[0074] In the formula, b is a constant used to define the shape of the spiral, l is a random number in (-1,1), and e bl ×cos(2πl) represents the mathematical model of the whale's spiral motion.

[0075] (3) Random Search

[0076] The random search of individual whales for target prey is achieved by controlling the distance adjustment parameter |A| during the shrinking and encircling process. When |A|≥1, the whales perform random encirclement according to equations (4)(5) to (8)(5). When |A|<1, the whales perform encirclement towards the optimal solution according to equations (9), (10) and (11)(10).

[0077] Although the WOA algorithm has a novel structure and relatively few control parameters, it reduces the impact of artificial experience parameters on the results. However, when faced with discrete and complex problems, it has the disadvantages of low global search efficiency, poor convergence, and easy to fall into local optimality. In order to enhance the global search efficiency of the WOA algorithm and enhance the convergence in the later stages of iteration, some embodiments of the present application provide an improved whale optimization algorithm IWOA (Improved Whale Optimization Algorithm), and the present application improves the algorithm.

[0078] (1) In the process of whale hunting, the probability of using the two mechanisms of shrinking and encircling and spiraling is 50% each. In WOA, the probability parameter is set to a random number [0,1], which cannot guarantee a good search capability in the early stage. Therefore, the present invention sets a heuristic probability search to enhance its shrinking and encircling capability in the early stage and its spiraling capability in the later stage. The expression of the heuristic probability p is as follows:

[0079]

[0080] (2) Adaptive inertia weight

[0081] In order to enhance the optimization capability in the WOA optimization KELM process, some embodiments of the present application propose an adaptive inertia weight method for updating individual positions to avoid falling into local optimality and improve the optimization performance of the algorithm. The expression of the adaptive inertia weight is shown in the following formula (13).

[0082]

[0083] In the formula, w max and w min The values ​​are set to 1 and 0.3 respectively. After introducing the adaptive weight, equation (4(4)) and equation (9(9)) are updated as follows:

[0084]

[0085] (3) Objective function and L1 regularization coefficient

[0086] In the KELM algorithm, a larger C value will increase the degree of fit to the training set samples, and at the same time will lead to a deterioration in the generalization ability and an increase in the error of the prediction results. When the root mean square error (RMSE) is selected as the objective function of the WOA algorithm, the algorithm's optimization parameters will inevitably lead to parameter C being too large, making it impossible to accurately predict the load. This application introduces L1 regularization in the fitness function to limit the output parameters of the KELM model, thereby ensuring the generalization ability of the algorithm and reducing the prediction error. The expression of the objective function is shown in equations (13) and (17), where represents the predicted load, y i represents the actual load, N represents the number of points in the data set, and w j represents the output weight of KELM, and λ represents the L1 regularization coefficient.

[0087]

[0088] Next, in some embodiments of the present application, the IWOA algorithm is used to optimize the KELM algorithm, and the method steps for ultra-short-term power load forecasting are as follows:

[0089] Step 1. Obtain power load data, select the historical active load data of the research object for 8 months as the training set, and normalize the training set;

[0090] Step 2. Set the input of the prediction model KELM to the data of the seven days before the prediction date, and the output to the data of the 24 hours on the prediction day;

[0091] Step 3. Set the population size and iteration number of the IWOA algorithm, and set the upper and lower limits of the parameters C and γ of the prediction model KELM, where C is the regularization coefficient and γ represents the generalization ability of the prediction model KELM and the degree of adaptability to the data;

[0092] Step 4. Randomly initialize the population to generate an initial solution of the parameters of the prediction model KELM;

[0093] Step 5. Determine whether the current number of iterations t of the IWOA algorithm reaches the maximum number of iterations T max ;

[0094] Step 6. After the conditions in step 5 are met, <T max Under the premise of , calculate the individual objective function value according to formula (16) and record the current optimal objective function value;

[0095] Step 7. Update the individual position of the whale according to the probability p and distance adjustment parameter |A| in the shrinking encirclement and spiral advance mechanism in the IWOA algorithm, and record the updated position; wherein the expression of the probability p is:

[0096] Step 8. Repeat steps 5 to 7;

[0097] Step 9. Output the calculation result of the IWOA algorithm to obtain the optimization parameters C and γ of the KELM;

[0098] Step 10. Using the optimization parameters C and γ as the configuration parameters of the KELM, input the historical power load data of seven days before the date to be predicted into the prediction model KELM to obtain the power load prediction result on the prediction date.

