Power demand side response strategy optimization method based on feedback information extraction

Abstract

The invention discloses a power demand side response strategy optimization method based on feedback information extraction, and relates to the technical field of energy management, and the method comprises the steps: calling a line architecture of a target region, obtaining a user set on a power transmission line, and collecting the power use data of users. And carrying out power and power consumption period disassembly analysis on the data to obtain a power concentration interval, a power consumption peak period and a power consumption valley period of the user. And formulating an initial power demand side response strategy. And collecting feedback information of the user through the feedback window. And in combination with the power interval and the response coefficient of the user, optimizing the initial strategy to form a final power demand side response strategy. The technical problem that an existing power demand side response strategy lacks dynamic adaptability to actual power consumption behaviors and response conditions of users is solved, and the technical effects that the accurate regulation and control capacity of power loads is improved, the influence of power peak and valley fluctuation on a power grid is reduced, and therefore the stability of the power grid is enhanced are achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of energy management, and in particular relates to a method for optimizing a power demand side response strategy based on feedback information extraction. Background Art

[0002] In the operation scenario of modern power systems, the rapid growth and volatility of power loads have brought huge challenges to the stability of the power grid, especially during peak hours, when the contradiction between supply and demand is more prominent. Reasonable allocation of power resources and improving the flexibility and responsiveness of the power grid have become key links in solving the dispatching pressure of the power system. Traditional power demand-side response strategies are often relatively simple and fixed, mainly relying on historical load data and preset rules, with insufficient analysis of real-time user feedback information, lack of dynamic adjustment capabilities, and difficulty in accurately matching user electricity consumption behavior characteristics. The limitations of traditional methods make the power system passive in dealing with complex load fluctuations, and it is difficult to meet the requirements of modern power systems for real-time and accuracy.

[0003] At present, there is a technical problem in the relevant technologies that the power demand side response strategy lacks dynamic adaptability to the users' actual power consumption behavior and response conditions. Summary of the invention

[0004] The purpose of the present invention is to provide a method for optimizing the power demand side response strategy based on feedback information extraction, which uses the line architecture of the target area to obtain the user set on the transmission line and collect the power usage data of the users. The data is analyzed by power and power consumption period to obtain the power concentration interval and peak and valley periods of power consumption of the users. An initial power demand side response strategy is formulated, including incentive intensity and response period, and distributed to each transmission line. The user's feedback information is collected through the feedback window, the response effect is analyzed and the user's response coefficient is calculated. The initial strategy is optimized in combination with the user's power range and response coefficient to form the final power demand side response strategy, which realizes the technical effect of improving the precise regulation capability of the power load, reducing the impact of power peak and valley fluctuations on the power grid, and thus enhancing the stability of the power grid.

[0005] The purpose of the present invention is achieved through the following technical solutions: A method for optimizing a power demand side response strategy based on feedback information extraction, characterized in that the method comprises the following steps: S100, calling the line architecture of the target area, obtaining K user sets of K transmission lines, where K is an integer greater than or equal to 1; S200, traversing the K user sets to collect power usage data in a preset monitoring window to obtain K user power usage data sets; S300, performing power and power consumption period disassembly analysis on the K user power usage data sets to obtain K user power concentration intervals, K peak power consumption concentration periods, K average power consumption concentration periods, and K valley power consumption concentration periods; S400, obtaining an initial power demand side response strategy, wherein the initial power demand side response strategy includes K response strategies, wherein the K response strategies include K incentive intensities and K response time periods; S500, distributing the K excitation intensities and K response time periods to the K transmission lines, and interacting with the K user sets in a preset feedback window to extract feedback information, and obtaining K user feedback parameter sets; S600, performing response analysis on the K user feedback parameter sets to obtain K user response coefficients; S700, optimize the initial power demand side response strategy based on the K user power concentration intervals, K peak power concentration time periods, K average power concentration time periods and K valley power concentration time periods, as well as the K user response coefficients to obtain an optimized power demand side response strategy.

[0006] Further, using power as an index, data extraction is performed on the K user power usage data sets to obtain K user power sets; Performing mean processing on the K user power sets to determine the mean values ​​of the K user powers; Performing center iteration on the K user power means in the K user power sets according to a preset iteration step length to determine the K user power center values; The K user power concentration intervals are determined by taking the K user power center values ​​as the interval center and setting the preset iteration step length as the distance from the two end points to the interval center.

[0007] A central iteration formula is pre-constructed, wherein the central iteration formula is: ; in, is the central value of the user power in the stage, An iterative neighborhood consisting of multiple user powers whose distance to the user power mean in each user power set is a preset iteration step length, is the first User power, is the average user power, is the Gaussian weight kernel function; Taking the K user power averages as matching objects, performing neighborhood searches in the K user power sets according to a preset iteration step length, and determining K iteration neighborhoods; Using the center iteration formula, center iteration is performed on the K user power means and the K iteration neighborhoods to obtain K stage user power center values; After multiple iterations, until the difference between the stage user power center values ​​obtained in two adjacent iterations is less than or equal to the preset difference, the iteration is stopped, and the K stage user power center values ​​obtained in the last iteration are used as the K user power center values.

