Virtual power plant power dispatching method and system based on intelligent optimization algorithm

Abstract

The invention discloses a virtual power plant power dispatching method and system based on an intelligent optimization algorithm, and relates to the technical field of power dispatching, and the method comprises the following steps: obtaining a historical load capacity and a historical temperature of a target power utilization area, and obtaining an initial prediction load capacity function based on the historical load capacity and the historical temperature; optimizing parameters of the initial prediction load capacity function based on a particle swarm optimization algorithm to obtain a final prediction load capacity function; obtaining the future temperature of the target power utilization area, and carrying out calculation based on the future temperature and the final prediction load capacity function to obtain a predicted future load capacity; performing power dispatching on the target power utilization area based on the predicted future load capacity; the method is used for solving the problem that the load demand predicted by a prediction model is inaccurate if the temperature change of the prediction model trained by using historical power data and the timestamp of the historical power data is large in a short time in the existing power dispatching technology.

Description

Technical Field

[0001] The present invention relates to the technical field of power dispatching, and specifically to a virtual power plant power dispatching method and system based on an intelligent optimization algorithm. Background Art

[0002] The virtual power plant technology provides a flexible and efficient solution for power dispatching by integrating and optimizing distributed energy resources such as solar energy, wind energy, energy storage systems, and controllable loads. It uses smart grid technology and demand response programs to achieve real-time monitoring, prediction, and dispatching of electricity; The existing virtual power plant power dispatching method trains a prediction model based on historical power data and the time stamps of historical power data, and then conducts power dispatching based on the prediction model. However, the actual load is related to temperature. When the temperature is different, the electrical appliances used and their load capacities are also different. For example, when it is too cold or too hot, air conditioners need to be turned on. If the temperature changes greatly in a short period of time, resulting in changes in the number of electrical appliances used and their load capacities, the prediction model trained based on historical power data and the time stamps of historical power data will be inaccurate. For example, in the patent application with the publication number CN117691583A, a power dispatching system and method for a virtual power plant are disclosed. Another example is the patent application with the publication number CN118469257A, which discloses a power plant optimization dispatching method and a virtual control system based on a machine learning model. Both of these two solutions use historical power data and the time stamps of historical power data to train a prediction model. If the temperature changes greatly in a short period of time, the predicted electrical load will be inaccurate. In the existing power dispatching technology, the prediction model trained using historical power data and the time stamps of historical power data will result in inaccurate predicted load demands if the temperature changes greatly in a short period of time. Summary of the Invention

[0003] The present invention aims to solve at least one of the technical problems in the existing technology to a certain extent. By obtaining an initial predicted load quantity function based on historical load quantity and historical temperature, then using the particle swarm optimization algorithm to obtain a final predicted load quantity function, calculating the predicted future load quantity by using the future temperature and the final predicted load quantity function, and performing power dispatching on the target power consumption area based on the predicted future load quantity, to solve the problem that in the existing power dispatching technology, if the temperature changes greatly in a short period of time, the predicted load demand by the prediction model will be inaccurate.

[0004] To achieve the above object, in the first aspect, the present application provides a virtual power plant power dispatching method based on an intelligent optimization algorithm, including the following steps: Obtain the historical load and historical temperature of the target power consumption area, and obtain the initial predicted load function based on the historical load and historical temperature; Optimize the parameters of the initial predicted load function based on the particle swarm optimization algorithm to obtain the final predicted load function; Obtain the future temperature of the target power consumption area, and calculate the predicted future load based on the future temperature and the final predicted load function; perform power dispatching on the target power consumption area based on the predicted future load; Obtain the real-time temperature and real-time load of the target power consumption area, and determine whether to update the final predicted load function based on the real-time temperature, real-time load and the final predicted load function.

[0005] Furthermore, obtaining the historical load and historical temperature of the target power consumption area includes the following sub-steps: Obtain the first quantity of historical loads and historical temperatures; where the historical load is the load within one week of the target power consumption area, and the historical temperature is the average temperature within one week corresponding to the historical load of the target power consumption area; Calculate the mean value of all historical loads corresponding to each historical temperature, and mark it as the historical load mean value; Establish a plane rectangular coordinate system with the historical temperature as the X-axis and the historical load mean value as the Y-axis, mark it as the historical load mean coordinate system, plot the historical temperature and the historical load mean value into the historical load mean coordinate system to obtain a scatter plot, and mark it as the historical load mean scatter plot.

[0006] Furthermore, obtaining the initial predicted load function based on the historical load and historical temperature includes the following sub-steps: Mark each coordinate point in the historical load mean scatter plot as a historical load mean coordinate point; Obtain the number of historical load mean coordinate points, and mark it as n; Calculate the slopes of the two historical load mean coordinate points with the closest abscissas among all, and mark it as the load slope. The calculation formula is: ; where Kx i is the slope of the i-th and i-1-th historical load mean coordinate points, Y i is the ordinate of the i-th historical load mean coordinate, X i is the abscissa of the i-th historical load mean coordinate, Y i-1 is the ordinate of the i-1-th historical load mean coordinate, X i-1 is the abscissa of the i-1-th historical load mean coordinate. At this time, the range of i is an integer from 2 to n, and the size of i corresponds to the size of the abscissa of the historical load mean coordinate.

