Demand response management method considering new energy prediction smoothing technology

Abstract

The invention relates to a demand response management method considering a new energy prediction smoothing technology, and belongs to the field of wind power prediction and demand correlation. According to the method, a prediction-filtering strategy is provided, and the whale algorithm WOA and the LSTM are combined to carry out short-term output prediction; and then, a Hampel-Butterworth-SG filtering strategy containing outlier regression and specific dangerous frequency band elimination is introduced, and then, the Hampel-Butterworth-SG filtering strategy containing outlier regression and specific dangerous frequency band elimination is introduced. And then, calculating a scheduling pre-configuration scheme of peak clipping and valley filling type energy storage according to the short-term output predicted value. And finally, according to the predicted new energy output and a time-of-use electricity price mechanism, under the action of a Logistic function, optimistic response and pessimistic response factors, calculating the demand response dynamic change of the user. At different electricity price moments, the user can carry out demand response dynamic adjustment according to the price, so as to realize optimal scheduling of resources.

Description

Technical Field

[0001] The present invention belongs to the field related to wind power prediction and demand, and particularly relates to a demand response management method considering new energy prediction smoothing technology. Background Art

[0002] New energy prediction has become an important guarantee for the optimal dispatching and stable operation of a new power system. Especially in terms of short-term voltage stability, it can guide a series of decision-making and management arrangements. However, renewable energy sources such as wind power have inherent characteristics such as randomness and uncertainty, which greatly restrict the utilization rate of wind power and the stability of the power grid system. And new energy prediction can reduce the impact caused by this randomness. Therefore, multi-time scale wind power prediction has been widely applied to the dispatching and control of wind power systems. Therefore, combining new energy prediction with energy storage is of great significance for the development of large-scale new energy.

[0003] Due to the inherent attributes of wind power output such as randomness and uncertainty, large-scale grid connection is bound to bring problems such as large power fluctuations, reduced power quality, and even serious power system disorders. New energy prediction can not only provide a basis for the dispatching and allocation of energy storage systems. More importantly, power producers can provide the total amount of new energy for the power spot market, and participating users can perform pre-demand response according to time-of-use electricity prices and electricity quantity reports. The benefit of this demand response is that it can reduce the uncertainty of the real-time spot market caused by the randomness of new energy and greatly reduce the revenue loss caused by uncertainty. At present, there are mainly four types of methods in new energy prediction, namely: physical models, conventional statistical analysis, artificial intelligence methods, and combined prediction model methods.

[0004] Energy prediction can also guide the demand response of the power spot market. When the output of new energy continues to expand and gradually participates in the power spot market, the prediction of demand response becomes particularly important. In recent years, China has gradually launched 8 pilot cities for the power spot market, summarizing and accumulating important data on the spot market under the participation mode of new energy. The research on combining new energy prediction with energy storage pre-dispatching and allocation and power spot market pre-regulation is insufficient, which is not conducive to promoting the development of new energy, energy storage, and the spot market to a large extent. Summary of the Invention

[0005] The purpose of the present invention is to provide a demand response management method considering new energy prediction smoothing technology to effectively guide the predicted output for the demand response of new energy participating in the power spot market.

[0006] To achieve the above purpose, the technical solution of the present invention is: a demand response management method considering new energy prediction smoothing technology, including:

[0007] Combine the Whale Optimization Algorithm (WOA) with Long Short-Term Memory (LSTM) to perform short-term output prediction;

[0008] Introduce a Hampel-Butterworth-SG filtering strategy with outlier regression and predetermined dangerous frequency band elimination to process the short-term output prediction values;

[0009] Calculate the dispatching pre-configuration plan of peak shaving and valley filling energy storage based on the processed short-term output prediction values, and predict new energy output and time-of-use electricity price mechanism;

[0010] According to the predicted new energy output and time-of-use electricity price mechanism, calculate the dynamic change of users' demand response under the action of the Logistics function, optimistic response factor and pessimistic response factor.