[0008] Furthermore, an electricity consumption coordinate system is constructed with time as the horizontal axis and electricity consumption as the vertical axis; Obtaining K user power consumption curve sets according to the K user power consumption data sets and the power consumption coordinate system; Fitting the K user power consumption curve sets respectively to obtain K fitted user power consumption curves; Based on the K fitted user power consumption curves, power consumption peak and valley analysis is performed to determine the K peak power consumption concentrated time periods, K average power consumption concentrated time periods and K valley power consumption concentrated time periods.

[0009] Further, the K excitation intensities and K response time periods are distributed to the K transmission lines, and the K user sets are interacted in a preset feedback window to extract feedback information, and K user feedback parameter sets are obtained, including: Obtaining K distribution time nodes of the K excitation intensities and the K response time periods; Extracting K receiving time node sets at which K users in the K transmission lines receive the K excitation intensities and the K response time periods; Performing feedback correction factor analysis based on the K distribution time nodes and the K receiving time node sets to determine K user feedback correction factor sets; The K user feedback parameter sets are modified based on the K user feedback modification factor sets.

[0010] Further, the K user feedback parameter sets include K response rate sets, K response duration sets, and K participation frequency sets.

[0011] Pre-built response factor identifier; The K response rate sets, the K response duration sets and the K participation frequency sets are analyzed by using the response coefficient identifier to determine the K user feedback parameter sets.

[0012] Calculate the ratio of the K user response coefficients to the sum of the K user response coefficients respectively, and use the reciprocal of the ratio as K strategy optimization coefficients; Taking the K strategy optimization coefficients as the optimization scale, and taking the K user power concentration intervals, K peak power concentration time periods, K flat power concentration time periods and K valley power concentration time periods as the optimization analysis data, the initial power demand side response strategy is optimized to obtain the optimized power demand side response strategy.

[0013] The beneficial effects of the present invention are: First, retrieve the line architecture of the target area, obtain the user set on the transmission line, and collect the user's power usage data. Decompose and analyze the data by power and power consumption period to obtain the user's power concentration range and peak and valley periods. Formulate the initial power demand side response strategy, including incentive intensity and response period, and distribute it to each transmission line. Collect user feedback information through the feedback window, analyze the response effect and calculate the user's response coefficient. Combined with the user's power range and response coefficient, optimize the initial strategy to form the final power demand side response strategy, which achieves the technical effect of improving the precise regulation capability of power load, reducing the impact of power peak and valley fluctuations on the power grid, and thus enhancing the stability of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A flow chart of a method for optimizing a power demand side response strategy based on feedback information extraction provided in an embodiment of the present application; Figure 2 A schematic diagram of the feedback correction process of a user feedback parameter set of the power demand side response strategy optimization method based on feedback information extraction provided in an embodiment of the present application. DETAILED DESCRIPTION

[0015] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0016] A power demand side response strategy optimization method based on feedback information extraction, such as Figure 1 As shown, the method includes: Step S100, retrieve the line architecture of the target area, and obtain K user sets of K transmission lines, where K is an integer greater than or equal to 1. Specifically, first establish a connection with the power system database management system, and use the power data management software to screen and extract line architecture information, including physical parameters of the transmission line, identification code, and connection points with the substation, based on conditions such as the geographical coordinate range or administrative area division of the target area, and sort out K transmission lines according to attributes such as voltage level and commissioning time. Then, with the help of the power marketing system customer and line association file records and the smart meter installation and line connection information, the user set is divided by the meter code or user account number as the association key through the data matching algorithm, and special user situations such as dual power supply and temporary power use are processed by combining manual verification with system automatic identification. At the same time, a dynamic update mechanism is established so that when new users are connected or line transformation causes power supply line changes, user set information can be adjusted in time to ensure accuracy and real-time.

[0017] Step S200, traverse the K user sets to collect power usage data of the preset monitoring window, and obtain K user power usage data sets. Specifically, analyze the characteristics of power consumption in the target area, and comprehensively consider factors such as the business peak hours in the commercial area, the seasonal impact of the residential area, the power system operation scheduling cycle, and the electricity price policy period division to determine the preset monitoring window. Install a smart meter or data acquisition terminal device with high-precision electricity metering, data storage and preprocessing capabilities at each user access point, and build a stable communication network of wired (such as power-dedicated optical fiber or urban integrated wiring system) and wireless (such as 4G / 5G or power wireless private network), and optimize the configuration of the communication network priority, bandwidth allocation, etc. According to the preset monitoring window start time, the collection equipment collects data at a set frequency and conducts a preliminary inspection, interpolates and fills in missing data, judges and marks the rationality of mutation data, and sends it to the data center after the monitoring window ends for cleaning and integration again, removes duplications and corrects erroneous formats, and finally forms a K user power usage data set containing comprehensive and accurate power usage data information.

[0018] Step S300, the power and power consumption period of the K user power usage data set are analyzed and disassembled to obtain K user power concentration intervals, K peak power consumption concentration periods, K flat power consumption concentration periods and K valley power consumption concentration periods. Specifically, the K user power usage data set is first cleaned, missing values ​​are processed, abnormal data are corrected and sorted in chronological order. Then the frequency distribution of each user's power data is calculated, the frequency of occurrence of power values ​​is statistically analyzed by grouping at a certain time interval, a power frequency histogram is drawn, and the cluster analysis algorithm is used to aggregate similar power values ​​to determine the main group, the group center value and the degree of dispersion are calculated, and the power concentration interval is determined according to the threshold. Then the sliding average method is used to smooth the power consumption data, the power consumption threshold is set, the average power consumption of each time period is calculated and compared with the threshold, and the time characteristics such as weekdays and weekends, seasonal factors are combined, and the power consumption pattern library is established with reference to historical data and matched and corrected, so as to determine the peak, flat and valley power consumption concentration periods. Finally, key data support is provided for the formulation and optimization of power demand side response strategy.