[0007] Further, obtaining the initial predicted load function based on historical load and historical temperature further includes the following sub-steps: Calculate the difference between two adjacent load slopes for all, marked as slope difference, and the calculation formula is: Kw i =Kx i+1 -Kx i ; where Kw i is the difference between the (i + 1)-th and i-th load slopes, Kx i+1 is the (i + 1)-th load slope, and Kx i is the i-th load slope. At this time, the range of i is an integer from 1 to n - 1, and the value of i corresponds to the abscissa of the historical load mean coordinate where the load slope is generated; Judge whether all slope differences exceed the difference range threshold. If so, calculate the error between the (i + 2)-th and i-th load slopes, marked as detection difference, and the calculation formula is: Kwd i =Kx i+2 -Kx i ; where Kwd i is the difference between the (i + 2)-th and i-th load slopes, and Kx i+2 is the (i + 2)-th load slope. At this time, the range of i is an integer from 1 to n - 2; Judge whether the detection error exceeds the difference range threshold. If so, obtain the abscissa of the i-th historical load mean coordinate corresponding to the i-th load slope, marked as abscissa threshold.

[0008] Further, obtaining the initial predicted load function based on historical load and historical temperature further includes the following sub-steps: Count the number of abscissa thresholds and mark it as m; Mark the abscissa thresholds in ascending order as T j , where j is an integer from 1 to m; Obtain the minimum abscissa of the historical load mean coordinate point, marked as T q , and obtain the maximum abscissa of the historical load mean coordinate point, marked as T z ; Obtain the segmentation ranges as: [Tq, T1], [T1, T2],..., [Tm, Tz].

[0009] Further, obtaining the initial predicted load function based on historical load and historical temperature further includes the following sub-steps: Taking the historical temperature as the X-axis and the historical load as the Y-axis, a rectangular coordinate system is established and marked as the historical load coordinate system. The historical temperature and the corresponding historical load are plotted into the historical load coordinate system to obtain a scatter plot, which is marked as the historical load scatter plot; Set the preset initial prediction load function as: Y = a β *X + b β , where a β and b β are the parameters of the preset initial prediction load function, and β is an integer from 1 to m + 1; Based on the historical load scatter plot and the segmented range, multi-segment linear fitting is performed to obtain the specific values of a β and b β of the preset initial prediction load function in each segmented range. Substitute the specific values of a β and b β into the preset initial prediction load function to obtain m + 1 initial prediction load functions.

[0010] Furthermore, based on the particle swarm optimization algorithm, optimizing the parameters of the initial prediction load function to obtain the final prediction load function includes the following sub-steps: Obtain the specific values of a β and b β of any one initial prediction load function, and mark them as as and bs respectively; Obtain the second quantity of particles, where the particle is represented as (w, e). Set the value range of w as [as - p1, as + p1], and the value range of e as [bs - p2, bs + p2]; p1 is the w range constant, and p2 is the e range constant; Calculate the fitness of each particle as: ; where MSE is the fitness of each particle, X i is the historical temperature, Y i is the historical load corresponding to X i , and N is the quantity of X i ; Record the particle with the minimum fitness as the optimal solution; Update the particle velocity and position. The particle velocity update formula is: Vn α = u * vo α + c1 * r1 * (pdb α - p α ) + c2 * r2 * (pdg α - p α ); where Vn α is the velocity of particle α after update; voα is the velocity before the update of particle α; u is the inertia factor; c1 and c2 are learning factors; r1 and r2 are random numbers between 0 and 1; pdb α is the optimal solution of particle α; pdg α is the global optimal solution; p α is the current particle position; The particle position update formula is: pn α =p α +Vn α ; where pn α is the position after the particle update; Repeat the update of the particle velocity and position until the optimal solution of the particle no longer changes; Obtain the global optimal solution of the particle, marked as (wy, ey); Substitute wy for as and ey for bs into the initial predicted load function to obtain the final predicted load function.

[0011] Furthermore, based on the future temperature and the final predicted load function, calculate the predicted future load as follows: Obtain the average temperature in the target power consumption area in the next week, marked as the future temperature; Substitute the future temperature into the final predicted load function in the corresponding segmented range to obtain the predicted future load; Schedule the load in the target power consumption area in the next week according to the predicted future load.

[0012] Furthermore, based on the real-time temperature, real-time load and the final predicted load function, judge whether to update the final predicted load function, including the following steps: Obtain the average temperature and load in the target power consumption area in the recent week, marked as the real-time temperature and real-time load respectively; Substitute the real-time temperature into the final predicted load function in the corresponding segmented range to obtain the predicted real-time load; Calculate the absolute value of the difference between the real-time load and the predicted real-time load, marked as the detection difference; Judge whether the detection difference exceeds the detection threshold. If it exceeds, add a third quantity of historical load and historical temperature, update the final predicted load function and optimize it using the particle swarm optimization algorithm.