[0011] In an embodiment of the present invention, at different electricity price moments, users will dynamically adjust their demand response according to the price level to achieve optimal resource scheduling.

[0012] In an embodiment of the present invention, the method includes the following steps:

[0013] Step S1, capture the original wind power output or related time series W(t);

[0014] Step S2, set the initial parameters of Long Short-Term Memory (LSTM) and Whale Optimization Algorithm (WOA), and calculate the optimal result;

[0015] Step S3, according to the theoretical method of WOA, screen the optimal result respectively according to spiral contraction, surrounding enclosure, and random exploration until the maximum number of iterations g max ;

[0016] Step S4, obtain the optimal LSTM parameters at a predetermined time scale, and calculate the predicted output situation at the predetermined time scale at this time;

[0017] Step S5, realize outlier regression of the predicted output through Hampel, then obtain the stable output power after attenuation in the predetermined frequency band of the Butterworth notch filter, that is, the Butterworth notch filter; finally, use the S-G filtering algorithm to calculate the low-frequency power and remaining fluctuations of the predicted output that meet the grid connection standard; analyze the data through MSE and MAE;

[0018] Step S6: The new energy operation and electricity seller maximize their profits by seeking the optimal demand response price, setting the objective function and assessment deviation of the new energy, and calculating the charging and discharging power of the energy storage element at this time;

[0019] Step S7: Based on the time-of-use electricity price, the user's dynamic demand response under the forecast condition is calculated through the Logistics function, the optimistic response curve, and the pessimistic response curve, and the charging and discharging power of the energy storage element before and after adjustment is calculated at this time.

[0020] In one embodiment of the present invention, step S3 is specifically implemented as follows:

[0021] Step S31: In the prey encirclement stage, assuming that the optimal solution is used as the target prey position, the behavior of the whale moving toward the optimal target position has the following equation:

[0022]

[0023] Where n represents the number of iterations; and Respectively represent the updated coefficient vector; and They represent the best position vector of the whale at the current moment and the current position vector of the whale respectively;

[0024] Coefficient vector They are obtained by the following equations:

[0025]

[0026] in, represents a random vector in the interval [0,1]; represents a coefficient vector that decreases linearly from 2 to 0 as the number of iterations increases;

[0027] The mathematical model equation of the spiral position update mechanism is as follows:

[0028]

[0029] in, represents the distance between the whale and its prey; b is a constant used to define the spiral shape; l is a random number between (-1,1); in order to simulate the whale's choice of hunting in a shrinking or spiraling manner, it is assumed that the whale chooses to hunt with probability p i <0.5 executes shrinking orbit, otherwise performs spiral position update; in the prey search phase, let The random value of makes the whale expand its search range and update the whale's position through random positions. The corresponding equation is as follows:

[0030]

[0031] Among them, represents the position vector randomly determined by the whale in the current population;

[0032] Step S32, when , randomly select a solution to update the position of the whale; when , select the current optimal solution to update the position of the whale; in addition, WOA switches freely between the spiral position update and the shrinking wrap-around mechanisms according to the probability of p i . During the continuous iteration process, the mean square deviation SD between the predicted value and the actual value is used as the fitness; assuming that the number of iterations is g, it is judged whether the termination condition is satisfied, that is, when g = g max , the iteration stops.

[0033] In an embodiment of the present invention, step S5 is specifically implemented as follows:

[0034] Step S51, there are fluctuations and deviations in the predicted power output, so it needs to be preprocessed; it is considered that when the deviation value W n of the sample that is more than 2 standard deviations different from the median is > 2, this point is a "single-point" outlier; replace the outlier of the "single-point" with the median of the surrounding window, that is where the data set W(t) = {w 1 (t), w 2 (t), …, w n (t)}; W = {w 1 (t) - median[W(t)], w 2 (t) - median[W(t)], …, w n (t) - median[W(t)]}, median(·) represents taking the median; then, use the Butterworth notch filter as the method to eliminate the predetermined frequency band;