[0019] In a possible implementation, the K user power usage data sets are subjected to power and power usage period disassembly analysis to obtain K user power concentration intervals, K peak power usage concentration periods, K average power usage concentration periods, and K valley power usage concentration periods. Step S300 further includes: Step S310, using power as an index, extract data from the K user power usage data sets to obtain K user power sets. Specifically, from the comprehensive database of power usage data sets, based on power as the retrieval basis, for each user's power usage data set, traverse each record therein, and filter out the data fields related to power in the record. For example, if the data record format is [timestamp, power consumption, power value, voltage value], extract the power value column data. Perform the same operation on the K user power usage data sets to obtain power data subsets corresponding to the K users, and these subsets are combined to form K user power sets. During the extraction process, it is necessary to ensure the accuracy and completeness of the data, and pre-process some situations where the format is not standardized or the data is missing, such as supplementing the missing power data or eliminating invalid records according to certain rules.

[0020] Step S320, average processing is performed on the K user power sets to determine the K user power averages. Specifically, for each user power set, all power values ​​in the set are accumulated and summed, and then the sum is divided by the number of power values ​​to obtain the average value of the user power set. The calculation process can be automated by writing a data processing program, using array operations and mathematical calculation functions in the programming language. During the calculation process, attention should be paid to the accuracy of the data type to avoid inaccurate calculation results due to data accuracy loss. For example, in some programming languages, if an integer type is used for calculation, the decimal part may be truncated. Therefore, it is necessary to select a suitable data type according to the actual situation, such as a floating point type for accurate calculation, so as to determine K accurate user power averages.

[0021] Step S330, center iteration is performed on the power mean values ​​of the K users in the power sets of the K users according to a preset iteration step length, and the power center values ​​of the K users are determined. Specifically, the value of the preset iteration step length is first determined, which is pre-set according to factors such as data accuracy requirements and data distribution characteristics, and the power mean value of each user is taken as the starting point. In the corresponding power set, the search range is gradually expanded according to the iteration step length. For example, the initial power mean value is P0, and the iteration step length is ΔP. ​​In the first iteration, the power value is searched in the range of P0-ΔP to P0+ΔP, and the weighted average value of the power value in the range is calculated (the weight can be determined according to factors such as the distance from the mean value), as the new center value P1, and then based on P1, the search range is expanded again according to the iteration step length for the next iteration until the preset iteration termination condition is met, such as the difference between the center values ​​obtained in two adjacent iterations is less than a certain minimum value or reaches the preset maximum number of iterations. Through the iterative process, the accurate power center values ​​of the K users are finally determined.

[0022] Step S340, taking the power center values ​​of the K users as the center of the interval, taking the preset iteration step as the distance from the two end points to the center of the interval, and determining the power concentration intervals of the K users. Specifically, for each user, the power center value has been obtained, and taking the center value as a reference, taking the preset iteration step as a distance metric, a range of intervals is determined, for example, if the power center value is Pc and the iteration step is ΔP, then the power concentration interval is [Pc-ΔP, Pc+ΔP]. The interval represents the range of relatively concentrated distribution of the power data of the user. After determining the interval, the rationality of the interval can also be verified, for example, the proportion of the number of power data points falling into the interval to the total number of data points is counted. If the proportion reaches a certain threshold (such as more than 80%), it means that the determined power concentration interval is reasonable and effective, so as to determine an accurate power concentration interval for each user.

[0023] In a possible implementation, performing center iteration on the K user power means in the K user power sets according to a preset iteration step length to determine the K user power center values, step S330 further includes: Step S331, pre-constructing a central iteration formula, wherein the central iteration formula is: ;in, is the central value of the user power in the stage, An iterative neighborhood consisting of multiple user powers whose distance to the user power mean in each user power set is a preset iteration step length, is the power of the i-th user in the iterative neighborhood, is the average user power, is the Gaussian weight kernel function. Specifically, the central iteration formula is constructed based on mathematical principles and data characteristics. The numerator in the formula is Indicates that in the iterative neighborhood Each user power Multiply by the corresponding Gaussian weight kernel function value Post-summation, Gaussian weighted kernel function The role is based on the power value With the mean Different weights are assigned to the distance. The closer the distance to the mean, the greater the weight, and the farther the distance, the smaller the weight. This can highlight the influence of the data points near the center. The weights are summed to normalize the numerator so that the calculated user power center value is It is more reasonable and accurate, and provides a mathematical model basis for subsequent central value calculations.

