Real-time demand response user screening method based on multi-dimensional evaluation indexes

Abstract

The invention discloses a real-time demand response user screening method based on a multi-dimensional evaluation index, and the method systematically integrates a plurality of links, such as data preprocessing, power grid and user dual-view analysis, multi-dimensional evaluation index system construction and weight optimization, and user response capability quantification and sorting. Quantitative evaluation and linear normalization processing are carried out on the multi-dimensional response capability of each demand response participating user, so that effective screening and identification of high-quality demand response users are realized. The method has the advantages that by introducing multi-dimensional evaluation indexes including the effective response load, the satisfaction rate and the response capacity ratio of the power grid level and the power utilization mode, the electricity fee expenditure satisfaction degree and the like of the user level, comprehensive evaluation of the response ability and willingness of the user is achieved; according to the method, high-quality users with good response potential and actual participation intentions can be identified more accurately, and the screening accuracy and effectiveness are improved.

Description

Technical Field

[0001] The present invention relates to a real-time demand response user screening method, in particular to a real-time demand response user screening method based on multi-dimensional evaluation indicators, and belongs to the technical field of smart grids. Background Art

[0002] With the rapid development of smart grid technology and the continuous optimization of energy structure, the supply and demand balance of power system is facing unprecedented challenges. In order to effectively cope with the peak of power demand, optimize the allocation of power resources and improve energy efficiency, demand response (DR) as an important means of flexibly adjusting power supply and demand is gradually becoming a key link in the operation of power system.

[0003] The demand response mechanism allows power users to adjust their power consumption behavior according to their actual situation after receiving grid dispatch instructions, such as reducing load and shifting peak power consumption periods, so as to help the grid maintain stable operation and alleviate the tight power supply situation. However, in actual operation, how to efficiently identify and screen high-quality users with good response capabilities and who can actively participate in demand response plans has become a major problem in the current power system operation and management.

[0004] In the prior art, there is a method for automatically screening demand response users disclosed in announcement number CN105096040A, which first selects a part of the remaining responding users as candidate users for this round of screening, and then establishes an integer linear programming model based on the power load gap and the load reduction capacity of the candidate users, and uses the branch and bound method to solve it, interact with the user, and determine whether the next round of screening is needed based on the feedback, and repeat the cycle until the power load gap issued by the superior system is completed. Most of the existing user screening methods rely on simple load data analysis and user historical behavior records, which lack comprehensiveness and accuracy. These methods often ignore the differences in users' subjective willingness, electricity usage habits, electricity cost sensitivity, etc. during the response process, resulting in unsatisfactory screening results and difficulty in meeting the actual needs of power grid dispatching. Based on this, the present application proposes a real-time demand response user screening method based on multi-dimensional evaluation indicators. Summary of the invention

[0005] The purpose of the present invention is to provide a real-time demand response user screening method based on multi-dimensional evaluation indicators in order to solve at least one of the above technical problems, so as to improve the limitations and shortcomings of the current demand response user screening method.

[0006] The present invention achieves the above-mentioned purpose through the following technical scheme: a real-time demand response user screening method based on multi-dimensional evaluation indicators, comprising the following steps:

[0007] S1. Collect user historical operation data and basic information of demand response events, and pre-process user historical operation data;

[0008] S2. From the perspective of the power grid, the effective response load and satisfaction rate of each user in different response periods are determined based on the actual load and baseline load of the user's participation in the last demand response, and the response capacity ratio is further calculated;

[0009] S3. From the user level, the difference between the actual load and the original load after the user participated in the last demand response, as well as the change in electricity bill expenditure before and after the user participated in the last demand response, is used to evaluate the user's electricity consumption mode and electricity bill expenditure satisfaction;

[0010] S4. Based on the four aspects of user response evaluation indicators, a multi-dimensional user response evaluation indicator system is established, and the weights of the indicator system are calculated and corrected using the analytic hierarchy process and entropy weight method;

[0011] S5. Calculate the multi-dimensional response evaluation index of each demand response participating user, perform linear normalization on the multi-dimensional response evaluation index, screen each user as a high-quality user, and obtain high-quality demand response users.