[0035] Step S52, the preprocessed predicted power is not the actual grid-connected power, and the S-G filtering algorithm needs to be used for filtering; in the case of data in the 2M + 1 interval, use the polynomial p i for fitting:

[0036]

[0037] where, use the polynomial SGF i for fitting; F is the number of local fittings of the polynomial; λ is the fitting order; a λ is the local fitting coefficient of the polynomial; i represents the length of the data;

[0038] In scenarios with strict power specifications, the upper limits of volatility at the 1-minute and 30-minute time scales are 2% and 7% respectively, considering the 30-minute volatility evaluation index; in practical applications, the 30-minute constraint condition is relatively harsh, and the percentage of the over-limit state of volatility is introduced to describe the filtering effect:

[0039]

[0040] where s is the volatility; s max represents the upper limit of volatility; m represents the length of the power data at this time.

[0041] In an embodiment of the present invention, the equivalent transfer function of the Butterworth notch filter is as follows:

[0042]

[0043] where ε(s) and x(s) respectively represent the input and output of the Butterworth notch filter; ω 0 represents the center frequency of the Butterworth notch filter; μ and τ respectively represent the iteration step and the sampling interval; A represents the amplitude corresponding to the Butterworth notch filter; s represents the complex frequency domain variable of the system.

[0044] In an embodiment of the present invention, in step S7, in the participation of new energy in the electricity spot market, the introduction of time-of-use electricity prices will inevitably cause changes in the dynamic demand response of users. The changes in demand response caused by such electricity price differences are within a certain range of fluctuations, which are called optimistic response fluctuations and pessimistic response fluctuations. According to the optimistic response fluctuations and pessimistic response fluctuations, the entire demand response is divided into three regions, which are respectively called: static region, dynamic region, and saturation region; the Logistic function can reliably improve the fitting accuracy of time-of-use electricity prices to dynamic demand response; and in the dynamic region, users will be more willing to make a dynamic adjustment of demand response when participating in the electricity market. Therefore, the membership value of the optimistic response is introduced to reflect a probability to describe this dynamic response situation.

[0045] In an embodiment of the present invention, the Logistic function model is as follows:

[0046]

[0047] where Δp w,t represents the electricity spot market price difference; φ(·) represents the load transfer rate function; α represents the range of function values; δ represents the abscissa corresponding to the function value (α / 2 + β); μ represents the known quantity in the Logistic function; β is a variable parameter used to move the function curve up and down; when the price difference Δp w,t = 0, it means that the time-of-use electricity price strategy is not adopted at this time, so there is no dynamic demand response of users.

[0048] In one embodiment of the present invention, the dynamic response equation introducing the membership value of the optimistic response is expressed as follows:

[0049]

[0050] Wherein, represents the peak-valley transfer rate of renewable energy load; and respectively represent the peak-valley load transfer ratios of demand response under optimistic and pessimistic conditions; α w,t and β w,t respectively represent the critical value points of the price difference division area; during the entire dynamic demand response process, in the static area, due to the too small electricity price difference at this time, the power behavior of individual users in the entire power market has great uncertainty, so it is considered that this process is described by the mean value of optimistic and pessimistic responses; while in the dynamic area, as the electricity price difference continuously increases, users fully participate in the power market based on the electricity price difference, and users are more optimistic and actively respond to the entire process, and its essence is defined as an optimistic response model; while in the saturation area, the response behavior of users has reached a threshold, so it is described by the maximum value of the load transfer;

[0051] Based on the dynamic response equation introducing the membership value of the optimistic response, the peak-flat and flat-valley load transfer rates are solved The load transfer amount Ω generated under demand response w,t has the following equation:

[0052]

[0053] Wherein, and respectively represent the average load of each time period before the implementation of time-of-use electricity price; t p 、t f 、t v respectively represent the peak period, flat period and valley period of time-of-use electricity price.