[0024] Step S332: Taking the K user power averages as matching objects, neighborhood searches are performed in the K user power sets according to the preset iteration step length to determine K iteration neighborhoods. Specifically, for each user power set, taking the determined user power average as a reference point, the power set is searched for power values ​​whose distance from the average is within the preset iteration step length range according to the preset iteration step length. These power values ​​constitute the iteration neighborhood. , for example, if the preset iteration step size is Δ, the mean is , then find the power set that satisfies All ,Will The combination determines an iterative neighborhood. This operation is performed on the K user power means respectively, thereby obtaining K corresponding iterative neighborhoods. In the retrieval process, an efficient data search algorithm should be established to quickly and accurately find the power value that meets the conditions.

[0025] Step S333: Use the center iteration formula to perform center iteration on the K user power mean values ​​and the K iteration neighborhoods to obtain the K-stage user power center values. Specifically, the power mean value of each user and the corresponding iterative neighborhood Substituting into the center iteration formula, for each power value in the iteration neighborhood , first calculate the Gaussian weight kernel function value , and then calculate the values ​​of the numerator and denominator according to the formula, and finally get the stage user power center value By performing such calculations for K users respectively, we can obtain the power center values ​​of the K-stage users. During the calculation process, attention should be paid to the precision control of the data and the accuracy of the function calculation to avoid result deviation due to calculation errors.

[0026] Step S334, after multiple iterations, until the difference between the stage user power center values ​​obtained in two adjacent iterations is less than or equal to the preset difference, the iteration is stopped, and the K stage user power center values ​​obtained in the last iteration are used as the K user power center values. Specifically, when the next iteration starts, the stage user power center value obtained last time is used as the new mean value, and the steps of neighborhood search and center iteration calculation are repeated. After each iteration, the difference between the stage user power center value obtained this time and the value obtained last time is calculated. When the difference is less than or equal to the preset difference (such as 0.01), it means that the iteration has converged and the power center value has stabilized. At this time, the iteration is stopped, and the K stage user power center values ​​obtained in the last iteration are determined as the final K user power center values, which are used for subsequent operations such as determining the power concentration interval.

[0027] In a possible implementation, the K user power usage data sets are subjected to power and power usage period disassembly analysis to obtain K user power concentration intervals, K peak power usage concentration periods, K average power usage concentration periods, and K valley power usage concentration periods. Step S300 further includes: Step S350, constructing an electricity consumption coordinate system with time as the horizontal axis and electricity consumption as the vertical axis. Specifically, clearly mark time as an independent variable on the horizontal axis (horizontal axis), divide the scale according to a certain time interval, such as in hours or determine the scale interval according to the data collection frequency, and for the vertical axis electricity consumption, set a suitable scale value according to the actual electricity consumption range to ensure that the numerical changes of different electricity consumption can be clearly displayed. At the same time, determine the origin position, unit length and other parameters of the coordinate system so that the entire coordinate system has a unified standard and specification, and use the graphics drawing function in the drawing software or data analysis tool to construct this two-dimensional coordinate system to prepare for the subsequent drawing of the electricity consumption curve.

[0028] Step S360, obtaining K user power consumption curve sets according to the K user power consumption data sets and the power consumption coordinate system. Specifically, for each user's power consumption data set, extract the time data and the corresponding power consumption data from the set in sequence, correspond the extracted time data to the horizontal axis, and correspond the power consumption data to the vertical axis. In the power consumption coordinate system, connect each data point in chronological order, so as to draw the power consumption curve of each user. Since there may be some small fluctuations and errors in the data collection process, the drawn curve may have some jagged or rough parts. By performing such operations on K users respectively, a set of K user power consumption curves can be obtained, which intuitively shows the changes in power consumption of each user at different times.

[0029] Step S370, respectively fit the K user power consumption curve sets to obtain K fitted user power consumption curves. Specifically, select a suitable fitting algorithm, such as polynomial fitting, exponential fitting or a curve fitting algorithm based on machine learning, etc. For each user's power consumption curve, input the curve data into the fitting algorithm, and the fitting algorithm determines the parameters of the fitting curve by minimizing the error between the actual data point and the fitting curve. For example, in polynomial fitting, select a suitable polynomial degree according to the data characteristics, and determine the coefficient of the polynomial by solving the equation group. After fitting, a smoother fitting user power consumption curve that can reflect the power consumption trend is obtained. The power consumption curves of K users are respectively fitted to obtain K fitted user power consumption curves, which provide a more accurate data basis for the subsequent power consumption peak and valley analysis.

[0030] Step S380, based on the K fitting user power consumption curves, power consumption peak and valley analysis is performed to determine the K peak power consumption concentrated time periods, K flat power consumption concentrated time periods and K valley power consumption concentrated time periods. Specifically, the threshold standard for peak and valley judgment is set, which can generally be determined according to the relative size of power consumption or the difference with the average power consumption, etc., along the fitting user power consumption curve, the power consumption value corresponding to each time point is analyzed from left to right in turn, when the power consumption is higher than the peak power consumption threshold and lasts for a period of time (such as several consecutive time points), this period is determined as the peak power consumption concentrated time period; when the power consumption is lower than the valley power consumption threshold and meets a certain duration condition, it is determined as the valley power consumption concentrated time period; between the peak and valley thresholds is the flat power consumption concentrated time period, and at the same time, a comprehensive judgment is made in combination with the user's power consumption habits, seasonal factors, and the difference between working days and rest days. For example, on summer working days, the office space has high power consumption during the day, which may be a peak period; the residential power consumption is reduced at night, which is a valley period. Such analysis is performed on the K fitting user power consumption curves respectively, and finally the peak, flat, and valley power consumption concentrated time periods corresponding to each user are determined.