[0012] As a further solution of the present invention: collecting user historical operation data and demand response event basic information, and preprocessing the user historical operation data, specifically including the following steps:

[0013] S11. Collect historical user operation data and basic information of demand response events. The collection cycle is 15 minutes per interval. The user's historical operation data includes the historical operation load at time t on the Nth day of the past week. The basic information of the demand response event includes the last response invitation quantity P invite , the number of the last demand response period N dr , the time period set T of the last demand response event dr 、Time-of-use electricity price unit price λ t out , response subsidy unit price λ subsidy ;

[0014] S12. Use the Median Absolute Deviation (MAD) method to identify missing values ​​and outliers in the user's historical operation data. The specific steps are as follows:

[0015] ① Define the user's historical operating load at time t on day N of the past week Calculate the median of the historical operating load at all times t in the past week (P t history ) where N = {1,...,7}, t∈T history And T history is a collection of historical operating load periods;

[0016] ② Calculate the median of the absolute deviation values ​​at all times t in the past week

[0017] ③ By setting a reasonable proportional factor constant k, usually k = 1.4826, MAD t Approximately equivalent to the standard deviation), so the reasonable range of the historical operating load fluctuation is determined as:

[0018]

[0019] S13, use Lagrange interpolation (LI) to fill in the missing values ​​in the user's historical operation data and correct the abnormal values. For the data points that need to be filled, select the point t N The normal data points at the first m moments and the last m moments are used as samples to calculate the Lagrange polynomial:

[0020]

[0021] Where: t N,i and t N,j are the data sampling points at the i-th and j-th moments of the N-th day in the past week; P N,i is the historical operating load at the i-th moment of the Nth day in the past week; finally, the missing moment t N Substitute P N,t ’ can obtain the filling value of missing and abnormal moments.

[0022] As a further solution of the present invention: the calculation of the response capacity ratio specifically includes the following steps:

[0023] S21. According to the actual load of the user participating in the last demand response and the baseline load, the user's effective load is identified and the user's demand response period ratio is calculated. The formula is:

[0024]

[0025] Where: η time are the effective load identification variable and the effective response period proportion of the user’s last demand response, respectively; P t actual , are the actual load and baseline maximum load of the user participating in the last demand response at time t, respectively; Ndr is the number of the last demand response period; T dr is the time period set of the last demand response event;

[0026] S22. Calculate the difference between the actual average load of the user participating in the last demand response and the baseline average load based on the effective load of the user participating in the last demand response. The formula is:

[0027]

[0028] Where: ΔP poor P is the difference between the actual average load of the user participating in the last demand response and the baseline average load; base line,t is the baseline load of the user participating in the last demand response at time t;

[0029] According to the difference between the actual average load of the user participating in the last demand response and the average load baseline, the response capacity ratio of the user participating in the last demand response is calculated. The formula is:

[0030]

[0031] Where: η capacity P is the response capacity ratio of users participating in the last demand response; invite The number of invitations to respond to the last demand event.

[0032] As a further solution of the present invention: the evaluation of the user's electricity usage mode and electricity fee expenditure satisfaction specifically includes the following steps:

[0033] S31. The user's electricity consumption satisfaction can be expressed by the difference between the actual load of the user participating in the last demand response and the original electricity load. The formula is:

[0034]

[0035] Where: η way P is the user's satisfaction with electricity consumption in the last demand response; t primitive is the original power load at time t when the user participated in the last demand response; M is the total number of days in the week before the user participated in the last demand response, that is, M = max(N);

[0036] S32. The user's electricity expenditure satisfaction can be expressed by the change between the electricity expenditure after the user participated in the last demand response and the electricity expenditure when the user did not participate in the demand response. The formula is:

[0037]

[0038] Where: η disburse The user's satisfaction with electricity expenditure in the last demand response; t out , subsidy They are time-of-use electricity price unit price and response subsidy unit price respectively.

[0039] As a further solution of the present invention: the calculation and correction of the weight of the index system by the hierarchical analysis method specifically includes:

[0040] AHP can decompose complex problems into several combined factors, group these combined factors according to the dominant relationship, form a top-down hierarchical structure, and obtain the relative importance of factors at the same level by comparing them two by two.

[0041] The main steps of using AHP are:

[0042] ① Determine the evaluation objectives of the evaluation plan, collect the standard criteria based on which to base the evaluation, and determine the evaluation index system;

[0043] ② Establish an evaluation model, which generally consists of a target layer, a criterion layer, and a solution layer, and determine the inclusion factors of each layer;

[0044] ③ Establish a judgment matrix corresponding to the factors at each level. In the evaluation model established in the previous step, the factors contained in each layer are compared with each other using a factor in the upper layer as a comparison criterion; in the established judgment matrix, A=(a ij ) n×n a ij Indicates the importance of the i-th factor in this level relative to the j-th factor;

[0045] ④ Solve the maximum eigenvalue of the judgment matrix A and the eigenvector w corresponding to A, and then obtain the weight of the hierarchical factor after normalization;

