Load management center demand response dispatching method considering continuous response demand

Abstract

The present application relates to a kind of load management center demand response scheduling method considering continuous response demand, it includes the following steps: the objective function of the multi-objective optimization model of continuous working day demand response scheduling strategy is constructed;The constraint condition of the multi-objective optimization model of continuous working day demand response scheduling strategy is constructed;The multi-objective optimization solution algorithm of demand response scheduling strategy based on Tchebycheff method is proposed to solve the multi-objective optimization model of continuous working day demand response scheduling strategy.The load management center demand response scheduling method considering continuous response demand proposed in the present application can coordinate each power type user to respond in different time periods, avoid the load gap in the later period of continuous demand response, improve the reliability of load management center to carry out the demand response scheduling of continuous multi-day.

Description

Technical Field

[0001] This invention relates to the field of power systems, and in particular to a load management center demand response scheduling method that takes into account continuous response requirements. Background Technology

[0002] Currently, national electricity consumption and peak load are showing a rapid growth trend. The sustained high temperatures have led to a surge in electricity consumption in many provinces. Rising primary energy prices and low water levels are among the factors constraining power supply capacity. The complex and severe situation regarding power supply due to multiple factors on both the supply and demand sides has prompted nearly 20 provincial power grids to implement orderly power consumption measures. In recent years, various provinces in my country have conducted pilot projects for differentiated demand response (DR). However, the domestic demand response market is still in its early stages of development, with widespread problems such as small response scale, low willingness to respond, and an imperfect economic compensation mechanism. There is an urgent need to study refined scheduling strategies for demand-side resources. Many scholars have studied the relationship between demand response volume and response cost in various demand response projects, proposing incentive measures for demand response and control strategies for aggregating resources. However, under extreme weather conditions such as sustained high temperatures, drought, and low winds, the system may experience continuous peak-shaving demand. Demand response scheduling strategies that only consider dispatch costs may result in insufficient demand to be met in the later stages of multi-day demand responses, thus affecting the security of the power system. Summary of the Invention

[0003] To address the problems existing in the above-mentioned background technology, the present invention proposes a load management center demand response scheduling method that considers continuous response requirements.

[0004] This invention is achieved using the following technical solution:

[0005] A load management center demand response scheduling method that considers continuous response requirements includes the following steps:

[0006] 1) Construct the objective function of a multi-objective optimization model for the continuous working day demand response scheduling strategy;

[0007] 2) Construct a multi-objective optimization model for the continuous working day demand response scheduling strategy, and define its constraints.

[0008] 3) A multi-objective optimization algorithm based on the Tchebycheff method is proposed to solve the multi-objective optimization model of the continuous working day demand response scheduling strategy.

[0009] In the above technical solution, further, in step 1), the objective function of the multi-objective optimization model for the continuous working day demand response scheduling strategy is constructed as follows:

[0010] Considering the uncertainty of continuous demand response and user demand response within the week, a multi-objective optimization model for continuous weekday demand response scheduling strategy is constructed with the objectives of minimizing the call cost of the load management center and maximizing user satisfaction with continuous response. The demand response call cost comprises three parts: the daytime peak-shaving demand response electricity cost, the valley-filling demand response capacity cost, and the intraday hourly peak-shaving demand response electricity cost. User satisfaction with continuous response is the sum of the satisfaction levels of all users. Therefore, the objective function of the multi-objective optimization can be expressed as:

[0011]

[0012]

[0013]

[0014] In the formula, f1 and f2 are the objective functions for the load management center's call cost and user continuous response satisfaction, respectively. and These represent the load management center call cost and user continuous response satisfaction on day d, respectively. The electricity price responds to user i's peak demand at time t on day d; Electricity pricing is adjusted to meet peak demand during the day. To meet demand in the valley of demand, capacity pricing is required. The demand response power for all hourly peak-shaving demand response users at time t on day d; The capacity to meet user i's valley filling demand on day d; For industrial user i, the satisfaction level with the demand response on day d; P P,i,d,t To meet the peak demand response load of user i at time t on day d; For user i, the satisfaction level with the response to their needs on day d; P P,i,d,t To handle the peak demand response load for user i at time t on day d.

[0015] User satisfaction with response to user needs can be expressed as:

[0016]

[0017] In the formula, and These are the daily call count metric, power consumption reduction metric, and multi-day continuous response rate metric for user i; α i,j (j=1,2,3) are the weight coefficients of each indicator for user i, satisfying It is determined by the user based on their subjective wishes.