[0031] Step S400, obtaining the initial power demand side response strategy, wherein the initial power demand side response strategy includes K response strategies, wherein the K response strategies include K incentive intensities and K response time periods. Specifically, collect the past demand side response activity records of the power system and evaluate the current power market supply and demand situation, analyze the power consumption behavior characteristics of users in the target area, and use big data analysis to mine the power consumption data rules. According to the difference in user sensitivity to price and reward, formulate different incentive intensities, consider the cost and expected effect of the power company, determine the scope through simulation calculation and cost-benefit analysis, set electricity price discounts for price-sensitive users, and set reward plans for users who focus on non-monetary incentives. Set the response time period in combination with the peak and valley change law of power grid load and the distribution characteristics of power generation resources, analyze the historical load curve to find out the peak and valley time characteristics, set the response time period according to the power generation resource situation, such as wind power is high at night and power consumption is low, and consider the user's life and production rules, communicate and negotiate with relevant parties to solicit opinions and adjustments, and finally determine the incentive intensity and response time period corresponding to each response strategy to form the initial power demand side response strategy.

[0032] Step S500, distribute the K incentive intensities and K response time periods to the K transmission lines, and interact with the K user sets in the preset feedback window to extract feedback information, and obtain K user feedback parameter sets. Specifically, the K incentive intensities and response time period data are synchronized from the strategy formulation system to the relevant business system through various methods such as publishing notices and setting up query pages on the official website of the power company, using the marketing system to send mobile phone text messages with links, arranging customer managers to provide door-to-door services to large and important users and issuing paper materials, and setting pop-up prompts on smart meters. Encryption technology is used to ensure transmission security and verification is performed before display or notification to ensure that the information is accurately conveyed to the K user sets. An interactive platform combining online (developing mobile applications or mini-programs with feedback portals and equipping them with online customer service) and offline (setting up a consulting service desk in the power supply business hall) is built in front of the preset feedback window. Smart meters are used to report changes in electricity consumption data. During the feedback window, feedback from various channels is continuously collected and sorted, and key parameters such as user satisfaction with incentive intensity, feedback on the rationality of the response period, and electricity consumption change data are extracted. Fuzzy or incomplete information is improved to form a set of K user feedback parameters, providing a reliable data basis for response analysis.

[0033] In one possible implementation, Figure 2 As shown, the K excitation intensities and K response periods are distributed to the K transmission lines, and the K user sets are interacted in a preset feedback window to extract feedback information to obtain K user feedback parameter sets. Step S500 further includes: Step S510, obtaining K distribution time nodes of the K incentive intensities and the K response time periods. Specifically, a monitoring program is set in the policy distribution system, and when the incentive intensity and response time period information are distributed to the users corresponding to each transmission line, the precise time point is automatically recorded. The time recording process can use the system's log function or a specially developed timestamp recording module. For different distribution channels, such as SMS sending, website announcement publishing, meter push, etc., the respective distribution start time is recorded, and the data is sorted and classified. According to the number of the transmission line or the identifier of the user set, the corresponding distribution time nodes are accurately associated to form a clear record of K distribution time nodes, so as to facilitate the subsequent time difference analysis.

[0034] Step S520, extracting K receiving time node sets at which K users in the K transmission lines receive the K incentive intensities and the K response time periods. Specifically, the receiving time information is obtained by setting a feedback monitoring mechanism on the user terminal device (such as a mobile phone, a computer, a smart meter, etc.). When the user opens a text message, browses a website announcement, checks the meter prompt information, etc., the device automatically transmits the receiving time data back to the system. For some situations where the network is unstable or the device function is limited, multiple attempts to return and data caching mechanisms are used to ensure that the receiving time data can be accurately obtained. After the receiving time data of each user is collected, it is classified and sorted according to the corresponding transmission line and user set to form K receiving time node sets. The set includes the specific time point when each user receives the information, providing basic data for subsequent analysis.

[0035] Step S530, based on the K distribution time nodes and the K receiving time node sets, feedback correction factor analysis is performed to determine K user feedback correction factor sets. Specifically, for each group of corresponding distribution time nodes and receiving time nodes, the time difference is calculated, and the calculation of the time difference can be accurate to the second or even millisecond level to accurately reflect the delay of information transmission. According to the pre-set algorithm rules, the time difference is converted into a feedback correction factor. Generally speaking, if the time difference is long, it means that there may be obstacles in the information transmission process or the user does not receive the information in time, and the corresponding feedback correction factor will be large. For example, a simple linear relationship function can be set. When the time difference is within a certain range, the feedback correction factor is calculated in proportion. The time difference of K users is calculated and converted respectively, so as to determine the K user feedback correction factor set, which reflects the delay and potential influencing factors of each user in the process of receiving information.