[0046] ⑤Perform consistency check, for each maximum eigenvalue λ of A max And the eigenvector w, the eigenvector can be regarded as the relative importance scale of the factors at the same level when the factors at the upper level are used as the comparison criteria; because the comparison scale may be inconsistent when the factors at the same level are relatively compared, there may be errors when sorting each level; therefore, consistency check is required, and the steps are as follows:

[0047] First, calculate the consistency index according to the following formula

[0048]

[0049] Secondly, according to n, check the average random consistency index R I; Finally, calculate the consistency ratio of the judgment matrix, that is, use the following formula to calculate:

[0050]

[0051] If the obtained C R <0.1, then it can be determined that the current judgment matrix meets the consistency requirements; otherwise, the judgment matrix needs to be modified again;

[0052] ⑥ Comprehensively summarize the weight results of each level to get the final conclusion.

[0053] As a further solution of the present invention: the calculation and correction of the weight of the index system by the entropy weight method specifically includes:

[0054] Since the AHP only considers the subjective judgment of experts when comparing the elements of each layer, when there are many fuzzy factors in the problem, it may lead to low accuracy and a large difference from the actual situation. Therefore, on the basis of the AHP, the application of the entropy weight method is combined, the AHP and the entropy weight method are combined, and the information of the entropy value is used to correct the subjective judgment results obtained by the AHP method, so as to form a more objective and accurate evaluation result.

[0055] The entropy weight method can calculate the objective weighting results. According to the irregularity of each factor, the weight of each factor can be calculated, and then the weight of each indicator can be corrected according to the entropy weight to obtain a more objective weight result.

[0056] Information represents the order of the system; entropy represents the disorder of the system; if the probability of a system being in a certain situation is p i (i=1,2,...,m), define the system entropy

[0057]

[0058] When p i =1 / m(i=1,2,...,m), the probability of being in each situation is equal, and the entropy takes the corresponding maximum result:

[0059] e=l nm

[0060] e max =1nm

[0061] For the corresponding n evaluation indicators in the m schemes to be evaluated, the evaluation matrix R = (r ij )m×n, for a certain indicator r j There is information entropy:

[0062]

[0063] If the entropy value e j The smaller it is, the more obvious its irregularity is, so it can be considered that this indicator can provide more information and its given weight is greater;

[0064] The entropy weight method calculation process, the specific steps are as follows:

[0065] Assume that there are m projects to be evaluated and n evaluation indicators, forming the original data matrix R = (r ij )m×n:

[0066]

[0067] In the formula, r ij is the evaluation value of i to j. On this basis, the indicator weight is calculated;

[0068] ①The weight of i to j's index p ij :

[0069]

[0070] ② Calculate the entropy value e of the jth indicator j :

[0071]

[0072] ③ Calculate the entropy weight w of the jth indicator j :

[0073]

[0074] ④Determine the comprehensive weight of the indicator β j :

[0075] Using the weight of AHP as α j , where j = 1, 2, ..., n, this weight is combined with the weight of the entropy weight method. The method of combining the subjective weight obtained by the AHP method with the entropy weight is the entropy weight correction method:

[0076]

[0077] As a further solution of the present invention: the calculation of the multi-dimensional response evaluation index of each demand response participating user specifically includes the following steps:

[0078] S51. Calculation of multi-dimensional user response evaluation indicators:

[0079] U synthesis =β 1 η time +β 2 η capacity +β 3 ηway +β 4 η disburse

[0080] Among them, U synthesis It is a multi-dimensional user response evaluation indicator; β 1 , β 2 , β 3 , β 4 They are the weight coefficients of response time proportion, response capacity proportion, electricity consumption mode satisfaction, and electricity expense satisfaction;

[0081] S52. Normalize the multi-dimensional user response evaluation index of the user using a linear ratio normalization method:

[0082]

[0083] Where: is the normalized multi-dimensional user response evaluation index of the i-th user;

[0084] are the maximum and minimum values ​​of the multi-dimensional user response evaluation indicators for all users, respectively.

[0085] S53. Response to user evaluation score:

[0086] According to the normalized multi-dimensional user response evaluation index, the response user evaluation score is calculated. The formula is:

[0087]

[0088] Where: U score In response to user evaluation scores;

[0089] S54. Screening high-quality demand response users

[0090] U high =f(U score )≥0.8

[0091] Where: U high Respond to users' needs for quality.