[0099] Some embodiments of the present application also provide an ultra-short-term power load forecasting system, which includes a memory for storing computer program instructions and a processor for executing the program instructions, wherein when the computer program instructions are executed by the processor, the system is triggered to execute the aforementioned ultra-short-term power load forecasting method.

[0100] In summary, the solution provided by the present application obtains power load data, divides the training set, and normalizes the training set; sets the input prediction date of the prediction model KELM; sets the number of IWOA algorithm populations and the number of iterations, and sets the upper and lower limits of KELM parameters C and γ; randomly initializes the population to generate the initial solution of KELM parameters; determines whether the current number of iterations t of the IWOA algorithm reaches the maximum number of iterations T max ; When satisfying t <T max Under the premise of (16), calculate the individual objective function value and record the current optimal objective function value; update the individual position of the whale according to the parameter and record the updated position; output the IWOA algorithm operation result to obtain the optimization parameters C and γ of KELM; use the optimization parameters C and γ as the configuration parameters of KELM, and obtain the power load forecast result according to the input data. This application can optimize the parameters of the power load forecasting model KELM, so that the model can make more accurate predictions under different substation load data sets.

[0101] It should be noted that the present application can be implemented in software and / or a combination of software and hardware, for example, can be implemented using an application specific integrated circuit (ASIC), a general purpose computer or any other similar hardware device. In one embodiment, the software program of the present application can be executed by a processor to implement the steps or functions described above. Similarly, the software program of the present application (including relevant data structures) can be stored in a computer-readable recording medium, for example, a RAM memory, a magnetic or optical drive or a floppy disk and similar devices. In addition, some steps or functions of the present application can be implemented using hardware, for example, as a circuit that cooperates with a processor to perform each step or function.

[0102] In a typical configuration of the present application, the terminal and the network device each include one or more processors (CPU), input / output interface, network interface and memory.

[0103] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0104] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include non-transitory media such as modulated data signals and carrier waves.

[0105] In addition, a part of the present application may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present application through the operation of the computer. The program instruction for calling the method of the present application may be stored in a fixed or removable recording medium, and / or transmitted through a data stream in a broadcast or other signal-bearing medium, and / or stored in a working memory of a computer device that runs according to the program instruction. Here, according to an embodiment of the present application, a device is included, the device including a memory for storing computer program instructions and a processor for executing program instructions, wherein, when the computer program instruction is executed by the processor, the device is triggered to run the method and / or technical solution based on the aforementioned multiple embodiments according to the present application.

[0106] It is obvious to those skilled in the art that the present application is not limited to the details of the exemplary embodiments described above, and that the present application can be implemented in other specific forms without departing from the spirit or essential features of the present application. Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive, and the scope of the present application is defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present application.

Claims

1. A method for ultra-short-term power load forecasting, characterized in that: The following steps are involved: Step 1. Obtain historical power load data, select historical active load data of a certain period of time as a training set, and normalize the training set; Step 2. Set the input of the prediction model KELM to the data of the seven days before the prediction date, and the output to the data of the 24 hours on the prediction day; Step 3. Set the population size and iteration number of the IWOA algorithm, and set the upper and lower limits of the parameters C and γ of the prediction model KELM, where C is the regularization coefficient and γ represents the generalization ability of the prediction model KELM and the degree of adaptability to the data; Step 4. Randomly initialize the population to generate an initial solution of the parameters of the prediction model KELM; Step 5. Determine whether the current number of iterations t of the IWOA algorithm reaches the maximum number of iterations T max ; Step 6. When the t is satisfied <T max Under the premise of , calculate the individual objective function value and record the current optimal objective function value; Step 7. Update the individual position of the whale according to the probability p and distance adjustment parameter |A| in the shrinking encirclement and spiral advance mechanism in the IWOA algorithm, and record the updated position; wherein the expression of the probability p is: Step 8. Repeat steps 5 to 7; Step 9. Output the calculation result of the IWOA algorithm to obtain the optimization parameters C and γ of the KELM; Step 10. Use the optimization parameters C and γ as configuration parameters of the prediction model KELM, input the historical power load data of seven days before the date to be predicted into the prediction model KELM, and obtain the power load prediction result on the prediction date.