[0013] In a second aspect, the present application also provides a virtual power plant power dispatching system based on an intelligent optimization algorithm, including: a function acquisition module, a function optimization module, a power dispatching module, and a function update module; The function acquisition module is used to obtain the historical load and historical temperature of the target power consumption area, and obtain the initial predicted load function based on the historical load and historical temperature; The function optimization module is used to optimize the parameters of the initial predicted load quantity function based on the particle swarm optimization algorithm to obtain the final predicted load quantity function; The power dispatching module is used to obtain the future temperature of the target power consumption area, calculate the predicted future load quantity based on the future temperature and the final predicted load quantity function; and perform power dispatching on the target power consumption area based on the predicted future load quantity; The function update module is used to obtain the real-time temperature and real-time load quantity of the target power consumption area, and determine whether to update the final predicted load quantity function based on the real-time temperature, real-time load quantity and the final predicted load quantity function.

[0014] Advantages of the present invention: The present invention obtains the initial predicted load quantity function based on historical load quantity and historical temperature, then uses the particle swarm optimization algorithm to obtain the final predicted load quantity function, calculates the predicted future load quantity based on the future temperature and the final predicted load quantity function, and performs power dispatching on the target power consumption area based on the predicted future load quantity. The advantages are that the future load quantity of the target area can be obtained according to the predicted future temperature, and the corresponding load quantity can be provided for the target area based on the future load quantity. When the short-term temperature changes greatly, this method can accurately predict the load quantity of the target power consumption area and improve the accuracy of the predicted future load quantity; The present invention optimizes the parameters of the initial predicted load quantity function based on the particle swarm optimization algorithm to obtain the final predicted load quantity function. The advantage is that more optimal parameters are found through the particle swarm optimization algorithm based on the parameters of the initial predicted load quantity function, which improves the accuracy of predicting the future load quantity of the target power consumption area. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is the principle block diagram of the system of the present invention; Figure 2 It is the schematic diagram of the historical load mean scatter plot of the present invention; Figure 3 It is the schematic diagram of the historical load quantity scatter plot of the present invention; Figure 4 It is the step flow chart of the method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0017] Embodiment 1, please refer to Figure 1As shown in the figure, the present application provides a virtual power plant power dispatching system based on an intelligent optimization algorithm, including: a function acquisition module, a function optimization module, a power dispatching module, and a function update module; The function acquisition module is used to obtain the historical load and historical temperature of the target power consumption area, and obtain an initial predicted load function based on the historical load and historical temperature; The function acquisition module is configured with a strategy for obtaining a historical load mean scatter plot. The strategy for obtaining a historical load mean scatter plot includes: Obtain the first quantity of historical loads and historical temperatures; where the historical load is the load within one week of the target power consumption area, and the historical temperature is the average temperature within one week corresponding to the historical load of the target power consumption area; Calculate the mean value of all historical loads corresponding to each historical temperature, and mark it as the historical load mean; Establish a plane rectangular coordinate system with the historical temperature as the X-axis and the historical load mean as the Y-axis, mark it as the historical load mean coordinate system, and plot the historical temperature and the historical load mean into the historical load mean coordinate system to obtain a scatter plot, marked as the historical load mean scatter plot.