[0092] The beneficial effects of the present invention are:

[0093] 1. The present invention achieves a comprehensive evaluation of user response capabilities and willingness by introducing multi-dimensional evaluation indicators, including effective response load, satisfaction rate, response capacity ratio at the grid level, and electricity consumption mode and electricity fee expenditure satisfaction at the user level. Compared with the traditional single-dimensional screening method, this comprehensive evaluation method can more accurately identify high-quality users with good response potential and actual participation willingness, thereby improving the accuracy and effectiveness of screening;

[0094] 2. The present invention helps the power grid to achieve more flexible load regulation during periods of tight power supply and demand or peak hours by accurately screening high-quality users to participate in demand response, thereby effectively alleviating power supply pressure and enhancing the stability and reliability of the power system. This is of great significance for ensuring power supply security and reducing the risk of power outages;

[0095] 3. The method of the present invention can allocate power resources more scientifically, make more reasonable use of limited power resources, guide users to adjust their power consumption behavior, realize peak load shaving and valley filling of power load, help reduce power grid operation costs, improve energy utilization efficiency, and promote sustainable development of the power industry;

[0096] 4. When evaluating the user's response ability, the present invention fully considers the user's electricity usage habits and electricity fee sensitivity, which helps to formulate a response strategy that better meets the user's needs; this can not only stimulate the user's enthusiasm for participating in demand response, but also improve the user's satisfaction and loyalty while protecting the user's electricity rights and interests;

[0097] The proposal and implementation of the present invention provide new ideas and methods for technological innovation in the field of demand response user screening. By introducing advanced data processing and analysis technologies such as hierarchical analysis method and entropy weight method, it provides strong technical support for the intelligent management of power systems and the effective implementation of demand response mechanisms. BRIEF DESCRIPTION OF THE DRAWINGS

[0098] Figure 1 It is a flow chart of the present invention;

[0099] Figure 2 Provide a flowchart for preprocessing historical operation data for users;

[0100] Figure 3 This is the flow chart of the entropy weight correction method. DETAILED DESCRIPTION

[0101] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0102] Embodiment 1, as Figures 1 to 3 As shown, a real-time demand response user screening method based on multi-dimensional evaluation indicators includes the following steps:

[0103] S1. Collect user historical operation data and basic information of demand response events, and pre-process user historical operation data;

[0104] S2. From the perspective of the power grid, the effective response load and satisfaction rate of each user in different response periods are determined based on the actual load and baseline load of the user's participation in the last demand response, and the response capacity ratio is further calculated;

[0105] S3. From the user level, the difference between the actual load and the original load after the user participated in the last demand response, as well as the change in electricity bill expenditure before and after the user participated in the last demand response, is used to evaluate the user's electricity consumption mode and electricity bill expenditure satisfaction;

[0106] S4. Based on the four aspects of user response evaluation indicators, a multi-dimensional user response evaluation indicator system is established, and the weights of the indicator system are calculated and corrected using the analytic hierarchy process and entropy weight method;

[0107] S5. Calculate the multi-dimensional response evaluation index of each demand response participating user, perform linear normalization on the multi-dimensional response evaluation index, screen each user as a high-quality user, and obtain high-quality demand response users.

[0108] Embodiment 2: In addition to all the technical features in Embodiment 1, this embodiment also includes: collecting user historical operation data and demand response event basic information, and preprocessing the user historical operation data, specifically including the following steps:

[0109] S11. Collect historical user operation data and basic information of demand response events. The collection cycle is 15 minutes per interval. The user's historical operation data includes the historical operation load at time t on the Nth day of the past week. The basic information of the demand response event includes the last response invitation quantity P invite , the number of the last demand response period N dr , the time period set T of the last demand response event dr 、Time-of-use electricity price unit price λ t out , response subsidy unit price λ subsidy ;

[0110] S12. Use the Median Absolute Deviation (MAD) method to identify missing values ​​and outliers in the user's historical operation data. The specific steps are as follows:

[0111] ① Define the user's historical operating load at time t on day N of the past week Calculate the median of the historical operating load at all times t in the past week (P t history ) where N = {1,...,7}, t∈T history And T history is a collection of historical operating load periods;

[0112] ② Calculate the median of the absolute deviation values ​​at all times t in the past week

[0113] ③ By setting a reasonable proportional factor constant k, usually k = 1.4826, MAD t Approximately equivalent to the standard deviation), so the reasonable range of the historical operating load fluctuation is determined as:

[0114]

[0115] S13, use Lagrange interpolation (LI) to fill in the missing values ​​in the user's historical operation data and correct the abnormal values. For the data points that need to be filled, select the point t N The normal data points at the first m moments and the last m moments are used as samples to calculate the Lagrange polynomial:

[0116]

[0117] Where: t N,i and t N,j are the data sampling points at the i-th and j-th moments of the N-th day in the past week; P N,i is the historical operating load at the i-th moment of the Nth day in the past week; finally, the missing moment t N Substitute P N,t ’ can obtain the filling value of missing and abnormal moments.