2. The method according to claim 1, characterized in that The IWOA algorithm searches for the optimal solution process including: Each whale in the population represents an individual, and a position of the individual in the search space represents a solution to the optimization problem; the position of the target prey represents the target solution. If a random whale individual in the current population is represented by The position of the optimal solution in the population is expressed as S represents the dimension of the solution; In addition, in order to enhance the optimization ability in the KELM process, an adaptive inertia weight method is used to update the individual position. The expression of the adaptive inertia weight w is shown in the following formula (1): In the formula, w max and w min The values ​​are set to 1 and 0.3 respectively; After the introduction of the adaptive inertia weight, all individuals in the population will surround the optimal solution position according to the model constructed according to the following formulas (2)(5) to (6)(9): A=2a·rand-a (4) B=2·rand (5) a=2×(1-t / T max ) (6) In the formula, t represents the current iteration number, T max Indicates the maximum number of iterations; represents the distance between the position of the i-th whale and the position of an individual in the population; A and B are distance adjustment parameters; a represents the convergence factor, which decreases linearly from 2 to 0 as the number of iterations increases.

3. The method according to claim 2, characterized in that The IWOA algorithm search for the optimal solution process also includes: During the predation process, the whale changes its position through two mechanisms: contraction and spiral advance. The contraction and spiral mechanism performs linear decreasing encirclement by reducing the convergence factor a. The mechanism expression is shown in equations (7), (8) and (10), where represents the distance between the position of the i-th whale and the optimal individual position in the population; In the formula, represents the optimal individual position in the current population, Indicates the position of the new generation of individuals in the population; The spiral forward mechanism is that the whale moves towards the target prey in a spiral motion trajectory to achieve the search for the optimal solution. The expression of the search mechanism is shown in equations (9) and (12): In the formula, b is a constant used to define the shape of the spiral, l is a random number in (-1,1), and e bl ×cos(2πl) represents the mathematical model of the whale's spiral motion.

4. The method according to claim 3, characterized in that The IWOA algorithm search for the optimal solution process also includes: The individual whales randomly search for target prey during the shrinking and encircling process by controlling the distance adjustment parameter |A|. When |A|≥1, the whales perform random encirclement according to formula (2)(5) to formula (6)(5). When A|<1, the whales perform encirclement toward the optimal solution according to formula (7), (8) and formula (9)(10).

5. The method according to claim 1, characterized in that The objective function of the IWOA algorithm is expressed as shown in equations (10) and (17): In the formula, represents the predicted power load, y i represents the actual power load, N represents the number of points in the data set, and w j represents the output weight of KELM, and λ represents the L1 regularization coefficient.

6. The method according to claim 1, characterized in that The expression of the prediction model KELM algorithm is shown in formula (11)(1): Where I represents the identity matrix of the same dimension as the kernel function matrix; the regularization coefficient C is set by the user to ensure the balance between the output weight and the training error; S represents the expected output; K(x,x1) represents the kernel function; Ω ELM Represents the ELM kernel function matrix. The KELM algorithm maps the input space to a high-dimensional feature space through the kernel function, where: Oh ELMi,j =K(x i ,x j ) (12).

7. The method according to claim 6, characterized in that The kernel function of the KELM algorithm adopts the radial basis RBF kernel function, and its expression is shown in formula (13)(4): K(x,z)=exp(-γ||xz|| 2 ); (13) In the formula, ||xz|| 2 It represents the squared Euclidean distance between two feature vectors. The parameter γ affects the generalization ability of the model and its adaptability to the data.

8. An ultra-short-term power load forecasting system, the system comprising a memory for storing computer program instructions and a processor for executing the program instructions, wherein: When the computer program instructions are executed by the processor, the system is triggered to execute the method according to any one of claims 1 to 7.

Images