[0018] 1) Daily call count metric

[0019] The daily call count metric represents the relationship between the total number of peak shaving and valley filling demand response control commands received by a user on a given day and the user's contracted maximum daily scheduling count. It reflects the impact of control frequency on user production scheduling and can be expressed as:

[0020]

[0021] In the formula, v i,d,t The flag for the start of demand response regulation for user i at time t on day d is a binary variable. A value of 1 indicates that the user starts peak shaving or valley filling demand response at that time. The maximum number of daily calls allowed for user i to sign up.

[0022] 2) Electricity reduction index

[0023] The power reduction index represents the relationship between a user's total daily power reduction and the baseline total power consumption. When a user participates in both peak shaving and valley filling demand response, they can compensate for lost output during valley filling periods. The total daily power reduction reflects the impact of demand response on user output, and can be expressed as follows:

[0024]

[0025] P i,d,t =P P,i,d,t -P V,i,d,t

[0026] In the formula, P i,d,t P represents the demand response load reduction for user i at time t on day d. This value is positive when participating in peak shaving demand response and negative when participating in valley filling demand response. P,i,d,t and P V,i,d,t These represent the peak-shaving demand response load and valley-filling demand response load for user i at time t on day d; Q i,d This represents the baseline total electricity consumption of user i on day d.

[0027] 3) Multi-day continuous response level indicators

[0028] The multi-day continuous response rate index represents the relationship between the number of times a user's request response is scheduled over multiple consecutive days and the average number of times all users respond. Since the impact of historical continuous call occurrences on the current day decreases, a smaller weight is assigned to response counts from more distant dates, while a larger weight is assigned to response counts from more recent dates. An exponential smoothing method is used to apply a weighted average to the historical continuous response rate, which can be expressed as follows:

[0029]

[0030]

[0031] In the formula, ρ is the weighted average of the number of consecutive historical calls made by user i on day d; H This is the weighting coefficient for exponential smoothing. Since responses from dates closer to the original response have a higher weight, this coefficient must satisfy ρ. H ∈[0.5,1], when ρ H When the value is 1, the multi-day continuous response index only considers the number of responses on the previous day. As a normalized baseline value, this invention takes the maximum value among the maximum daily call counts of all users' subscriptions; N D This represents the total number of users who participated in the demand response.

[0032] Step 2) establishes the constraints for the multi-objective optimization model of the continuous working day demand response scheduling strategy, as follows:

[0033] The constraints of the demand response scheduling strategy model include load reduction index constraints, regulation potential constraints, regulation interval time constraints, maximum number of regulation constraints, regulation duration constraints, and response uncertainty constraints, expressed as follows:

[0034] 1) Pressure drop load index constraint

[0035] The day-ahead demand response load and intraday demand response load must be higher than the load management center's drop load target, expressed as:

[0036]

[0037] In the formula, This represents the load drop index of the load management center at time t on day d.

[0038] 2) Adjustment potential constraints

[0039]

[0040]

[0041]

[0042] In the formula, u P,i,d,t and u V,i,d,t These are the flags for user i's peak-shaving and valley-filling demand response at time t on day d, and are binary variables; The potential response to user i's needs at time t; i,d,t For industrial user i, the demand response fault state variable is l when the user actually issues a load control command. i,d,t If the value is 1, otherwise it is considered a fault state. i,d,t It is 0.

[0043] 3) Adjusting the duration constraint

[0044] v P,i,d,t +z P,i,d,t ≤1

[0045] u P,i,d,t -u P,i,d,t-1 =v P,i,d,t -z P,i,d,t

[0046] v V,i,d,t +z V,i,d,t ≤1

[0047] u V,i,d,t -u V,i,d,t-1 =v V,i,d,t -z V,i,d,t

[0048]

[0049]

[0050]

[0051]

[0052]

[0053]

[0054] In the formula, v P,i,d,t / v V,i,d,t These are the flags indicating the start of peak shaving / valley filling demand response calls. A value of 1 indicates that user i starts peak shaving / valley filling demand response at time t on day d; z V,i,d,t / z P,i,d,t These are the flags indicating the end of the peak shaving / valley filling demand response call. A value of 1 indicates that user i ended the peak shaving / valley filling demand response at time t on day d. This indicates the duration of peak shaving / valley filling demand response for user i at time t on day d; These represent the maximum duration of peak-shaving / valley-filling demand response for user i; Let be the minimum peak-shaving / valley-filling demand response duration for user i. The Big-M method is used to linearize the above constraints, transforming them into...

[0055]

[0056]

[0057] 4) Adjusting the interval time constraint

[0058]

[0059]

[0060]

[0061]

[0062] In the formula, These represent the peak-shaving / valley-filling demand response intervals for user i at time t on day d; Let be the minimum demand response interval for user i at time t on day d. The Big-M method is used to linearize the above constraints, transforming them into...