[0118] The calculation of the response capacity ratio specifically includes the following steps:

[0119] S21. According to the actual load of the user participating in the last demand response and the baseline load, the user's effective load is identified and the user's demand response period ratio is calculated. The formula is:

[0120]

[0121] Where: η time are the effective load identification variable and the effective response period proportion of the user’s last demand response, respectively; P t actual , are the actual load and baseline maximum load of the user participating in the last demand response at time t, respectively; N dr is the number of the last demand response period; Tdr is the time period set of the last demand response event;

[0122] S22. Calculate the difference between the actual average load of the user participating in the last demand response and the baseline average load based on the effective load of the user participating in the last demand response. The formula is:

[0123]

[0124] Where: ΔP poor P is the difference between the actual average load of the user participating in the last demand response and the baseline average load; base line,t is the baseline load of the user participating in the last demand response at time t;

[0125] According to the difference between the actual average load of the user participating in the last demand response and the average load baseline, the response capacity ratio of the user participating in the last demand response is calculated. The formula is:

[0126]

[0127] Where: η capacity P is the response capacity ratio of users participating in the last demand response; invite The number of invitations to respond to the last demand event.

[0128] The evaluation of user electricity usage and electricity bill expenditure satisfaction includes the following steps:

[0129] S31. The user's electricity consumption satisfaction can be expressed by the difference between the actual load of the user participating in the last demand response and the original electricity load. The formula is:

[0130]

[0131] Where: η way P is the user's satisfaction with electricity consumption in the last demand response; t primitive is the original power load at time t when the user participated in the last demand response; M is the total number of days in the week before the user participated in the last demand response, that is, M = max(N);

[0132] S32. The user's electricity expenditure satisfaction can be expressed by the change between the electricity expenditure after the user participated in the last demand response and the electricity expenditure when the user did not participate in the demand response. The formula is:

[0133]

[0134] Where: η disburse The user's satisfaction with electricity expenditure in the last demand response;t out , subsidy They are time-of-use electricity price unit price and response subsidy unit price respectively.

[0135] The calculation and correction of the weight of the index system by the hierarchical analysis method specifically include:

[0136] AHP can decompose complex problems into several combined factors, group these combined factors according to the dominant relationship, form a top-down hierarchical structure, and obtain the relative importance of factors at the same level by comparing them two by two.

[0137] The main steps of using AHP are:

[0138] ① Determine the evaluation objectives of the evaluation plan, collect the standard criteria based on which to base the evaluation, and determine the evaluation index system;

[0139] ② Establish an evaluation model, which generally consists of a target layer, a criterion layer, and a solution layer, and determine the inclusion factors of each layer;

[0140] ③ Establish a judgment matrix corresponding to the factors at each level. In the evaluation model established in the previous step, the factors contained in each layer are compared with each other using a factor in the upper layer as a comparison criterion; in the established judgment matrix, A=(a ij ) n×n a ij It indicates the importance of the i-th factor in this level relative to the j-th factor; its values ​​are shown in Table 1.

[0141] Table 1 Scale representation and corresponding explanation

[0142] Scale Interpretation 1 i and j have the same importance 3 i is slightly more important than j 5 i is obviously more important than j 7 i is more important than j 9 i is more important than j 2，4，6，8 Indicates that the importance is between the above two

[0143] ④ Solve the maximum eigenvalue of the judgment matrix A and the eigenvector w corresponding to A, and then obtain the weight of the hierarchical factor after normalization;

[0144] ⑤Perform consistency check, for each maximum eigenvalue λ of A max And the eigenvector w, the eigenvector can be regarded as the relative importance scale of the factors at the same level when the factors at the upper level are used as the comparison criteria; because the comparison scale may be inconsistent when the factors at the same level are relatively compared, there may be errors when sorting each level; therefore, consistency check is required, and the steps are as follows:

[0145] First, calculate the consistency index according to the following formula

[0146]

[0147] Secondly, according to n, check the average random consistency index RI ; Finally, calculate the consistency ratio of the judgment matrix, that is, use the following formula to calculate:

[0148]

[0149] If the obtained C R <0.1, then it can be determined that the current judgment matrix meets the consistency requirements; otherwise, the judgment matrix needs to be modified again; R I See Table 2 for the values.

[0150] ⑥ Comprehensively summarize the weight results of each level to get the final conclusion.