[0054] The present invention also provides a computer-readable storage medium, on which computer program instructions that can be run by a processor are stored. When the processor runs the computer program instructions, the method steps as described in any one of the above can be implemented.

[0055] Compared with the prior art, the present invention has the following beneficial effects: The method of the present invention proposes a prediction-filtering strategy. First, the whale optimization algorithm (WOA) is combined with the long short-term memory (LSTM) to perform short-term output prediction. Then, a Hampel-Butterworth-Savitzky-Golay (Hampel-Butterworth-SG) filtering strategy including outlier regression and specific dangerous frequency band elimination is introduced. Then, according to the short-term output prediction value, a dispatching pre-configuration scheme for peak shaving and valley filling energy storage is calculated. Finally, according to the predicted new energy output and the time-of-use electricity price mechanism, under the action of the Logistics function, optimistic response, and pessimistic response factors, the dynamic change of the user's demand response is calculated. At different electricity price moments, users will dynamically adjust the demand response according to the price level to achieve optimal resource scheduling. The method of the present invention can effectively guide the demand response of the predicted output for the new energy participating in the electricity spot market. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a flowchart of the method of the present invention.

[0057] Figure 2 It is a three-dimensional position vector and their potential position diagram of the present invention.

[0058] Figure 3 It is a contraction and surrounding diagram of the present invention.

[0059] Figure 4 It is a spiral position update diagram of the present invention.

[0060] Figure 5 It is an improved prediction and filtering diagram of the present invention.

[0061] Figure 6 It is a dynamic demand response diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0062] The technical solutions of the present invention will be specifically described below with reference to the drawings.

[0063] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present application belongs.

[0064] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0065] The present invention provides a demand response management method considering new energy prediction smoothing technology, including:

[0066] Combining the Whale Optimization Algorithm (WOA) with Long Short-Term Memory (LSTM) for short-term output prediction;

[0067] Introducing a Hampel-Butterworth-SG filtering strategy including outlier regression and predetermined dangerous frequency band elimination to process the short-term output prediction values;

[0068] Calculating a scheduling pre-configuration plan for peak shaving and valley filling energy storage based on the processed short-term output prediction values, predicting new energy output and time-of-use electricity price mechanism;

[0069] Calculating the dynamic change of user demand response under the action of the Logistics function, optimistic response factor and pessimistic response factor according to the predicted new energy output and time-of-use electricity price mechanism.

[0070] At different electricity price moments, users will dynamically adjust demand response according to the price level to achieve optimal resource scheduling.

[0071] The following is the specific implementation process of the present invention.

[0072] As Figure 1 shown, this embodiment provides a demand response management method considering new energy prediction smoothing technology, realizing new energy prediction smoothing, which specifically includes the following steps:

[0073] Step S1: Capture the original wind power output or related time series W(t);

[0074] Step S2: Set the initial parameters of Long Short-Term Memory (LSTM) and Whale Optimization Algorithm (WOA), and calculate the optimal result;

[0075] Step S3: According to the theoretical method of WOA, screen the optimal result according to spiral contraction, surrounding encirclement, and random exploration respectively until the maximum number of iterations g max ;

[0076] Step S4: Obtain the optimal LSTM parameters at a predetermined time scale, and calculate the predicted output situation at this predetermined time scale;

[0077] Step S5: Implement outlier regression for the predicted output power through Hampel. Then, after attenuation in the predetermined frequency band of the Butterworth notch filter, a stable output power is obtained. Finally, use the S-G filtering algorithm to calculate the low-frequency power and remaining fluctuations of the predicted output power that meet the grid connection standard; analyze the data through MSE and MAE.

[0078] Step S6: New energy operators and electricity retailers maximize their interests by seeking the optimal demand response price, set the objective function and assessment deviation of new energy, and calculate the charging and discharging power conditions of energy storage elements at this time.

[0079] Step S7: Based on time-of-use electricity prices, through the Logistics function, as well as the optimistic response curve and pessimistic response curve, calculate the dynamic demand response of users under the predicted situation, and at the same time calculate the charging and discharging power of the energy storage element before and after adjustment at this time.