[0018] The function acquisition module is configured with a strategy for obtaining a load slope. The strategy for obtaining a load slope includes: Mark each coordinate point in the historical load mean scatter plot as a historical load mean coordinate point; Obtain the number of historical load mean coordinate points, marked as n; Calculate the slope of the two historical load mean coordinate points with the closest abscissas, marked as the load slope. The calculation formula is: ; where Kx i is the slope of the i-th and i-1-th historical load mean coordinate points, Y i is the ordinate of the i-th historical load mean coordinate, X i is the abscissa of the i-th historical load mean coordinate, Y i-1 is the ordinate of the i-1-th historical load mean coordinate, X i-1 is the abscissa of the i-1-th historical load mean coordinate. At this time, the range of i is an integer from 2 to n, and the size of i corresponds to the size of the abscissa of the historical load mean coordinate; where i-1 is at least 1, so the range of is an integer from 2 to n; In practical applications, if the i-th and i-1-th historical load mean coordinate points are (24, 10) and (22.5, 10.18), that is, Kx i is (10 - 10.18) / (24 - 22.5) = -0.12; The function acquisition module is configured with a strategy for acquiring the abscissa threshold, and the strategy for acquiring the abscissa threshold includes: Calculate the difference between all adjacent two load slopes, and mark it as the slope difference. The calculation formula is: Kw i =Kx i+1 -Kx i ; Where Kw i is the difference between the (i + 1)-th and i-th load slopes, Kx i+1 is the (i + 1)-th load slope, and Kx i is the i-th load slope. At this time, the range of i is an integer from 1 to n - 1, and the size of i corresponds to the abscissa of the historical load mean coordinate that generates the load slope; where the minimum value of i is 1, so the range is an integer from 1 to n - 1; Since the temperature and the load amount are roughly linearly related, but when the temperature is too high or too low, the linear change trend of the load amount will be different, that is, the slopes of the linear relationships are different. Therefore, in order to find the linear relationships in different temperature ranges, a detection difference is set, and the slope differences are judged in turn to find the temperature values where the slope changes; Judge whether all the slope differences exceed the difference range threshold. If they exceed, calculate the error between the (i + 2)-th and i-th load slopes, and mark it as the detection difference. The calculation formula is: Kwd i =Kx i+2 -Kx i ; Where Kwd i is the difference between the (i + 2)-th and i-th load slopes, and Kx i+2 is the (i + 2)-th load slope. At this time, the range of i is an integer from 1 to n - 2; increasing the detection difference is set to exclude mutation values in the linear relationship to find the correct segmentation range; the setting of the difference range threshold. The difference threshold range is set. Since the slope range is between [-1, 1], the difference threshold range should not be set too large but can distinguish the difference between two slopes. The specific difference threshold range is set to [-0.2, 0.2]; Judge whether the detection error exceeds the difference range threshold. If it exceeds, obtain the abscissa of the i-th historical load mean coordinate corresponding to the i-th load slope, and mark it as the abscissa threshold; where the error threshold is set In practical applications, Kx i is -0.12, Kx i+1 is 0.662, Kx i+2It is 0.667. The slope difference is calculated as 0.782. The difference threshold range is set as [-0.2, 0.2]. Since the slope difference of 0.782 exceeds the set difference threshold range, the detection difference is calculated as 0.787, and the detection difference also exceeds the set difference threshold range. The abscissa of the i-th historical load mean coordinate corresponding to the i-th load slope is obtained as 24, that is, 24 is the abscissa threshold; The function acquisition module is configured with a strategy for acquiring the segmentation range, and the strategy for acquiring the segmentation range includes: Count the number of abscissa thresholds and mark it as m; Mark the abscissa thresholds in ascending order as T j , where j is an integer from 1 to m; Obtain the minimum abscissa of the historical load mean coordinate point and mark it as T q , obtain the maximum abscissa of the historical load mean coordinate point and mark it as T z ; The obtained segmentation ranges are: [Tq, T1], [T1, T2],..., [Tm, Tz]; In practical applications, please refer to Figure 2 As shown, the number of abscissa thresholds m is 2, T1 = 0 °C, T2 = 24 °C. The minimum abscissa T q of the historical load mean coordinate point is obtained as -20 °C, and the maximum abscissa T z of the historical load mean coordinate point is obtained as 36 °C. Therefore, the segmentation ranges are [-20, 0], [0, 24], [24, 36]; The function acquisition module is configured with a strategy for acquiring the initial predicted load quantity function, and the strategy for acquiring the initial predicted load quantity function includes: Establish a plane rectangular coordinate system with historical temperature as the X-axis and historical load quantity as the Y-axis, mark it as the historical load quantity coordinate system, plot the historical temperature and the corresponding historical load quantity into the historical load quantity coordinate system to obtain a scatter plot, and mark it as the historical load quantity scatter plot; Set the preset initial predicted load quantity function as: Y = a β *X + b β , where a β and b β are the parameters of the preset initial predicted load quantity function, and β is an integer from 1 to m + 1; Based on the historical load quantity scatter plot and the segmentation range, perform multi-segment linear fitting to obtain the specific values of a β and b β of the preset initial predicted load quantity function in each segmentation range. Substitute the specific values of a β and b β into the preset initial predicted load quantity function to obtain m + 1 initial predicted load quantity functions.

[0019] The function optimization module is used to optimize the parameters of the initial predicted load function based on the particle swarm optimization algorithm to obtain the final predicted load function; In practical applications, please refer to Figure 3 As shown, the initial predicted load function obtained by multi-segment linear fitting is a standard function, that is, a β The range is [-1, 1]. The initial predicted load function in the temperature range of [-20, 0] is Y = -0.4*X + 13; the initial predicted load function in the temperature range of [0, 24] is Y = -0.125*X + 13; the initial predicted load function in the temperature range of [24, 36] is Y = 0.667*X - 6; The function optimization module is configured with a strategy for obtaining the final predicted load function. The strategy for obtaining the final predicted load function includes: Obtain the a β and b β Specific values are respectively marked as as and bs; Obtain a second number of particles, where the particle is represented as (w, e). Set the value range of w to [as - p1, as + p1], and the value range of e to [bs - p2, bs + p2]; p1 is the range constant of w, and p2 is the range constant of e; among them, the particle swarm optimization algorithm is a kind of intelligent optimization algorithm. The particle swarm optimization algorithm is inspired by the foraging behavior of bird flocks and solves problems by simulating the social behavior of biological groups such as bird flocks or fish schools; since the range of as is [-1, 1], the range of w should not be set too large. For example, p1 is 0.2. If as + p1 > 1, as + p1 can be set to 1. If as - p1 < -1, as - p1 can be set to -1, and p2 is set to 5; the particle (w, e) is set to take values uniformly within the range; The principle applied to the present invention is to regard the particles in the particle swarm optimization algorithm as the a β and b β Specific values of. Because as and bs are linearly fitted, based on as and bs, the particle swarm optimization algorithm can be used to continue searching for the values near as and bs to obtain more suitable a β and b β Specific values; Calculate the fitness of each particle as: ; where MSE is the fitness of each particle, X i is the historical temperature, Y i is the historical load corresponding to X i , and N is X iThe quantity; substituting historical temperatures with different (w, e) into the initial predicted load quantity function to obtain the predicted load quantity, calculating the mean square error between the predicted load quantity and the historical load quantity as MSE, and using MSE to evaluate the particles with appropriate parameters. The smaller the MSE, the smaller the error of the initial predicted load quantity function; Denote the particle with the minimum fitness value as the optimal solution; Update the particle velocity and position. The particle velocity update formula is: Vn α =u*vo α +c1*r1*(pdb α -p α )+c2*r2*(pdg α -p α ); Where Vn α is the velocity of particle α after update; vo α is the velocity of particle α before update; u is the inertia factor; c1 and c2 are learning factors; r1 and r2 are random numbers between 0 and 1; pdb α is the optimal solution of particle α; pdg α is the global optimal solution; p α is the current particle position, that is, the current (w, e); The particle position update formula is: pn α =p α +Vn α ; where pn α is the position of the particle after update, that is, the updated (w, e); Repeat the update of the particle velocity and position until the optimal solution of the particle no longer changes; Obtain the global optimal solution of the particle, marked as (wy, ey); the global optimal solution of the particle is the most suitable parameter of the initial predicted load quantity function; Substitute wy for as and ey for bs into the initial predicted load quantity function to obtain the final predicted load quantity function.