[0151] Table 2R I Value

[0152] n 1 2 3 4 5 6 7 8 9 10 <![CDATA[R I ]]> 0 0 0.58 0.90 1.12 1.24 1.32 1.41 1.45 1.49

[0153] The calculation and correction of the index system weight by entropy weight method specifically include:

[0154] Since the AHP only considers the subjective judgment of experts when comparing the elements of each layer, when there are many fuzzy factors in the problem, it may lead to low accuracy and a large difference from the actual situation. Therefore, on the basis of the AHP, the application of the entropy weight method is combined, the AHP and the entropy weight method are combined, and the information of the entropy value is used to correct the subjective judgment results obtained by the AHP method, so as to form a more objective and accurate evaluation result.

[0155] The entropy weight method can calculate the objective weighting results. According to the irregularity of each factor, the weight of each factor can be calculated, and then the weight of each indicator can be corrected according to the entropy weight to obtain a more objective weight result.

[0156] Information represents the order of the system; entropy represents the disorder of the system; if the probability of a system being in a certain situation is p i (i=1,2,...,m), define the system entropy

[0157]

[0158] When p i =1 / m(i=1,2,...,m), the probability of being in each situation is equal, and the entropy takes the corresponding maximum result:

[0159] e=l nm

[0160] e max =lnm

[0161] For the corresponding n evaluation indicators in the m schemes to be evaluated, the evaluation matrix R = (r ij )m×n, for a certain indicator r j There is information entropy:

[0162]

[0163] If the entropy value e j The smaller it is, the more obvious its irregularity is, so it can be considered that this indicator can provide more information and its given weight is greater;

[0164] The entropy weight method calculation process, the specific steps are as follows:

[0165] Assume that there are m projects to be evaluated and n evaluation indicators, forming the original data matrix R = (r ij )m×n:

[0166]

[0167] In the formula, r ij is the evaluation value of i to j. On this basis, the indicator weight is calculated;

[0168] ①The weight of i to j's index p ij :

[0169]

[0170] ② Calculate the entropy value e of the jth indicator j :

[0171]

[0172] ③ Calculate the entropy weight w of the jth indicator j :

[0173]

[0174] ④Determine the comprehensive weight of the indicator β j :

[0175] Using the weight of AHP as α j , where j = 1, 2, ..., n, this weight is combined with the weight of the entropy weight method. The method of combining the subjective weight obtained by the AHP method with the entropy weight is the entropy weight correction method:

[0176]

[0177] Furthermore, the calculation of the multi-dimensional response evaluation index of each demand response participating user specifically includes the following steps:

[0178] S51. Calculation of multi-dimensional user response evaluation indicators:

[0179] U synthesis =β 1 η time +β 2 η capacity +β 3 η way +β 4 η disburse

[0180] Among them, U synthesis It is a multi-dimensional user response evaluation indicator; β 1 , β 2 , β 3 , β 4 They are the weight coefficients of response time proportion, response capacity proportion, electricity consumption mode satisfaction, and electricity expense satisfaction;

[0181] S52. Normalize the multi-dimensional user response evaluation index of the user using a linear ratio normalization method:

[0182]

[0183] Where: is the normalized multi-dimensional user response evaluation index of the i-th user;

[0184] are the maximum and minimum values ​​of the multi-dimensional user response evaluation indicators for all users, respectively.

[0185] S53. Response to user evaluation score:

[0186] According to the normalized multi-dimensional user response evaluation index, the response user evaluation score is calculated. The formula is:

[0187]

[0188] Where: U score In response to user evaluation scores;

[0189] S54. Screening high-quality demand response users

[0190] U high =f(U score )≥0.8

[0191] Where: U high Respond to users' needs for quality.

[0192] Working principle: Collect historical operation data of users and basic information of demand response events, and pre-process historical operation data of users; from the perspective of the power grid, determine the effective response load and satisfaction rate of each user in different response periods according to the actual load and baseline load of the user participating in the last demand response, and further calculate the response capacity ratio; from the perspective of users, evaluate the user's electricity consumption mode and satisfaction with electricity expenditure according to the difference between the actual load of the user participating in the last demand response and the original electricity load, as well as the changes in electricity expenditure before and after the user participated in the last demand response; establish a multi-dimensional user response evaluation index system based on four aspects of user response evaluation indicators, and use the hierarchical analysis method and entropy weight method to calculate and correct the weights of the index system; calculate the multi-dimensional response evaluation index of each demand response participating user, perform linear normalization on the multi-dimensional response evaluation index, screen each user as a high-quality user, and obtain high-quality demand response users.