[0063]

[0064]

[0065] 5) Maximum number of adjustments constraint

[0066] In user request response contracts, a daily maximum number of adjustments is typically limited, while a weekly limit on the number of calls is set to ensure user comfort. Therefore, the maximum number of adjustments constraint can be expressed as follows:

[0067] v i,d,t =v P,i,d,t +v V,i,d,t

[0068]

[0069]

[0070] In the formula, and These represent the maximum number of adjustments per day and the maximum number of adjustments per week for user i, respectively.

[0071] 6) Responding to uncertainty constraints

[0072] To ensure power balance under the condition of minimum demand-side reserve response, the improved Nk uncertainty constraint must be satisfied.

[0073]

[0074]

[0075]

[0076]

[0077] g t ≥0,h i,t ≥0

[0078] In the formula, and For user i, the peak-shaving / valley-filling demand response load actually invoked at time t on day d; g t and h i,t There are two dual variables.

[0079] Furthermore, step 3) proposes a multi-objective optimization algorithm for the demand response scheduling strategy based on the Tchebycheff method, as follows:

[0080] (1) The objective function, after being transformed by the Tchebycheff method, can be expressed as follows:

[0081]

[0082] In the formula, f i OP This represents the ideal point for each objective, i.e., the optimal function value obtained by solving for f1 and f2 as single objectives respectively. The solution objective of the model consists of two parts: the maximum deviation between each objective and its ideal point is the main optimization part, α. i The weights for each objective are α1 + α2 = 1; the second term of the objective can avoid the uncertainty brought to the solution by the local ill-conditioned nature of the feasible region of the model, M S It is a very small constant, taken as 0.0001.

[0083] Considering the difference in dimensions between f1 and f2, normalization is required. Therefore, it is further transformed into

[0084]

[0085] In the formula, f i NE Let be the negative ideal point of each objective, and be the function value obtained by solving for each other as a single objective; let α i By substituting values ​​from 0 to 1 into the objective function at a certain step size, the multi-objective optimization model of the continuous working day demand response scheduling strategy is solved, and the Pareto front solution set is obtained.

[0086] (2) The analytic hierarchy process (AHP) is combined with the entropy weight method to comprehensively consider subjective and objective weights. By calculating the comprehensive weight, an optimal demand response scheduling strategy is finally determined in the Pareto front solution set.

[0087] First, based on the importance relationships between elements, a analytic hierarchy process (AHP) judgment matrix is ​​formed.

[0088]

[0089] In the formula, o mnIndicates the importance relationship between two elements;

[0090] The subjective weighting factors of each objective function can be expressed as:

[0091]

[0092] U = [u1, u2, ..., u n ] T

[0093] In the formula, U is the subjective weighting factor of objective function i; U is the largest eigenvalue λ. max The corresponding feature vector, u i Let i be the i-th element in U;

[0094] Then, the objective weighting algorithm of the entropy weight method is used to calculate the objective weighting factors of each objective function, including the following steps:

[0095] Step 1: Forming a data matrix

[0096]

[0097] In the formula, X ji Let m be the value of the i-th objective function in the j-th Pareto solution, m be the total number of Pareto front solutions, and n be the total number of objective functions;

[0098] Step 2: Non-negative data processing

[0099] Since the entropy method calculates the ratio of a certain indicator to the total value of the same indicator in each scheme, it is not affected by dimensions and does not require standardization. If the data contains negative numbers, it needs to be non-negative. Furthermore, to avoid the logarithm becoming meaningless when calculating entropy, data shifting is necessary.

[0100]

[0101] Step 3: Calculate the proportion of the j-th solution to the i-th objective function.

[0102]

[0103] Step 4: Calculate the entropy of the i-th objective function.

[0104]

[0105] Step 5: Calculate the difference coefficient g of the i-th objective function. i

[0106] For the i-th objective function, the index value X ijThe greater the difference, the greater its impact on the evaluation of the scheme, and the smaller the entropy value. Let g j =1-e j Then we have g j The larger the indicator, the more important it is.

[0107] Step Six: Calculate the objective weights of the objective function i, which can be expressed as...

[0108]

[0109] Obtain subjective weighting factors and objective weighting factors Then, based on the consistency of subjective and objective attribute values, subjective weights and objective weights are combined proportionally to obtain combined weights. An optimization model is constructed with the objective of minimizing the sum of subjective and objective deviations among all power generation enterprises. The objective function of the model can be expressed as:

[0110]

[0111] In the formula, ω is the allocation coefficient between subjective weight and objective weight;

[0112] Then, the combined weights are calculated using the optimal subjective and objective weight allocation coefficients, expressed as follows:

[0113]

[0114] In the formula, The optimal combination weights for objective function i are... The optimal demand response invocation strategy is obtained by solving the model with weight coefficients.