[0020] In practical applications, obtain (wy, ey), substitute wy for as and ey for bs into the initial predicted load quantity function to obtain the final predicted load quantity function. For example, the final predicted load quantity function in the temperature range of [-20, 0] is Y = -0.41*X + 13.1; the final predicted load quantity function in the temperature range of [0, 24] is Y = -0.126*X + 13.1; the final predicted load quantity function in the temperature range of [24, 36] is Y = 0.668*X - 6.2; The power dispatching module is used to obtain the future temperature of the target power consumption area, calculate the predicted future load based on the future temperature and the final predicted load function; and perform power dispatching on the target power consumption area based on the predicted future load. The power dispatching module is configured with a power dispatching strategy, and the power dispatching strategy includes: Obtain the average temperature of the target power consumption area in the next week, and mark it as the future temperature; Substitute the future temperature into the final predicted load function in the corresponding segmented range to obtain the predicted future load; Dispatch the load of the target power consumption area in the next week with the predicted future load; In practical applications, the future temperature obtained is 5°C, and substituting it into the final predicted load function in the temperature range of [0, 24] is Y = -0.126*X + 13.1; the load is 124,700 kilowatts; dispatch the preset load of the target power consumption area in the next week with a load of 124,700 kilowatts. The function update module is used to obtain the real-time temperature and real-time load of the target power consumption area, and judge whether to update the final predicted load function based on the real-time temperature, real-time load and the final predicted load function. The function update module is configured with a function update strategy, and the function update strategy includes: Obtain the average temperature and load of the target power consumption area in the recent week, and mark them as the real-time temperature and real-time load respectively; Substitute the real-time temperature into the final predicted load function in the corresponding segmented range to obtain the predicted real-time load; Calculate the absolute value of the difference between the real-time load and the predicted real-time load, and mark it as the detection difference; Judge whether the detection difference exceeds the detection threshold. If it exceeds, add a third quantity of historical loads and historical temperatures, update the final predicted load function and optimize it using the particle swarm optimization algorithm; because as the times develop, electrical equipment will be updated and the load will change, so the final predicted load function will change and needs to be updated in real time; among them, the detection threshold can be set according to the maximum difference between the historical load and the load predicted by the historical temperature and the final predicted load function. In practical applications, obtain the average temperature = 2°C and load = 128,580 kilowatts of the target power consumption area in the recent week. Substitute 2°C into the final predicted load function in the temperature range of [0, 24] as Y = -0.126*X + 13.1; obtain the predicted real-time load = 128,480 kilowatts, calculate the detection difference as 10,000 kilowatts, obtain the detection threshold as 23,000 kilowatts, and 10,000 kilowatts is less than 23,000 kilowatts, so the final predicted load function does not need to be updated.

[0021] Embodiment 2, please refer to Figure 4 As shown, the present application provides a virtual power plant power dispatching method based on an intelligent optimization algorithm, including the following steps: Step S1, obtaining the historical load and historical temperature of a target power consumption area, and obtaining an initial predicted load function based on the historical load and historical temperature; Step S1 includes the following sub-steps: Step S101, obtaining a first quantity of historical loads and historical temperatures; where the historical load is the load within one week of the target power consumption area, and the historical temperature is the average temperature within one week corresponding to the historical load of the target power consumption area; Step S102, calculating the mean value of all historical loads corresponding to each historical temperature, and marking it as the historical load mean value; Step S103, establishing a plane rectangular coordinate system with the historical temperature as the X-axis and the historical load mean value as the Y-axis, marking it as the historical load mean coordinate system, and plotting the historical temperature and the historical load mean value into the historical load mean coordinate system to obtain a scatter plot, marking it as the historical load mean scatter plot.