[0193] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0194] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A real-time demand response user screening method based on multi-dimensional evaluation indicators, characterized in that: The following steps are involved: S1. Collect user historical operation data and basic information of demand response events, and pre-process user historical operation data; S2. From the perspective of the power grid, the effective response load and satisfaction rate of each user in different response periods are determined based on the actual load and baseline load of the user's participation in the last demand response, and the response capacity ratio is further calculated; S3. From the user level, the difference between the actual load and the original load after the user participated in the last demand response, as well as the change in electricity bill expenditure before and after the user participated in the last demand response, is used to evaluate the user's electricity consumption mode and electricity bill expenditure satisfaction; S4. Based on the four aspects of user response evaluation indicators, a multi-dimensional user response evaluation indicator system is established, and the weights of the indicator system are calculated and corrected using the analytic hierarchy process and entropy weight method; S5. Calculate the multi-dimensional response evaluation index of each demand response participating user, perform linear normalization on the multi-dimensional response evaluation index, screen each user as a high-quality user, and obtain high-quality demand response users.

2. The real-time demand response user screening method according to claim 1, characterized in that: In S1, the user's historical operation data and demand response event basic information are collected, and the user's historical operation data is preprocessed, which specifically includes the following steps: S11. Collect historical user operation data and basic information of demand response events. The collection cycle is 15 minutes per interval. The user's historical operation data includes the historical operation load at time t on the Nth day of the past week. The basic information of the demand response event includes the last response invitation quantity P invite , the number of the last demand response period N dr , the time period set T of the last demand response event dr , time-of-use electricity price unit price λtout, response subsidy unit price λ subsidy ; S12. Use the absolute median method to identify missing values ​​and outliers in the user's historical operation data. The specific steps are as follows: ① Define the user's historical operating load at time t on day N of the past week Calculate the median of the historical operating load at all times t in the past week (P t history ) where N = {1,...,7}, t∈T history And T history is a collection of historical operating load periods; ② Calculate the median of the absolute deviation values ​​at all times t in the past week ③ By setting a reasonable proportional factor constant k, taking k = 1.4826, MAD t It is approximately equivalent to the standard deviation, so the reasonable range of the historical operating load fluctuation is determined as: S13, use Lagrange interpolation (LI) to fill in the missing values ​​in the user's historical operation data and correct the abnormal values. For the data points that need to be filled, select the point t N The normal data points at the first m moments and the last m moments are used as samples to calculate the Lagrange polynomial: Where: t N,i and t N,j are the data sampling points at the i-th and j-th moments of the N-th day in the past week; P N,i is the historical operating load at the i-th moment of the Nth day in the past week; finally, the missing moment t N Substitute P N,t ’ can obtain the filling value of missing and abnormal moments.

3. The real-time demand response user screening method according to claim 1, characterized in that: In S2, the calculation of the response capacity ratio specifically includes the following steps: S21. According to the actual load of the user participating in the last demand response and the baseline load, the user's effective load is identified and the user's demand response period ratio is calculated. The formula is: Where: η time are the effective load identification variable and the effective response period proportion of the user’s last demand response, respectively; P t actual , are the actual load and baseline maximum load of the user participating in the last demand response at time t, respectively; N dr is the number of the last demand response period; T dr is the time period set of the last demand response event; S22. Calculate the difference between the actual average load of the user participating in the last demand response and the baseline average load based on the effective load of the user participating in the last demand response. The formula is: Where: ΔP poor P is the difference between the actual average load of the user participating in the last demand response and the baseline average load; base line,t is the baseline load of the user participating in the last demand response at time t; According to the difference between the actual average load of the user participating in the last demand response and the average load baseline, the response capacity ratio of the user participating in the last demand response is calculated. The formula is: Where: η capacity P is the response capacity ratio of users participating in the last demand response; invite The number of invitations to respond to the last demand event.

4. The real-time demand response user screening method according to claim 1, characterized in that: In S3, the evaluation of the user's electricity usage mode and electricity fee expenditure satisfaction specifically includes the following steps: S31. The user's electricity consumption satisfaction can be expressed by the difference between the actual load of the user participating in the last demand response and the original electricity load. The formula is: Where: η way P is the user's satisfaction with electricity consumption in the last demand response; t primitive is the original power load at time t when the user participated in the last demand response; M is the total number of days in the week before the user participated in the last demand response, that is, M = max(N); S32. The user's electricity expenditure satisfaction can be expressed by the change between the electricity expenditure after the user participated in the last demand response and the electricity expenditure when the user did not participate in the demand response. The formula is: Where: η disburse The user's satisfaction with electricity expenditure in the last demand response; t out , subsidy They are time-of-use electricity price unit price and response subsidy unit price respectively.