[0115] The beneficial effects of this invention are:

[0116] The method of this invention can effectively complete demand response scheduling. The model constructed by this invention can reflect multi-dimensional user satisfaction with the limits of the number of calls, the degree of power reduction, and the degree of continuous response over multiple days under continuous response. Compared with traditional scheduling methods, the method proposed in this invention can effectively avoid load gaps in the later stages of continuous demand response and improve the reliability of demand response calls carried out by the load management center over multiple days. Attached Figure Description

[0117] Figure 1 This is a flowchart of a load management center demand response scheduling method considering continuous response requirements in an embodiment of the present invention;

[0118] Figure 2 This is a flowchart of the multi-objective optimization algorithm for the demand response scheduling strategy based on the Tchebycheff method in this embodiment of the invention.

[0119] Figure 3The results of user demand response for each electricity consumption type after scheduling using the method of the present invention (a) and the existing method (b) are shown. Detailed Implementation

[0120] To better understand the purpose, technical solution, and technical effects of this invention, the invention will be further explained below in conjunction with the accompanying drawings.

[0121] refer to Figure 1 , Figure 1 The diagram illustrates a specific load management center demand response scheduling method that considers continuous response requirements, comprising the following steps:

[0122] S10, Construct the objective function of the multi-objective optimization model for the continuous working day demand response scheduling strategy:

[0123] Considering the uncertainty of continuous demand response and user demand response within the week, a multi-objective optimization model for continuous weekday demand response scheduling strategy is constructed with the objectives of minimizing the call cost of the load management center and maximizing user satisfaction with continuous response. The demand response call cost comprises three parts: the daytime peak-shaving demand response electricity cost, the valley-filling demand response capacity cost, and the intraday hourly peak-shaving demand response electricity cost. User satisfaction with continuous response is the sum of the satisfaction levels of all users. Therefore, the objective function of the multi-objective optimization can be expressed as:

[0124]

[0125]

[0126]

[0127] In the formula, f1 and f2 are the objective functions for the load management center's call cost and user continuous response satisfaction, respectively. and These represent the load management center call cost and user continuous response satisfaction on day d, respectively. The electricity price responds to user i's peak demand at time t on day d; Electricity pricing is adjusted to meet peak demand during the day. To meet demand in the valley of demand, capacity pricing is required. The demand response power for all hourly peak-shaving demand response users at time t on day d; The capacity to meet user i's valley filling demand on day d; For industrial user i, the satisfaction level with the demand response on day d; P P,i,d,t To meet the peak demand response load of user i at time t on day d; For user i, the satisfaction level with the response to their needs on day d; P P,i,d,tTo handle the peak demand response load for user i at time t on day d.

[0128] User satisfaction with response to user needs can be expressed as:

[0129]

[0130] In the formula, and These are the daily call count metric, power consumption reduction metric, and multi-day continuous response rate metric for user i; α i,j (j=1,2,3) are the weight coefficients of each indicator for user i, satisfying It is determined by the user based on their subjective wishes.

[0131] 1) Daily call limit index

[0132] The daily call count metric represents the relationship between the total number of peak shaving and valley filling demand response control commands received by a user on a given day and the user's contracted maximum daily scheduling count. It reflects the impact of control frequency on user production scheduling and can be expressed as:

[0133]

[0134] In the formula, v i,d,t The flag for the start of demand response regulation for user i at time t on day d is a binary variable. A value of 1 indicates that the user starts peak shaving or valley filling demand response at that time. The maximum number of daily calls allowed for user i to sign up.

[0135] 2) Electricity reduction index

[0136] The power reduction index represents the relationship between a user's total daily power reduction and the baseline total power consumption. When a user participates in both peak shaving and valley filling demand response, they can compensate for lost output during valley filling periods. The total daily power reduction reflects the impact of demand response on user output, and can be expressed as follows:

[0137]

[0138] P i,d,t =P P,i,d,t -P V,i,d,t

[0139] In the formula, P i,d,t P represents the demand response load reduction for user i at time t on day d. This value is positive when participating in peak shaving demand response and negative when participating in valley filling demand response. P,i,d,t and P V,i,d,t These represent the peak-shaving demand response load and valley-filling demand response load for user i at time t on day d; Q i,dThis represents the baseline total electricity consumption of user i on day d.