[0022] Step S104, obtaining a first quantity of historical loads and historical temperatures; where the historical load is the load within one week of the target power consumption area, and the historical temperature is the average temperature within one week corresponding to the historical load of the target power consumption area; Step S105, calculating the mean value of all historical loads corresponding to each historical temperature, and marking it as the historical load mean value; Step S106, establishing a plane rectangular coordinate system with the historical temperature as the X-axis and the historical load mean value as the Y-axis, marking it as the historical load mean coordinate system, and plotting the historical temperature and the historical load mean value into the historical load mean coordinate system to obtain a scatter plot, marking it as the historical load mean scatter plot; Step S107, marking each coordinate point in the historical load mean scatter plot as a historical load mean coordinate point; Step S108, obtaining the number of historical load mean coordinate points, and marking it as n; Step S109, calculating the slope of the two historical load mean coordinate points with the closest abscissas, marking it as the load slope, and the calculation formula is: ; where Kx i is the slope of the i-th and (i - 1)-th historical load mean coordinate points, Y i is the ordinate of the i-th historical load mean coordinate, X i is the abscissa of the i-th historical load mean coordinate, Y i-1 is the ordinate of the (i - 1)-th historical load mean coordinate, X i-1 is the abscissa of the (i - 1)-th historical load mean coordinate, and at this time, the range of i is an integer from 2 to n, and the size of i corresponds to the size of the abscissa of the historical load mean coordinate; Step S110, calculate the difference between every two adjacent load slopes, marked as slope difference, and the calculation formula is: Kw i =Kx i+1 -Kx i ; where Kw i is the difference between the (i + 1)-th and i-th load slopes, Kx i+1 is the (i + 1)-th load slope, and Kx i is the i-th load slope. At this time, the range of i is an integer from 1 to n - 1, and the value of i corresponds to the abscissa of the historical load mean coordinate for generating the load slope; Step S111, determine whether all slope differences exceed the difference range threshold. If they exceed, calculate the error between the (i + 2)-th and i-th load slopes, marked as detection difference, and the calculation formula is: Kwd i =Kx i+2 -Kx i ; where Kwd i is the difference between the (i + 2)-th and i-th load slopes, and Kx i+2 is the (i + 2)-th load slope. At this time, the range of i is an integer from 1 to n - 2; Step S112, determine whether the detection error exceeds the difference range threshold. If it exceeds, obtain the abscissa of the i-th historical load mean coordinate corresponding to the i-th load slope, marked as abscissa threshold.

[0023] Step S113, count the number of abscissa thresholds, marked as m; Step S114, mark the abscissa thresholds in ascending order as T j , where j is an integer from 1 to m; Step S115, obtain the minimum abscissa of the historical load mean coordinate points, marked as T q , and obtain the maximum abscissa of the historical load mean coordinate points, marked as T z ; Step S116, obtain the segmentation ranges as: [Tq, T1], [T1, T2],..., [Tm, Tz]; Step S117, establish a plane rectangular coordinate system with historical temperature as the X-axis and historical load as the Y-axis, marked as historical load coordinate system. Plot the historical temperature and the corresponding historical load into the historical load coordinate system to obtain a scatter plot, marked as historical load scatter plot; Step S118, set the preset initial prediction load function as: Y = a β *X + b β , where a β and b β are the parameters of the preset initial prediction load function, and β is an integer from 1 to m + 1; Step S119: Based on the historical load scatter plot and the segmented ranges, perform multi-segment linear fitting to obtain the specific values of a and b in the preset initial predicted load function for each segmented range. Substitute the specific values of a and b into the preset initial predicted load function to obtain m + 1 initial predicted load functions. β and b β The specific values of a β and b β are substituted into the preset initial predicted load function to obtain m + 1 initial predicted load functions.

[0024] Step S2: Optimize the parameters of the initial predicted load function based on the particle swarm optimization algorithm to obtain the final predicted load function. Step S2 includes the following sub-steps: Step S201: Obtain the specific values of a and b of any one of the initial predicted load functions, and mark them as as and bs respectively. β and b β The specific values are marked as as and bs respectively. Step S202: Obtain the second quantity of particles, where the particle is represented as (w, e). Set the value range of w to [as - p1, as + p1], and the value range of e to [bs - p2, bs + p2]; p1 is the w range constant, and p2 is the e range constant. Step S203: Calculate the fitness of each particle as: ; where MSE is the fitness of each particle, X i is the historical temperature, Y i is the historical load corresponding to X i and N is the quantity of X i . Step S204: Denote the particle with the minimum fitness as the optimal solution. Step S205: Update the particle velocity and position. The particle velocity update formula is: Vn α =u*vo α +c1*r1*(pdb α -p α )+c2*r2*(pdg α -p α ); where Vn α is the velocity of particle α after update; vo α is the velocity of particle α before update; u is the inertia factor; c1 and c2 are the learning factors; r1 and r2 are random numbers between 0 and 1; pdb α is the optimal solution of particle α; pdg α is the global optimal solution; p α is the current particle position. Step S206: The particle position update formula is: pn α =p α +Vn α ; where pnα is the position after particle update; Step S207, repeatedly update the particle velocity and position until the optimal solution of the particle no longer changes; Step S208, obtain the optimal solution of the global particle, marked as (wy, ey); Step S209, substitute wy for as and ey for bs into the initial predicted load function to obtain the final predicted load function.