[0036] Step S540, based on the K user feedback correction factor sets, the K user feedback parameter sets are corrected. Specifically, the feedback correction factor set is associated and matched with the feedback parameter set, and each parameter in the feedback parameter set is adjusted according to the corresponding feedback correction factor. If the feedback correction factor of a certain user is large, it means that there are certain difficulties or delays in receiving information. In this case, a certain degree of tolerance needs to be given when analyzing its feedback parameters. For example, for the response willingness parameter of user feedback, if it is originally at a low level, but considering that it receives information late and may not fully understand the content of the strategy, the weight of the parameter can be appropriately increased or its value range can be adjusted. In this way, the K user feedback parameter sets are comprehensively corrected, so that the subsequent strategy adjustment and analysis based on the feedback parameters are more scientific and reasonable, and can fully consider the actual situation of the user in the information receiving link.

[0037] In a possible implementation, the K excitation intensities and K response time periods are distributed to the K transmission lines, and the K user sets are interacted in a preset feedback window to extract feedback information, and K user feedback parameter sets are obtained. Step S500 further includes step S550, and the K user feedback parameter sets include K response rate sets, K response duration sets, and K participation frequency sets. Specifically, during the preset feedback window, the power consumption change data is collected at short time intervals and smoothed by the smart meter and the monitoring system. The rate of change of power consumption at each time point relative to the previous moment is calculated from the moment of receiving the excitation information. After analyzing and removing unstable data, it is comprehensively characterized by statistics, and representative values ​​are determined in combination with actual scenarios to form K response rate sets. The start and end time points of the user response are recorded, and the effective response duration period is screened by comparing different data. The data set of each user is summarized and sorted, and the typical range or representative value is determined by cluster analysis to generate K response duration sets. The number of times each user participates in demand-side response within a certain period of time is counted from multiple channels. Manual review and data analysis are combined to confirm the data, the number of participations per unit time is calculated and the rationality is verified, and K participation frequency sets are compiled to provide data support for subsequent strategy optimization and other work.

[0038] In a possible implementation, the K incentive intensities and K response periods are distributed to the K transmission lines, and the K user sets are interacted in a preset feedback window to extract feedback information, and K user feedback parameter sets are obtained. Step S500 further includes step S560, pre-building a response coefficient identifier. Specifically, the algorithm core of the response coefficient identifier is determined based on statistical principles and professional knowledge of power demand side response, and the relationship between different response rates, durations and participation frequencies and their influence weights on the overall response effect are studied. For example, through the analysis of a large amount of historical data, it is found that when the response rate is fast and the duration is long, the contribution to the power demand side response is large, but when the participation frequency is unstable, it will also affect the overall effect. Taking these factors into consideration, the calculation method and the correlation formula of each parameter in the algorithm are determined, and the classification and regression algorithms in the machine learning algorithm are combined to improve the accuracy and adaptability of the recognition. The classification algorithm can classify different user response situations, such as excellent response, general response, poor response and other categories; the regression algorithm is used to predict the response effect value under different response parameter combinations, collect user feedback data from past power demand side response activities, including response rate, duration, participation frequency and corresponding actual response effect data, divide the data into training set and test set, and use the training set data to train the response coefficient identifier model. During the training process, the parameters and model structure in the algorithm are continuously adjusted to minimize the prediction error. For example, the parameters of the model are optimized through the gradient descent algorithm to improve the fitting ability of the model. The trained model is verified and evaluated using the test set data, and the model is further optimized according to the evaluation results, such as adjusting the classification threshold, increasing or decreasing the complexity of the model, etc., until the model reaches the predetermined accuracy and stability indicators. After multiple iterative optimizations, a response coefficient identifier that can accurately identify the user response coefficient is finally constructed.

[0039] Step S570: Utilize the response coefficient identifier to analyze the K response rate sets, the K response duration sets, and the K participation frequency sets to determine the K user feedback parameter sets. Specifically, the data in the K response rate sets, K response duration sets and K participation frequency sets are sorted and formatted to meet the input requirements of the response coefficient identifier. For example, the data is standardized and the data of different dimensions are converted into a unified standard scale so that the model can accurately process it. The missing values ​​are filled and the missing data are supplemented by appropriate methods such as mean filling and interpolation to ensure the integrity of the data. The processed data are input into the response coefficient identifier in sequence. The response rate, duration and participation frequency data corresponding to each user are used as a set of input data. The response coefficient identifier analyzes and calculates the input data according to the pre-built algorithm model. For each set of input data, the category to which the user response belongs is determined through the classification part in the model; the specific response coefficient value is calculated through the regression part, and the classification result and response coefficient value of each user are combined to form K user feedback parameter sets. The set not only contains the classification information of the user response situation, but also accurately quantifies the response degree of each user, which provides a key basis for subsequent optimization of the power demand side response strategy and other work. By continuously accumulating new data and updating and optimizing the identifier, it can more accurately analyze and process user feedback parameters.

[0040] Step S600, perform response analysis on the K user feedback parameter sets to obtain K user response coefficients. Specifically, first perform Z-score standardization on the response rate, response duration and participation frequency data in the K user feedback parameter sets to make the data comparable. Then, a response analysis model is constructed according to the power demand side response principle and objectives, and the weights of each parameter are determined by combining historical data regression analysis and expert experience. A weighted summation method is used to establish a mathematical relationship with response rate, duration and participation frequency as input variables. Substitute the standardized data into the model, calculate each user response coefficient according to the weight, and pay attention to data accuracy control during the calculation. Finally, the response coefficient is evaluated and verified by comparing with the actual power system operation data and using internal verification methods such as cross-validation. If there is a deviation, re-check and adjust the data processing, model construction and weight distribution links until reliable K user response coefficients are obtained.