[0140] 3) Multi-day continuous response level indicators

[0141] The multi-day continuous response rate index represents the relationship between the number of times a user's request response is scheduled over multiple consecutive days and the average number of times all users respond. Since the impact of historical continuous call occurrences on the current day decreases, a smaller weight is assigned to response counts from more distant dates, while a larger weight is assigned to response counts from more recent dates. An exponential smoothing method is used to apply a weighted average to the historical continuous response rate, which can be expressed as follows:

[0142]

[0143]

[0144] In the formula, ρ is the weighted average of the number of consecutive historical calls made by user i on day d; H This is the weighting coefficient for exponential smoothing. Since responses from dates closer to the original response have a higher weight, this coefficient must satisfy ρ. H ∈[0.5,1], when ρ H When the value is 1, the multi-day continuous response index only considers the number of responses on the previous day. As a normalized baseline value, this invention takes the maximum value among the maximum daily call counts of all users' subscriptions; N D This represents the total number of users who participated in the demand response.

[0145] S20, Constraints for constructing a multi-objective optimization model for continuous workday demand response scheduling strategy:

[0146] The constraints of the demand response scheduling strategy model include load reduction index constraints, regulation potential constraints, regulation interval time constraints, maximum number of regulation constraints, regulation duration constraints, and response uncertainty constraints, expressed as follows:

[0147] 1) Pressure drop load index constraint

[0148] The day-ahead demand response load and intraday demand response load must be higher than the load management center's drop load target, expressed as:

[0149]

[0150] In the formula, This represents the load drop index of the load management center at time t on day d.

[0151] 2) Adjustment potential constraints

[0152]

[0153]

[0154]

[0155] In the formula, u P,i,d,t and u V,i,d,t These are the flags for user i's peak-shaving and valley-filling demand response at time t on day d, and are binary variables; The potential response to user i's needs at time t; i,d,t For industrial user i, the demand response fault state variable is l when the user actually issues a load control command. i,d,t If the value is 1, otherwise it is considered a fault state. i,d,t It is 0.

[0156] 3) Adjusting the duration constraint

[0157] v P,i,d,t +z P,i,d,t ≤1

[0158] u P,i,d,t -u P,i,d,t-1 =v P,i,d,t -z P,i,d,t

[0159] v V,i,d,t +z V,i,d,t ≤1

[0160] u V,i,d,t -u V,i,d,t-1 =v V,i,d,t -z V,i,d,t

[0161]

[0162]

[0163]

[0164]

[0165]

[0166]

[0167] In the formula, v P,i,d,t / v V,i,d,t These are the flags indicating the start of peak shaving / valley filling demand response calls. A value of 1 indicates that user i starts peak shaving / valley filling demand response at time t on day d; z V,i,d,t / z P,i,d,t These are the flags indicating the end of the peak shaving / valley filling demand response call. A value of 1 indicates that user i ended the peak shaving / valley filling demand response at time t on day d. This indicates the duration of peak shaving / valley filling demand response for user i at time t on day d; These represent the maximum duration of peak-shaving / valley-filling demand response for user i; Let be the minimum peak-shaving / valley-filling demand response duration for user i. The Big-M method is used to linearize the above constraints, transforming them into...

[0168]

[0169]

[0170] 4) Adjusting the interval time constraint

[0171]

[0172]

[0173]

[0174]

[0175] In the formula, These represent the peak-shaving / valley-filling demand response intervals for user i at time t on day d; Let be the minimum demand response interval for user i at time t on day d. The Big-M method is used to linearize the above constraints, transforming them into...

[0176]

[0177]

[0178] 5) Maximum number of adjustments constraint

[0179] In user request response contracts, a daily maximum number of adjustments is typically limited, while a weekly limit on the number of calls is set to ensure user comfort. Therefore, the maximum number of adjustments constraint can be expressed as follows:

[0180] v i,d,t =v P,i,d,t +v V,i,d,t

[0181]

[0182]

[0183] In the formula, and These represent the maximum number of adjustments per day and the maximum number of adjustments per week for user i, respectively.

[0184] 6) Responding to uncertainty constraints

[0185] To ensure power balance under the condition of minimum demand-side reserve response, the improved Nk uncertainty constraint must be satisfied.

[0186]

[0187]

[0188]

[0189]

[0190] g t ≥0,h i,t ≥0

[0191] In the formula, and For user i, the peak-shaving / valley-filling demand response load actually invoked at time t on day d; g t and h i,t There are two dual variables.

[0192] S30, a multi-objective optimization algorithm based on the Tchebycheff method is proposed to solve the multi-objective optimization model of the continuous workday demand response scheduling strategy:

[0193] Figure 2 The diagram shows the flow of a multi-objective optimization algorithm for a demand response scheduling strategy based on the Tchebycheff method.