[0080] In this embodiment, as Figures 2 - 4 shown, step S3 is specifically implemented as follows:

[0081] Step S31: In the prey surrounding stage, assuming that the optimal solution is the position of the target prey, the behavior of the whale moving towards the optimal target position has the following equation:

[0082]

[0083] where n represents the number of iterations; and respectively represent the updated coefficient vectors; and respectively represent the best position vector of the whale at the current moment and the position vector of the current whale;

[0084] The coefficient vector is respectively obtained by the following equations:

[0085]

[0086] where, represents a random vector within the interval [0, 1]; represents a coefficient vector that linearly decreases from 2 to 0 as the number of iterations increases;

[0087] The mathematical model equation of the spiral position update mechanism is as follows:

[0088]

[0089] where, represents the distance between the whale and its prey; b is a constant used to define the spiral shape; l is a random number between (-1, 1); to simulate the hunting method of the whale choosing shrinking encirclement or spiral encirclement, it is assumed that the whale performs shrinking encirclement with probability p i < 0.5 to perform shrinking encirclement, otherwise perform spiral position update; in the stage of searching for prey, let a random value of makes the whale expand the search range, and updates the position of the whale through a random position. The corresponding equation is as follows:

[0090]

[0091] where, represents the position vector randomly determined by the whale in the current population;

[0092] Step S32, when a solution is randomly selected to update the position of the whale; when the current optimal solution is selected to update the position of the whale; in addition, WOA freely switches between the two mechanisms of spiral position update and shrinking encirclement according to the probability p i ; during the continuous iteration process, the mean square deviation SD between the predicted value and the actual value is used as the fitness; assuming that the number of iterations is g, it is judged whether the termination condition is satisfied, that is, when g = g max the iteration stops.

[0093] In this embodiment, as Figure 5 shown, the specific implementation of step S5 is as follows:

[0094] There are fluctuations and deviations in the predicted power output, so it needs to be preprocessed; it is considered that when the deviation value W n > 2 of the sample that differs from the median by more than 2 standard deviations, this point is a "single-point" outlier; replace the outlier of the "single-point" with the median of the surrounding window, that is where the data set W(t) = {w 1 (t), w 2 (t), …, w n (t)}; W = {w 1 (t) - median[W(t)], w 2 (t) - median[W(t)], …, w n (t) - median[W(t)]}, median(·) represents taking the median; then, a Butterworth notch filter is used as the method to eliminate the predetermined frequency band;

[0095] Step S52: The predicted power after preprocessing is not the actual grid-connected power, and the S-G filtering algorithm needs to be used for filtering (SG (Savitzky-Golay) is a filtering method based on local polynomial least squares fitting in the time domain); in the case of data in the 2M+1 interval, the polynomial p i is used for fitting:

[0096]

[0097] Among them, the polynomial SGF i is used for fitting; F is the number of local fittings of the polynomial; λ is the fitting order; a λ is the local fitting coefficient of the polynomial; i represents the length of the data;

[0098] In scenarios with strict power specifications, the upper limits of volatility at the 1min and 30min time scales are 2% and 7% respectively, and the 30min volatility evaluation index is considered; in practical applications, the 30min constraint condition is relatively harsh, and the percentage of the volatility overrun state is introduced to describe the filtering effect:

[0099]

[0100] Among them, s is the volatility; s max represents the upper limit of the volatility; m represents the length of the power data at this time.

[0101] In an embodiment of the present invention, the equivalent transfer function of the Butterworth notch filter is as follows:

[0102]

[0103] Among them, ε(s) and x(s) respectively represent the input and output of the Butterworth notch filter; ω 0 represents the center frequency of the Butterworth notch filter; μ and τ respectively represent the iteration step and the sampling interval; A represents the amplitude corresponding to the Butterworth notch filter; s represents the complex frequency domain variable of the system.