[0041] Step S700, combining the K user power concentration intervals, K peak power concentration periods, K flat power concentration periods and K valley power concentration periods, and the K user response coefficients to optimize the initial power demand side response strategy, and obtain an optimized power demand side response strategy. Specifically, first summarize and integrate the user power concentration intervals, peak, flat and valley power consumption periods and response coefficient data, classify and organize the power interval data and statistically analyze them, and draw a power consumption period trend chart to compare the power consumption differences of different users. Adjust the incentive intensity according to the user's power consumption behavior, increase the incentive for users with high peak power consumption and concentrated power to adjust their power consumption time, and optimize the cost adjustment incentive for users with low power and high power consumption in valleys; optimize the response period according to the response coefficient, extend the period for users with high response and long flat power concentration period, evaluate the rationality of the period for users with low response and adjust or guide according to feedback. Use simulation software to establish a model input adjusted parameters to simulate the system operation, compare the results with the optimization goals, and re-analyze and adjust and simulate multiple times if it does not meet expectations, and finally determine the strategy parameters to integrate into the optimization strategy and establish a dynamic adjustment mechanism to adapt to changes.

[0042] In a possible implementation, the initial power demand side response strategy is optimized in combination with the K user power concentration intervals, the K peak power concentration time periods, the K average power concentration time periods, the K valley power concentration time periods, and the K user response coefficients to obtain an optimized power demand side response strategy, and step S700 further includes: Step S710, respectively calculate the ratio of the K user response coefficients to the sum of the K user response coefficients, and use the reciprocal of the ratio as the K strategy optimization coefficients. Specifically, for each user's response coefficient, calculate its ratio to the sum of the K user response coefficients. This step requires first accurately calculating the sum of the K user response coefficients, and then dividing each user's response coefficient by the sum one by one. In the calculation process, attention should be paid to the control of data accuracy, and appropriate data types and calculation methods should be used to ensure the accuracy of the calculation results. For example, a high-precision numerical calculation library or software tool is used for calculation to avoid ratio calculation errors due to data accuracy problems. After the ratio is calculated, its reciprocal is obtained as the strategy optimization coefficient. Since the better the user feedback (the larger the response coefficient), the larger the ratio, the smaller the reciprocal (strategy optimization coefficient). This means that for users with positive responses and good feedback, the adjustment range in the subsequent strategy optimization is relatively small; while for users with small response coefficients and poor feedback, the strategy optimization coefficient is large, and a large adjustment is required during the optimization process. Through this calculation method, a coefficient is determined for each user that can reflect his or her adjustment weight in strategy optimization.

[0043] Step S720, taking the K strategy optimization coefficients as the optimization scale, taking the K user power concentration intervals, K peak power concentration time periods, K flat power concentration time periods and K valley power concentration time periods as the optimization analysis data, optimizing the initial power demand side response strategy, and obtaining the optimized power demand side response strategy. Specifically, the K strategy optimization coefficients are associated with the optimization analysis data such as the user power concentration interval, peak, flat and valley power concentration time periods. According to the number or identification of each user, ensure that the strategy optimization coefficient accurately matches the corresponding power consumption data. Further preprocessing is performed on the optimization analysis data, such as refining and classifying the power concentration interval data, dividing different level intervals according to the power size; performing statistics and analysis on the peak, flat and valley power concentration time period data over a time span, clarifying the typical power consumption characteristics of each time period, and adjusting the incentive intensity according to the strategy optimization coefficient. For users with small strategy optimization coefficients, the adjustment range of the incentive intensity is appropriately reduced, and fine-tuning is performed on the basis of maintaining their original power consumption habits and response modes. For users with large coefficients, the adjustment of incentive intensity is strengthened, and users are attracted to change their electricity consumption behavior by increasing the preferential range of electricity prices, increasing the amount of subsidies, or providing more value-added services. For example, for users with concentrated power during peak hours and large strategy optimization coefficients, higher peak-valley electricity price difference incentives can be provided to encourage them to shift their electricity consumption time to valley hours or normal hours, and the response time period is optimized by combining the strategy optimization coefficient and the user's electricity consumption time period data. For users with small strategy optimization coefficients and regular and stable electricity consumption periods, their original response time period settings are maintained or small-scale optimization adjustments are made. For users with large coefficients and irregular electricity consumption periods or poor response effects, their electricity consumption behavior characteristics are re-evaluated, and a more reasonable response time period is formulated for them in combination with the load demand and power generation resources of the power grid. For example, according to the peak-valley distribution of power generation resources in the user's area and the power grid load forecast curve, the response time period is adjusted to a time period that is more conducive to balancing the supply and demand relationship of the power grid, thereby optimizing the initial power demand side response strategy and obtaining an optimized power demand side response strategy that is more in line with the actual situation of the user and the power grid operation needs.

[0044] The present invention uses the line architecture of the target area to obtain the user set on the transmission line and collect the user's power usage data. The data is analyzed by disassembly of power and power consumption period to obtain the user's power concentration range and peak and valley periods. An initial power demand side response strategy is formulated, including incentive intensity and response period, and distributed to each transmission line. User feedback information is collected through the feedback window, the response effect is analyzed and the user's response coefficient is calculated. Combined with the user's power range and response coefficient, the initial strategy is optimized to form the final power demand side response strategy, which achieves the technical effect of improving the ability to accurately control the power load and reducing the impact of power peak and valley fluctuations on the power grid, thereby enhancing the stability of the power grid.