[0194] (1) The objective function, after transformation using the Tchebycheff method, is expressed as follows:

[0195]

[0196] In the formula, f i OP This represents the ideal point for each objective, i.e., the optimal function value obtained by solving for f1 and f2 as single objectives respectively. The solution objective of the model consists of two parts: the maximum deviation between each objective and its ideal point is the main optimization part, α. i The weights for each objective are α1 + α2 = 1; the second term of the objective can avoid the uncertainty brought to the solution by the local ill-conditioned nature of the feasible region of the model, M S It is a very small constant, which can be 0.0001.

[0197] Considering the difference in dimensions between f1 and f2, normalization is required. Therefore, it is further transformed into

[0198]

[0199] In the formula, f i NE Let be the negative ideal point of each objective, and be the function value obtained by solving for each other as a single objective; let α i By substituting values ​​from 0 to 1 into the objective function at a certain step size, the multi-objective optimization model of the continuous working day demand response scheduling strategy is solved, and the Pareto front solution set is obtained.

[0200] (2) The analytic hierarchy process (AHP) is combined with the entropy weight method to comprehensively consider subjective and objective weights. By calculating the comprehensive weight, an optimal demand response scheduling strategy is finally determined in the Pareto front solution set.

[0201] First, based on the importance relationships between elements, a analytic hierarchy process (AHP) judgment matrix is ​​formed.

[0202]

[0203] In the formula, o mn It indicates the importance relationship between two elements.

[0204] The subjective weighting factors of each objective function can be expressed as:

[0205]

[0206] U = [u1, u2, ..., u n ] T

[0207] In the formula, U is the subjective weighting factor of objective function i; U is the largest eigenvalue λ. max The corresponding feature vector, u i Let i be the i-th element in U.

[0208] Then, the objective weighting algorithm of the entropy weight method is used to calculate the objective weighting factors of each objective function, including the following steps:

[0209] Step 1: Forming a data matrix

[0210]

[0211] In the formula, X ji Let m be the value of the i-th objective function in the j-th Pareto solution, m be the total number of Pareto front solutions, and n be the total number of objective functions.

[0212] Step 2: Non-negative data processing

[0213] Since the entropy method calculates the ratio of a certain indicator to the total value of the same indicator in each scheme, it is not affected by dimensions and does not require standardization. If the data contains negative numbers, it needs to be non-negative. Furthermore, to avoid the logarithm becoming meaningless when calculating entropy, data shifting is necessary.

[0214]

[0215] Step 3: Calculate the proportion of the j-th solution to the i-th objective function.

[0216]

[0217] Step 4: Calculate the entropy of the i-th objective function.

[0218]

[0219] Step 5: Calculate the difference coefficient g of the i-th objective function. i

[0220] For the i-th objective function, the index value X ij The greater the difference, the greater its impact on the evaluation of the scheme, and the smaller the entropy value. Let g j =1-e j Then we have g j The larger the indicator, the more important it is.

[0221] Step Six: Calculate the objective weights of the objective function i, which can be expressed as...

[0222]

[0223] Obtain subjective weighting factors and objective weighting factors Then, based on the consistency of subjective and objective attribute values, subjective weights and objective weights are combined proportionally to obtain combined weights. An optimization model is constructed with the objective of minimizing the sum of subjective and objective deviations among all power generation enterprises. The objective function of the model can be expressed as:

[0224]

[0225] In the formula, ω is the allocation coefficient between subjective weight and objective weight.

[0226] Then, the combined weights are calculated using the optimal subjective and objective weight allocation coefficients, expressed as follows:

[0227]

[0228] In the formula, The optimal combination weights for objective function i are... The optimal demand response invocation strategy is obtained by solving the model with weight coefficients.

[0229] This study analyzes electricity users in a specific industry within a certain region, totaling 263 users. These users are categorized into five types: The first type is a typical peak-flat electricity consumption type, characterized by large-scale production, high automation levels, and near-24-hour operation, resulting in high and stable electricity consumption throughout the day; the second type is a typical peak-avoidance electricity consumption type, with concentrated electricity consumption at night, leading to higher nighttime consumption; the third, fourth, and fifth types are daytime double-peak electricity consumption types, but due to differences in production patterns and overtime habits, the duration of peak periods varies, and are referred to below as fixed double-peak electricity consumption, gradual double-peak electricity consumption, and shutdown double-peak electricity consumption, respectively. The minimum response interval is set at 2 hours, the valley-filling demand response capacity subsidy price is 5 yuan / (kW·day), and the hourly peak-shaving demand response electricity subsidy price is 4 yuan / kWh. The study focuses on a 5-day continuous high peak-shaving demand response index during periods of sustained high temperatures in summer. After demand response scheduling using the method proposed in this invention, there is no risk of load shedding, and demand response indicators for all time periods can be met. The optimal demand response results for each type of electricity user are shown in the appendix. Figure 3 As shown. Compared to traditional scheduling methods, the method proposed in this invention can effectively avoid load gaps in the later stages of continuous demand response, and improve the reliability of demand response calls that the load management center can make over several days.