[0104] In this embodiment, as Figure 6As shown, in step S7, in the participation of new energy in the electricity spot market, the introduction of time-of-use electricity prices will inevitably cause changes in the dynamic demand response of users. The changes in demand response caused by such electricity price differences are within a certain range of fluctuations, which are called optimistic response fluctuations and pessimistic response fluctuations. According to the optimistic response fluctuations and pessimistic response fluctuations, the entire demand response is divided into three regions, which are respectively called: static region, dynamic region, and saturation region; The Logistic function can reliably improve the fitting accuracy of time-of-use electricity prices to dynamic demand response; In the dynamic region, users are more willing to make dynamic adjustments to demand response when participating in the electricity market. Therefore, the membership value of optimistic response is introduced to reflect a probability to describe this dynamic response situation.

[0105] The Logistic function model is as follows:

[0106]

[0107] Among them, Δp w,t represents the electricity spot market price difference; φ(·) represents the load transfer rate function; α represents the range of function values; δ represents the abscissa corresponding to the function value (α / 2 + β); μ represents the known quantity in the Logistic function; β is a variable parameter used to move the function curve up and down; When the price difference Δp w,t = 0, it means that the time-of-use electricity price strategy is not adopted at this time, so the dynamic demand response of users is not involved.

[0108] The dynamic response equation introducing the membership value of optimistic response is expressed as follows:

[0109]

[0110] Among them, represents the peak-valley transfer rate of renewable energy load; and respectively represent the peak-valley load transfer ratios of demand response under optimistic and pessimistic conditions; α w,t and β w,t respectively represent the critical value points of the price difference division region; In the whole process of dynamic demand response, in the static region, due to the too small electricity price difference at this time, the electricity behavior of individual users in the whole electricity market has great uncertainty. Therefore, it is considered that the process is described by the mean value of optimistic and pessimistic responses; In the dynamic region, as the price difference of electricity prices continues to increase, users fully participate in the electricity market based on the electricity price difference, and users are more optimistic and actively respond to the whole process, and its essence is defined as an optimistic response model; When in the saturation region, the response behavior of users has reached a threshold, so it is described by the maximum value of the load transfer.

[0111] The peak-valley and flat-valley load transfer rates are solved by a dynamic response equation based on the membership values introducing optimistic responses. The load transfer amount Ω generated under demand response w,t has the following equation:

[0112]

[0113] where and respectively represent the average load of each time period before the implementation of time-of-use electricity price; t p , t f , t v respectively represent the peak period, flat period and valley period of time-of-use electricity price.

[0114] The present invention also provides a computer-readable storage medium, on which computer program instructions capable of being run by a processor are stored. When the processor runs the computer program instructions, the method steps as described in any of the above can be implemented.

[0115] Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0116] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the flows and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0117] 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 generate a manufactured article including instruction means, and the instruction means implements the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1The functions specified in one or more boxes.

[0118] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide for implementing the steps of the functions specified in one Figure 1 one process or more processes and / or boxes Figure 1 step of the functions specified in one or more boxes.

[0119] As mentioned above, it is only the preferred embodiment of the present invention, and it is not a limitation of the present invention in other forms. Any person skilled in the art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes. However, any simple modification, equivalent change, and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution content of the present invention still belong to the protection scope of the technical solution of the present invention.

Claims

1. A demand response management method considering new energy forecast smoothing technology, characterized in that: include: Combine the Whale Optimization Algorithm (WOA) with the Long Short-Term Memory (LSTM) to perform short-term output prediction; A Hampel-Butterworth-SG filtering strategy with outlier regression and predetermined dangerous frequency band elimination is introduced to process the short-term output forecast value. Calculate the dispatch pre-configuration plan of peak-shaving and valley-filling energy storage based on the processed short-term output forecast value, and predict the output of new energy and time-of-use electricity price mechanism; According to the predicted new energy output and time-of-use electricity price mechanism, the dynamic changes of user demand response are calculated under the action of Logistics function, optimistic response factors and pessimistic response factors.