[0025] Step S3, obtain the future temperature of the target power consumption area, calculate the predicted future load based on the future temperature and the final predicted load function; perform power dispatching for the target power consumption area based on the predicted future load; Step S3 includes the following sub-steps: Step S301, obtain the average temperature of the target power consumption area in the next week, marked as the future temperature; Step S302, substitute the future temperature into the final predicted load function in the corresponding segmented range to obtain the predicted future load; Step S303, dispatch the load of the target power consumption area in the next week based on the predicted future load.

[0026] Step S4, obtain the real-time temperature and real-time load of the target power consumption area, and judge whether to update the final predicted load function based on the real-time temperature, real-time load and the final predicted load function; Step S4 includes the following sub-steps: Step S401, obtain the average temperature and load of the target power consumption area in the recent week, marked as the real-time temperature and real-time load respectively; Step S402, substitute the real-time temperature into the final predicted load function in the corresponding segmented range to obtain the predicted real-time load; Step S403, calculate the absolute value of the difference between the real-time load and the predicted real-time load, marked as the detection difference; Step S404, judge whether the detection difference exceeds the detection threshold. If it exceeds, add a third quantity of historical load and historical temperature, update the final predicted load function and optimize it using the particle swarm optimization algorithm.

[0027] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. Among them, the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk. These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured article including an instruction device, and the instruction device implements the process Figure 1 in one process or multiple processes and / or blocks Figure 1 specified functions in one block or multiple blocks.

[0028] In the embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For another example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some communication interfaces, and the indirect coupling or communication connection of the devices or units can be in an electrical, mechanical or other form.

Claims

1. A virtual power plant power dispatch method based on an intelligent optimization algorithm, characterized in that, It includes the following steps: Obtain the historical load and historical temperature of the target power consumption area, and obtain the initial predicted load function based on the historical load and historical temperature; Optimize the parameters of the initial predicted load function based on the particle swarm optimization algorithm to obtain the final predicted load function; Obtain the future temperature of the target power consumption area, and calculate the predicted future load based on the future temperature and the final predicted load function; perform power dispatching on the target power consumption area based on the predicted future load; Obtain the real-time temperature and real-time load of the target power consumption area, and determine whether to update the final predicted load function based on the real-time temperature, real-time load and the final predicted load function.

2. The virtual power plant power dispatching method based on the intelligent optimization algorithm according to claim 1, wherein, Obtaining the historical load and historical temperature of the target power consumption area includes the following sub-steps: Obtain the first quantity of historical loads and historical temperatures; where the historical load is the load within one week of the target power consumption area, and the historical temperature is the average temperature within one week corresponding to the historical load of the target power consumption area; Calculate the mean value of all historical loads corresponding to each historical temperature, and mark it as the historical load mean value; Establish a plane rectangular coordinate system with the historical temperature as the X-axis and the historical load mean value as the Y-axis, mark it as the historical load mean coordinate system, plot the historical temperature and the historical load mean value into the historical load mean coordinate system to obtain a scatter plot, and mark it as the historical load mean scatter plot.

3. The virtual power plant power dispatch method based on the intelligent optimization algorithm according to claim 2, characterized in that Obtaining the initial predicted load function based on the historical load and historical temperature includes the following sub-steps: Mark each coordinate point in the historical load mean scatter plot as a historical load mean coordinate point; Obtain the number of historical load mean coordinate points, and mark it as n; Calculate the slope of the two historical load mean coordinate points with the closest abscissas among all, and mark it as the load slope. The calculation formula is: ； Among them, Kx i is the slope of the coordinate points of the mean historical load between the i-th and (i - 1)-th points, Y i is the ordinate of the coordinate of the i-th mean historical load, X i is the abscissa of the coordinate of the i-th mean historical load, Y i-1 is the ordinate of the coordinate of the (i - 1)-th mean historical load, X i-1 is the abscissa of the coordinate of the (i - 1)-th mean historical load. At this time, the range of i is an integer from 2 to n, and the value of i corresponds to the value of the abscissa of the coordinate of the mean historical load.

4. The virtual power plant power dispatching method based on the intelligent optimization algorithm according to claim 3, characterized in that Obtaining the initial predicted load function based on the historical load and historical temperature further includes the following sub-steps: Calculate the difference between all adjacent two load slopes, and mark it as the slope difference. The calculation formula is: Kw i =Kx i+1 -Kx i ; where Kw i is the difference between the (i + 1)-th and i-th load slopes, and Kx i+1 is the (i + 1)-th load slope, and Kx i is the i-th load slope. Here, the range of i is an integer from 1 to n - 1, and the value of i corresponds to the abscissa of the historical load mean coordinate for generating the load slope. Determine whether all slope differences exceed the difference range threshold. If so, calculate the error between the slopes of the (i + 2)-th and i-th loads, which is marked as the detected difference. The calculation formula is: Kwd i =Kx i+2 -Kx i ; Where Kwd i is the difference between the (i + 2)-th and i-th load slopes, and Kx i+2 is the (i + 2)-th load slope, where i ranges from 1 to n - 2 as an integer; Determine whether the detection error exceeds the difference range threshold. If it exceeds, obtain the abscissa of the i-th historical load mean coordinate corresponding to the i-th load slope, and mark it as the abscissa threshold.