Claims

1. A load management center demand response scheduling method considering continuous response requirements, characterized in that, Includes the following steps: 1) Construct the objective function of a multi-objective optimization model for the continuous workday demand response scheduling strategy; the specific method is as follows: Considering the uncertainty of continuous demand response and user demand response within the week, a multi-objective optimization model for continuous weekday demand response scheduling strategy is constructed with the objectives of minimizing the call cost of the load management center and maximizing user satisfaction with continuous response. The demand response call cost comprises three parts: daytime peak shaving demand response power cost, valley filling demand response capacity cost, and intraday hourly peak shaving demand response power cost. User continuous response satisfaction is the sum of the satisfaction levels of all users. Therefore, the objective function for multi-objective optimization is expressed as: In the formula, and The objective functions are the load management center call cost and the user's satisfaction with continuous response, respectively. and These represent the load management center call cost and user continuous response satisfaction on day d, respectively. The total number of all users who participated in the demand response; The electricity price responds to user i's peak demand at time t on day d; Electricity pricing is adjusted to meet peak demand during the day. To meet demand in the valley of demand, the price of capacity should be adjusted accordingly. The demand response power for all hourly peak-shaving demand response users at time t on day d; The capacity to meet user i's valley filling demand on day d; The satisfaction level of industrial user i with the demand response on day d; To meet the peak demand response load of user i at time t on day d; User satisfaction with response to user needs is expressed as: In the formula, , and These are the daily call count, power reduction, and multi-day continuous response indicators for user i. The daily call count indicator represents the relationship between the total number of peak shaving and valley filling demand response control commands received by the user on a given day and the maximum daily scheduling count that the user has signed up for, reflecting the impact of control frequency on the user's production schedule. The power reduction indicator represents the relationship between the user's total power reduction on a given day and the baseline total power consumption. When a user participates in both peak shaving and valley filling demand response, they can make up for lost production during valley filling periods. The total power reduction within a day reflects the impact of demand response on the user's production. The multi-day continuous response rate index represents the relationship between the number of times a user's demand response is scheduled over multiple consecutive days and the average situation for all users. Let be the weight coefficients of each indicator for user i, satisfying It is determined by the user based on their subjective wishes; 2) Construct a multi-objective optimization model for the continuous working day demand response scheduling strategy, including constraints on load reduction indicators, adjustment potential, adjustment interval time, maximum number of adjustments, adjustment duration, and response uncertainty. 3) A multi-objective optimization algorithm based on the Tchebycheff method is proposed to solve the multi-objective optimization model of the continuous working day demand response scheduling strategy and obtain the Pareto front solution set. The analytic hierarchy process is combined with the entropy weight method to comprehensively consider subjective and objective weights. By calculating the comprehensive weight, an optimal demand response scheduling strategy is finally determined in the Pareto front solution set.

2. The load management center demand response scheduling method considering continuous response requirements according to claim 1, characterized in that, In step 1): 1) The daily call count metric is expressed as: In the formula, The flag for the start of demand response regulation for user i at time t on day d is a binary variable. A value of 1 indicates that the user starts peak shaving or valley filling demand response at that time. The maximum number of daily calls allowed for user i to sign up; 2) The degree of power reduction is expressed as: In the formula, This represents the load reduction for user i at time t on day d. This value is positive when participating in peak shaving demand response and negative when participating in valley filling demand response. and These are the peak-shaving demand response load and valley-filling demand response load for user i at time t on day d, respectively. This represents the baseline total electricity consumption of user i on day d. 3) The multi-day continuous response level indicator is expressed as: In the formula, The weighted average of the number of consecutive calls made by user i on day d. The weighting coefficients for exponential smoothing must satisfy the following conditions: ,when When the value is 1, the multi-day continuous response index only considers the number of responses on the previous day; The normalized baseline value is the maximum value among the daily maximum number of calls signed by all users.