5. The real-time demand response user screening method according to claim 1, characterized in that: In S4, the calculation and correction of the weight of the index system by the hierarchical analysis method specifically includes: AHP can decompose complex problems into several combined factors, group these combined factors according to the dominant relationship, form a top-down hierarchical structure, and obtain the relative importance of factors at the same level by comparing them two by two. The main steps of using AHP are: ① Determine the evaluation objectives of the evaluation plan, collect the standard criteria based on which to base the evaluation, and determine the evaluation index system; ② Establish an evaluation model, which generally consists of a target layer, a criterion layer, and a solution layer, and determine the inclusion factors of each layer; ③ Establish a judgment matrix corresponding to the factors at each level. In the evaluation model established in the previous step, the factors contained in each layer are compared with each other using a factor in the upper layer as a comparison criterion; in the established judgment matrix, A=(a ij ) n×n a ij Indicates the importance of the i-th factor in this level relative to the j-th factor; ④ Solve the maximum eigenvalue of the judgment matrix A and the eigenvector w corresponding to A, and then obtain the weight of the hierarchical factor after normalization; ⑤Perform consistency check, for each maximum eigenvalue λ of A max And the eigenvector w, the eigenvector can be regarded as the relative importance scale of the factors at the same level when the factors at the upper level are used as the comparison criteria; because the comparison scale may be inconsistent when the factors at the same level are relatively compared, there may be errors when sorting each level; therefore, consistency check is required, and the steps are as follows: First, calculate the consistency index according to the following formula Secondly, according to n, check the average random consistency index R I ; Finally, calculate the consistency ratio of the judgment matrix, that is, use the following formula to calculate: If the obtained C R <0.1, then it can be determined that the current judgment matrix meets the consistency requirements; otherwise, the judgment matrix needs to be modified again; ⑥ Comprehensively summarize the weight results of each level to get the final conclusion.

6. The real-time demand response user screening method according to claim 1, characterized in that: In S4, the calculation and correction of the weight of the indicator system by the entropy weight method specifically includes: Since the AHP only considers the subjective judgment of experts when comparing the elements of each layer, when there are many fuzzy factors in the problem, it may lead to low accuracy and a large difference from the actual situation. Therefore, on the basis of the AHP, the application of the entropy weight method is combined, the AHP and the entropy weight method are combined, and the information of the entropy value is used to correct the subjective judgment results obtained by the AHP method, so as to form a more objective and accurate evaluation result. The entropy weight method can calculate the objective weighting results. According to the irregularity of each factor, the weight of each factor can be calculated, and then the weight of each indicator can be corrected according to the entropy weight to obtain a more objective weight result. Information represents the order of the system; entropy represents the disorder of the system; if the probability of a system being in a certain situation is p i (i=1,2,...,m), define the system entropy When p i =1 / m(i=1,2,...,m), the probability of being in each situation is equal, and the entropy takes the corresponding maximum result: e=lnm it max =1nm For the corresponding n evaluation indicators in the m schemes to be evaluated, the evaluation matrix R = (r ij )m×n, for a certain indicator r j There is information entropy: If the entropy value e j The smaller it is, the more obvious its irregularity is, so it can be considered that this indicator can provide more information and its given weight is greater; The entropy weight method calculation process, the specific steps are as follows: Assume that there are m projects to be evaluated and n evaluation indicators, forming the original data matrix R = (r ij )m×n: In the formula, r ij is the evaluation value of i to j; on this basis, the indicator weight is calculated; ①The weight of i to j's index p ij : ② Calculate the entropy value e of the jth indicator j : ③ Calculate the entropy weight w of the jth indicator j : ④Determine the comprehensive weight of the indicator β j : Using the weight of AHP as α j , where j = 1, 2, ..., n, this weight is combined with the weight of the entropy weight method. The method of combining the subjective weight obtained by the AHP method with the entropy weight is the entropy weight correction method:

7. The real-time demand response user screening method according to claim 1, characterized in that: In S5, the calculation of the multi-dimensional response evaluation index of each demand response participating user specifically includes the following steps: S51. Calculation of multi-dimensional user response evaluation indicators: U synthesis =β1η time +β2η capacity +β3η way +β4η disburse Among them, U synthesis is a multi-dimensional user response evaluation index; β1, β2, β3, and β4 are the weight coefficients of response time proportion, response capacity proportion, electricity consumption mode satisfaction, and electricity bill expenditure satisfaction, respectively; S52. Normalize the multi-dimensional user response evaluation index of the user using a linear ratio normalization method: Where: is the normalized multi-dimensional user response evaluation index of the i-th user; are the maximum and minimum values ​​of the multi-dimensional user response evaluation indicators for all users, respectively. S53. Response to user evaluation score: According to the normalized multi-dimensional user response evaluation index, the response user evaluation score is calculated. The formula is: Where: U score In response to user evaluation scores; S54. Screening high-quality demand response users IN high =f(U score )≥0.8 Where: U high Respond to users' needs for quality.

Images