5. The virtual power plant power dispatching method based on the intelligent optimization algorithm according to claim 4, characterized in that, Obtaining the initial predicted load function based on the historical load and historical temperature further includes the following sub-steps: Count the number of abscissa thresholds and mark it as m; The abscissa thresholds are sequentially labeled as T from smallest to largest j , where j is an integer from 1 to m; Obtain the minimum value of the abscissa of the historical load mean coordinate point, denoted as T q , obtain the maximum value of the abscissa of the historical load mean coordinate point, denoted as T z ; Obtain the segmented ranges as: [Tq, T1], [T1, T2],..., [Tm, Tz].

6. The virtual power plant power dispatch method based on the intelligent optimization algorithm according to claim 5, wherein, Obtaining the initial predicted load function based on the historical load and historical temperature further includes the following sub-steps: Establish a plane rectangular coordinate system with the historical temperature as the X-axis and the historical load as the Y-axis, mark it as the historical load coordinate system, plot the historical temperature and the corresponding historical load into the historical load coordinate system to obtain a scatter plot, and mark it as the historical load scatter plot; Set the preset initial prediction load function as: Y = a β *X + b β , where a β and b β are parameters of the preset initial prediction load function, and β is an integer from 1 to m + 1; Perform multi-segment linear fitting based on the historical load scatter plot and the segmented range to obtain the specific values of a β and b β in the preset initial predicted load function. Substitute the specific values of a β and b β into the preset initial predicted load function to obtain m + 1 initial predicted load functions.

7. The virtual power plant power dispatch method based on the intelligent optimization algorithm according to claim 6, characterized in that Optimizing the parameters of the initial predicted load function based on the particle swarm optimization algorithm to obtain the final predicted load function includes the following sub-steps: Obtain the a of any initial predicted load function β and b β Specific values, marked as as and bs respectively; Obtain the second quantity of particles, where the particle is represented as (w, e), and set the value range of w to [as - p1, as + p1], and the value range of e to [bs - p2, bs + p2]; p1 is the w range constant, and p2 is the e range constant; The fitness of each particle is calculated as follows: ； where MSE is the fitness of each particle, X i is the historical temperature, Y i is the historical load corresponding to X i , and N is the number of X i ; The particle with the minimum fitness is denoted as the optimal solution; The velocity and position of the particle are updated. The particle velocity update formula is as follows: Vn α =u*vo α +c1*r1*(pdb α -p α )+c2*r2*(pdg α -p α ) Among them, Vn α is the velocity of particle α after update; vo α is the velocity of particle α before update; u is the inertia factor; c1 and c2 are learning factors; r1 and r2 are random numbers between 0 and 1; pdb α is the optimal solution of particle α; pdg α is the global optimal solution; p α is the current particle position; The particle position update formula is: pn α = p α + Vn α ; where pn α is the position of the particle after update; Repeat the update of the particle velocity and position until the optimal solution of the particle no longer changes; Obtain the optimal solution of the global particle and mark it as (wy, ey); Substitute wy for as and ey for bs into the initial predicted load function to obtain the final predicted load function.

8. The virtual power plant power dispatch method based on the intelligent optimization algorithm according to claim 7, characterized in that, Based on the future temperature and the final predicted load function, the following sub-steps are performed to calculate the predicted future load: Obtain the average temperature of the target power consumption area in the next week and mark it as the future temperature; Substitute the future temperature into the final predicted load function in the corresponding segmented range to obtain the predicted future load; Dispatch the load of the target power consumption area in the next week based on the predicted future load.

9. The virtual power plant power dispatch method based on the intelligent optimization algorithm according to claim 8, characterized in that Based on the real-time temperature, real-time load and the final predicted load function, the steps to determine whether to update the final predicted load function are as follows: Obtain the average temperature and load of the target power consumption area in the recent week, and mark them as the real-time temperature and real-time load respectively; Substitute the real-time temperature into the final predicted load function in the corresponding segmented range to obtain the predicted real-time load; Calculate the absolute value of the difference between the real-time load and the predicted real-time load, and mark it as the detection difference; Judge whether the detection difference exceeds the detection threshold. If it exceeds, add a third quantity of historical load and historical temperature, update the final predicted load function and optimize it using the particle swarm optimization algorithm.

10. A virtual power plant power dispatching system based on an intelligent optimization algorithm, which is used to implement the virtual power plant power dispatching method based on an intelligent optimization algorithm according to any one of claims 1-9, characterized in that, It includes a function acquisition module, a function optimization module, a power dispatch module and a function update module; The function acquisition module is used to obtain the historical load and historical temperature of the target power consumption area, and obtain the initial predicted load function based on the historical load and historical temperature; The function optimization module is used to optimize the parameters of the initial predicted load function based on the particle swarm optimization algorithm to obtain the final predicted load function; The power dispatch module is used to obtain the future temperature of the target power consumption area, calculate the predicted future load based on the future temperature and the final predicted load function; perform power dispatch on the target power consumption area based on the predicted future load; The function update module is used to obtain the real-time temperature and real-time load of the target power consumption area, and judge whether to update the final predicted load function based on the real-time temperature, real-time load and the final predicted load function.

Images