3. The load management center demand response scheduling method considering continuous response requirements according to claim 1, characterized in that, Step 2) establishes the constraints for the multi-objective optimization model of the continuous working day demand response scheduling strategy, as follows: The constraints of the demand response scheduling strategy model are expressed as follows: 1) Pressure drop load index constraint The day-ahead demand response load and intraday demand response load must be higher than the load management center's drop load target, expressed as: In the formula, The load drop index of the load management center at time t on day d; 2) Adjustment potential constraints In the formula, and These are the flags for user i's peak-shaving and valley-filling demand response at time t on day d, and are binary variables; The potential to respond to user i's needs at time t; For industrial user i, the demand response fault state variable is defined when the user actually issues a load control command. If the value is 1, then it is considered a fault state. =0; 3) Adjusting the duration constraint In the formula, / These are the flags for initiating peak shaving / valley filling demand response calls. A value of 1 indicates that user i starts peak shaving / valley filling demand response at time t on day d. / These are the flags indicating the end of the peak shaving / valley filling demand response call. A value of 1 indicates that user i ended the peak shaving / valley filling demand response at time t on day d. / This indicates the duration of peak shaving / valley filling demand response for user i at time t on day d; / These represent the maximum duration of peak-shaving / valley-filling demand response for user i; / These represent the minimum peak-shaving / valley-filling demand response duration for user i; The Big-M method is used to linearize the above constraints, transforming them into... 4) Adjusting the interval time constraint In the formula, / These represent the peak-shaving / valley-filling demand response intervals for user i at time t on day d; / These represent the minimum interval for peak shaving / valley filling demand response for user i at time t on day d; The Big-M method is used to linearize the above constraints, transforming them into... 5) Maximum number of adjustments constraint In the formula, and These represent the maximum number of adjustments per day and the maximum number of adjustments per week for user i, respectively. 6) Responding to uncertainty constraints In the formula, and The peak-shaving / valley-filling demand response load actually invoked by user i at time t on day d; and There are two dual variables.

4. The load management center demand response scheduling method considering continuous response requirements according to claim 1, characterized in that, Step 3) proposes a multi-objective optimization algorithm for the demand response scheduling strategy based on the Tchebycheff method, as follows: (1) The objective function, after being transformed by the Tchebycheff method, is expressed as follows: In the formula, Representing the ideal point of each objective, that is, respectively using and The optimal function value obtained for a single objective is used; the solution objective of the model consists of two parts, with the maximum deviation of each objective from its ideal point being the main optimization component. The weighting factors for each objective satisfy... The second objective term can avoid the uncertainty introduced by local ill-conditioned conditions in the feasible region of the model in the solution. It is a constant, taken as 0.0001; Considering and The dimensions of these quantities differ, requiring normalization. Therefore, they are transformed into... In the formula, Let be the negative ideal point of each objective, and be the function value obtained by solving for each of the other objectives as a single objective; By substituting 0 to 1 into the objective function with a certain step size, the multi-objective optimization model of the continuous working day demand response scheduling strategy is solved to obtain the Pareto front solution set. (2) The analytic hierarchy process is combined with the entropy weight method to comprehensively consider subjective and objective weights. By calculating the comprehensive weights, an optimal demand response scheduling strategy is finally determined in the Pareto front solution set. First, based on the importance relationships between elements, a analytic hierarchy process (AHP) judgment matrix is ​​formed. In the formula, Indicates the importance relationship between two elements; The subjective weighting factors of each objective function are expressed as follows: In the formula, U is the subjective weighting factor for the objective function i; U is the largest eigenvalue. The corresponding feature vector, Let i be the i-th element in U; Then, the objective weighting algorithm of the entropy weight method is used to calculate the objective weighting factors of each objective function, including the following steps: Step 1: Forming a data matrix In the formula, Let m be the value of the i-th objective function in the j-th Pareto solution, m be the total number of Pareto front solutions, and n be the total number of objective functions. Step 2: Nonnegation of data Step 3: Calculate the proportion of the j-th solution to the i-th objective function. Step 4: Calculate the entropy of the i-th objective function. Step 5: Calculate the difference coefficient g of the i-th objective function. i Step Six: Calculate the objective weights of the objective function i, expressed as... Obtain subjective weighting factors and objective weighting factors Then, based on the consistency of subjective and objective attribute values, subjective weights and objective weights are combined proportionally to obtain combined weights. An optimization model is constructed with the objective of minimizing the sum of subjective and objective deviations among all power generation enterprises. The objective function of the model is expressed as: In the formula, This is the allocation coefficient between subjective and objective weights; Then, the combined weights are calculated using the optimal subjective and objective weight allocation coefficients, expressed as follows: In the formula, The optimal combination weights for objective function i are... The optimal demand response invocation strategy is obtained by solving the model with weight coefficients.

Images