[0045] The above specific implementation manner does not constitute a limitation to the protection scope of the present application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application. In some cases, the actions or steps recorded in the present application can be performed in an order different from that in the embodiment and can still achieve the desired results. In addition, the process depicted in the accompanying drawings does not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A method for optimizing power demand response strategy based on feedback information extraction, characterized in that: The method comprises the following steps: S100, calling the line architecture of the target area, obtaining K user sets of K transmission lines, where K is an integer greater than or equal to 1; S200, traversing the K user sets to collect power usage data in a preset monitoring window to obtain K user power usage data sets; S300, performing power and power consumption period disassembly analysis on the K user power usage data sets to obtain K user power concentration intervals, K peak power consumption concentration periods, K average power consumption concentration periods, and K valley power consumption concentration periods; S400, obtaining an initial power demand side response strategy, wherein the initial power demand side response strategy includes K response strategies, wherein the K response strategies include K incentive intensities and K response time periods; S500, distributing the K excitation intensities and K response time periods to the K transmission lines, and interacting with the K user sets in a preset feedback window to extract feedback information, and obtaining K user feedback parameter sets; S600, performing response analysis on the K user feedback parameter sets to obtain K user response coefficients; S700, optimize the initial power demand side response strategy based on the K user power concentration intervals, K peak power concentration time periods, K average power concentration time periods and K valley power concentration time periods, as well as the K user response coefficients to obtain an optimized power demand side response strategy.

2. The power demand side response strategy optimization method based on feedback information extraction according to claim 1 is characterized in that: include: Taking power as an index, extracting data from the K user power usage data sets to obtain K user power sets; Performing mean processing on the K user power sets to determine the mean values ​​of the K user powers; Performing center iteration on the K user power means in the K user power sets according to a preset iteration step length to determine the K user power center values; The K user power concentration intervals are determined by taking the K user power center values ​​as the interval center and setting the preset iteration step length as the distance from the two end points to the interval center.

3. The power demand side response strategy optimization method based on feedback information extraction according to claim 2 is characterized in that: include: A central iteration formula is pre-constructed, wherein the central iteration formula is: ； in, is the central value of the user power in the stage, An iterative neighborhood consisting of multiple user powers whose distance to the user power mean in each user power set is a preset iteration step length, is the first User power, is the average user power, is the Gaussian weight kernel function; Taking the K user power averages as matching objects, performing neighborhood searches in the K user power sets according to a preset iteration step length, and determining K iteration neighborhoods; Using the center iteration formula, center iteration is performed on the K user power means and the K iteration neighborhoods to obtain K stage user power center values; After multiple iterations, until the difference between the stage user power center values ​​obtained in two adjacent iterations is less than or equal to the preset difference, the iteration is stopped, and the K stage user power center values ​​obtained in the last iteration are used as the K user power center values.

4. The power demand side response strategy optimization method based on feedback information extraction according to claim 1 is characterized in that: include: Construct an electricity consumption coordinate system with time as the horizontal axis and electricity consumption as the vertical axis; Obtaining K user power consumption curve sets according to the K user power consumption data sets and the power consumption coordinate system; Fitting the K user power consumption curve sets respectively to obtain K fitted user power consumption curves; Based on the K fitted user power consumption curves, power consumption peak and valley analysis is performed to determine the K peak power consumption concentrated time periods, K average power consumption concentrated time periods and K valley power consumption concentrated time periods.

5. The power demand side response strategy optimization method based on feedback information extraction according to claim 1 is characterized in that: The K excitation intensities and K response periods are distributed to the K transmission lines, and the K user sets are interacted in a preset feedback window to extract feedback information, and K user feedback parameter sets are obtained, including: Obtaining K distribution time nodes of the K excitation intensities and the K response time periods; Extracting K receiving time node sets at which K users in the K transmission lines receive the K excitation intensities and the K response time periods; Performing feedback correction factor analysis based on the K distribution time nodes and the K receiving time node sets to determine K user feedback correction factor sets; The K user feedback parameter sets are modified based on the K user feedback modification factor sets.

6. The power demand side response strategy optimization method based on feedback information extraction according to claim 1 is characterized in that: The K user feedback parameter sets include K response rate sets, K response duration sets, and K participation frequency sets.

7. The power demand side response strategy optimization method based on feedback information extraction according to claim 6 is characterized in that: include: Pre-built response factor identifier; The K response rate sets, the K response duration sets and the K participation frequency sets are analyzed by using the response coefficient identifier to determine the K user feedback parameter sets.

8. The power demand side response strategy optimization method based on feedback information extraction according to claim 1 is characterized in that: include: Calculate the ratio of the K user response coefficients to the sum of the K user response coefficients respectively, and use the reciprocal of the ratio as K strategy optimization coefficients; Taking the K strategy optimization coefficients as the optimization scale, and taking the K user power concentration intervals, K peak power concentration time periods, K flat power concentration time periods and K valley power concentration time periods as the optimization analysis data, the initial power demand side response strategy is optimized to obtain the optimized power demand side response strategy.

Images