2. A demand response management method considering new energy forecast smoothing technology according to claim 1, characterized in that: At different electricity price times, users will dynamically adjust their demand response based on price levels to achieve optimal resource scheduling.

3. A demand response management method considering new energy forecast smoothing technology according to claim 1, characterized in that: The method comprises the following steps: Step S1, capturing the original wind power output or related time series W(t); Step S2, setting initial parameters for the long short-term memory (LSTM) and the whale optimization algorithm (WOA), and calculating the optimal result; Step S3: According to the theoretical method of WOA, the optimal result is screened according to spiral contraction, encirclement, and random exploration until the maximum number of iterations g is reached. max ; Step S4, obtaining the optimal LSTM parameters at a predetermined time scale, and calculating the predicted output at the predetermined time scale at this time; Step S5, implementing outlier regression of the predicted output through Hampel, and then obtaining a stable output power after attenuation in a predetermined frequency band of a Butterworth notch filter; finally, using the SG filtering algorithm to calculate the low-frequency power and residual fluctuation of the predicted output that meet the grid connection standard; and analyzing the data through MSE and MAE; Step S6: The new energy operation and electricity seller maximize their profits by seeking the optimal demand response price, setting the objective function and assessment deviation of the new energy, and calculating the charging and discharging power of the energy storage element at this time; Step S7: Based on the time-of-use electricity price, the user's dynamic demand response under the forecast condition is calculated through the Logistics function, the optimistic response curve, and the pessimistic response curve, and the charging and discharging power of the energy storage element before and after adjustment is calculated at this time.

4. A demand response management method considering new energy forecast smoothing technology according to claim 3, characterized in that: Step S3 is specifically implemented as follows: Step S31: In the prey encirclement stage, assuming that the optimal solution is used as the target prey position, the behavior of the whale moving toward the optimal target position has the following equation: Where n represents the number of iterations; and Respectively represent the updated coefficient vector; and They represent the best position vector of the whale at the current moment and the current position vector of the whale respectively; Coefficient vector They are obtained by the following equations: in, represents a random vector in the interval [0,1]; represents a coefficient vector that decreases linearly from 2 to 0 as the number of iterations increases; The mathematical model equation of the spiral position update mechanism is as follows: in, represents the distance between the whale and its prey; b is a constant used to define the spiral shape; l is a random number between (-1,1); in order to simulate the whale's choice of hunting in a shrinking or spiraling manner, it is assumed that the whale chooses to hunt with probability p i <0.5 executes shrinking orbit, otherwise performs spiral position update; in the prey search phase, let The random value of makes the whale expand its search range and update the whale's position through random positions. The corresponding equation is as follows: in, represents the randomly determined position vector of the whale in the current population; Step S32: When , a solution is randomly selected to update the whale's position; when When the current optimal solution is selected, the position of the whale is updated; in addition, WOA updates the position of the whale according to p i The probability of freely switching between the spiral position update and the shrinking wrapping mechanism; in the process of continuous iteration, the mean square error SD between the predicted value and the actual value is used as the fitness; assuming that the number of iterations is g, it is determined whether the termination condition is met, that is, when g = g max The iteration stops when .

5. A demand response management method considering new energy forecast smoothing technology according to claim 3, characterized in that: Step S5 is specifically implemented as follows: Step S51: The predicted power output has fluctuations and deviations, so it needs to be preprocessed; it is considered that when the deviation value W of the sample is greater than 2 standard deviations from the median n When >2, the point is a "single point" outlier; the "single point" outlier is replaced by the median of the surrounding window, that is, The data set W(t) = {w1(t),w2(t),…,w n (t)}; W={w1(t)-median[W(t)],w2(t)-median[W(t)],…,w n (t)-median[W(t)]}, where median(·) represents the median value; then, a Butterworth notch filter is used as a method to eliminate the predetermined frequency band; Step S52: The predicted power after preprocessing is not the actual grid-connected power, and the SG filtering algorithm needs to be used for filtering. In the case of 2M+1 interval data, the polynomial p is used. i Perform the fit: Among them, the polynomial SGF is used i Fitting; F is the number of local fittings of the polynomial; λ is the fitting order; a λ is the local fitting coefficient of the polynomial; i represents the length of the data; In the scenario with strict power specifications, the upper limits of volatility at 1min and 30min time scales are 2% and 7% respectively, and the 30min volatility evaluation index is considered; in practical applications, the 30min constraint is more stringent, and the percentage of volatility exceeding the limit state is introduced to describe the filtering effect: Where, s is volatility; s max represents the upper limit of the fluctuation rate; m represents the length of the power data at this time.

6. A demand response management method considering new energy forecast smoothing technology according to claim 5, characterized in that: The equivalent transfer function of the Butterworth notch filter is as follows: Among them, ε(s) and x(s) represent the input and output of the Butterworth notch filter respectively; ω0 represents the center frequency of the Butterworth notch filter; μ and τ represent the iteration step size and sampling interval respectively; A represents the amplitude corresponding to the Butterworth notch filter; s represents the complex frequency domain variable of the system.

7. A demand response management method considering new energy forecast smoothing technology according to claim 3, characterized in that: In step S7, in the electricity spot market where new energy sources participate, the introduction of time-of-use electricity prices will inevitably cause changes in users' dynamic demand responses. The changes in demand responses caused by the difference in electricity prices are in a range of fluctuations, which are called optimistic response fluctuations and pessimistic response fluctuations. According to the optimistic response fluctuations and pessimistic response fluctuations, the entire demand response is divided into three areas, namely: static area, dynamic area, and saturation area; the Logistic function can reliably improve the accuracy of the fitting of time-of-use electricity prices to dynamic demand responses; and in the dynamic area, users will be more willing to make a dynamic adjustment of demand responses when participating in the electricity market, so the membership value of the optimistic response is introduced to reflect a probability to describe the situation of this dynamic response.

8. A demand response management method considering new energy forecast smoothing technology according to claim 7, characterized in that: The Logistic function model is as follows: Among them, Δp w,t represents the price difference in the electricity spot market; φ(·) represents the load transfer rate function; α represents the range of the function value; δ represents the horizontal coordinate corresponding to the function value (α / 2+β); μ represents the known quantity in the Logistic function; β is a variable parameter used to move the function curve up and down; when the price difference Δp w,t =0, it means that the time-of-use electricity price strategy is not adopted at this time, so the user's dynamic demand response is not involved.

9. A demand response management method considering new energy forecast smoothing technology according to claim 8, characterized in that: The dynamic response equation introducing the membership value of optimistic response is expressed as follows: in, represents the peak-to-valley transfer rate of renewable energy load; and Respectively represent the peak-valley load transfer ratio of demand response under optimistic and pessimistic conditions; α w,t and β w,t They represent the critical value points of the price difference division area respectively; in the whole dynamic demand response process, in the static area, the price difference is too small at this time, resulting in a large uncertainty in the behavior of individual users in the whole power market, so it is considered that the process is described by the mean of optimistic and pessimistic responses; while in the dynamic area, as the price difference continues to increase, users fully participate in the power market on the basis of the price difference, and users are more optimistic and actively respond to the whole process, which is essentially defined as an optimistic response model; while in the saturated area, the user's response behavior has reached a threshold, so it is described by the maximum value that meets the transfer; The peak-to-flat and flat-to-valley load transfer rates are solved by the dynamic response equation based on the membership value of the optimistic response. Load transfer amount generated under demand response Ω w,t With the following equation: in, and They represent the average load in each period before the implementation of time-of-use electricity price; t p ,t f ,t v They respectively represent the peak period, normal period and valley period of time-of-use electricity price.

10. A computer-readable storage medium having stored thereon computer program instructions that can be executed by a processor, and when the processor executes the computer program instructions, the method steps according to any one of claims 1 to 9 can be implemented.

Images