A node load curve method considering demand response uncertainty

Abstract

The application discloses a kind of node load quasi-line method in the technical field of electric power system considering demand response uncertainty, comprising the following steps: constructing node load quasi-line calculation model;Model transformation and solution;Based on node load quasi-line, DR is carried out;The application proposes a kind of node load quasi-line method considering demand response uncertainty, issues respective load quasi-line for each node adjustable load, to be used as the guiding target of each node load participating in DR;Considering the uncertainty of adjustable load response, a robust and safe node load quasi-line calculation model is established;Based on the node load quasi-line obtained by the proposed model, large-scale DR is carried out, which can improve the economic efficiency of system operation while effectively ensuring the power flow safety of power grid, i.e., when adjustable load responds to DR within the range reported in advance, the power grid does not exist line power out-of-limit.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power systems, in particular to a node load quasi-line method considering demand response uncertainty. BACKGROUND

[0002] The regulation capability of the power supply side of the new power system is greatly weakened, and it is difficult to guarantee the reliable power supply and safe and stable operation of the new power system only by relying on the regulation capability of the power supply side. Demand response (DR) can tap the huge regulation potential of a large number of flexible loads on the demand side of the power, such as air conditioners, water heaters, electric vehicles and the like. In order to alleviate the huge regulation burden that the new power system will continue to face, DR should be carried out on a large scale and in a normal state.

[0003] However, the participation of a large number of adjustable loads with uncertainty in DR will lead to changes in the power flow of the power grid, bringing challenges to the safe operation of the power grid. The existing DR methods mainly include three types of price-type DR, baseline-type DR and quasi-line-type DR. The price-type DR is guided by the electricity price, and usually converts the network safety constraints into load node interaction response constraints to ensure the safe operation of the power grid during the implementation of DR, but does not consider the influence of the uncertainty of the adjustable load response on the system safety. The baseline-type DR eliminates the power flow out-of-limit of the power grid itself through load interruption and direct load control, but lacks research on the system safety problems caused by the uncertainty of the adjustable load response. The quasi-line-type DR is a mechanism that takes the load quasi-line as the evaluation standard, which describes what kind of load is friendly to the system and uses it as a target to guide the adjustable load to participate in DR. The quasi-line-type DR overcomes many controversial issues such as small scale, difficult promotion, opacity and easy rebound of the price-type DR and the baseline-type DR, and has the characteristics of large-scale promotion and normal implementation. However, the formulation of the load quasi-line does not consider the system network constraints and the uncertainty of the adjustable load response, and it is difficult to guarantee the safety of the power grid during large-scale DR.

[0004] In summary, based on the existing DR method, large-scale DR will face the problem of power flow out-of-limit of the power grid caused by the uncertainty of the adjustable load response, and cannot guarantee the safe operation of the power grid.

[0005] Therefore, it is urgent to design a node load quasi-line method considering demand response uncertainty. SUMMARY

[0006] The purpose of the present application is to provide a node load quasi-line method considering demand response uncertainty to solve the problems raised in the background.

[0007] To achieve the above purpose, the present application provides the following technical scheme: a node load quasi-line method considering demand response uncertainty, comprising the following steps:

[0008] S1: Constructing the node load curve calculation model;

[0009] S2: Model transformation and solution;

[0010] S3: Based on the node load curve to carry out DR.

[0011] Further, in the above node load curve method considering demand response uncertainty, the specific steps of S1 are:

[0012] First, the uncertainty of large-scale adjustable load response is studied and modeled, and the uncertainty set of the response deviation of adjustable load compared with the curve is constructed, and then the objective function of the model and the related constraint conditions are defined;

[0013] a uncertainty set

[0014] Define the uncertainty deviation P of adjustable load relative to the node load curve Δ,k,t :

[0015] P DR,k,t =A k (P CDL,k,t +P Δ,k,t )

[0016] t=1,2,L,T; k=1,2,L,N D (1)

[0017] In the formula: P DR,k,t is the actual response value of the kth adjustable load at t period; A k is the sum of the kth adjustable load in T periods, calculated by formula (2); P CDL,k,t is the value of the kth node load curve at t period; P Δ,k,t is the deviation of the kth adjustable load at t period relative to the node load curve; T is the total period number; N D is the total number of adjustable loads;

[0018]

[0019] In the formula: is the day-ahead forecast value of the kth adjustable load at t period;

[0020] According to the response ability of the adjustable load, the deviation is constrained to determine the fluctuation range of the deviation. Specifically, it includes two levels of range constraints, local and overall;

[0021] Local level, the deviation of each period has upper and lower limits:

[0022]

[0023] t = 1, 2, L, T; k = 1, 2, L, N D (3)

[0024] In the formula: P Δ,k,t and respectively represent the lower limit and the upper limit of the response deviation of the kth adjustable load at the t period;

[0025] On the whole level, the baseline response of the adjustable load is defined, which is calculated based on the actual power consumption curve of the user after participating in the DR, and the baseline response is:

[0026]

[0027] In the formula: E k is the baseline response of the kth adjustable load, and ε is a given constant;

[0028] Before the DR is carried out, the actual baseline response of the adjustable load cannot be obtained, so it is stipulated that the baseline response of the adjustable load is above a certain threshold:

[0029]

[0030] In the formula: is the baseline response threshold of the kth adjustable load;

[0031] Therefore, the fluctuation range of the load deviation of the adjustable load constitutes the following uncertainty set:

[0032] X k = {P Δ,k | formula (3), formula (5)} (6)

[0033] In the formula: X k is the uncertainty set constituted by the response deviation of the kth adjustable load; P Δ,k is the time sequence vector of the response deviation of the kth adjustable load;

[0034] b Objective function

[0035] Through the DR, the adjustable load participates in the system regulation, realizes the purpose of promoting the economic operation of the system and improving the consumption of new energy; the closer the curve shape of the adjustable load is to the node load baseline, the lower the operation cost of the controllable unit of the system, and the higher the consumption of new energy; therefore, the curve shape of the adjustable load corresponding to the node load baseline should make the system operation cost and the energy rejection cost minimum; in addition, in order to ensure the consistency of the node load baseline and the uniqueness of the model solution under the condition that the line capacity is infinite, a regularization term represented by the difference degree of the node load baseline is introduced in the objective function; therefore, the objective function of the model is to minimize the sum of the controllable unit operation cost, the energy rejection cost of new energy and the regularization term, as shown in formula (7):

[0036]

[0037] N G is the number of controllable generator units; C i is the operating cost of the ith controllable generator unit; P G,i,t is the active power output of the ith controllable generator unit at time period t; N R is the number of new energy generator units; C j is the curtailment cost of the jth new energy generator unit; P R,j,t is the output of the jth new energy generator unit at time period t; λ is the regularization coefficient; F k is the difference degree of the kth node load curve; P CDL,k is the time series vector of the kth node load curve; decision variables include P G,i,t , P R,j,t , P CDL,k,t ;

[0038] C i , C j , F k , the expressions of which are respectively:

[0039] C i (P G,i,t ) = a i (P G,i,t ) 2 +b i P G,i,t +c i

[0040] t = 1, 2, L, T; i = 1, 2, L, N G (8)

[0041]

[0042]

[0043] a i , b i , c i are coefficients of the cost curve of the ith controllable generator unit; c R is the curtailment penalty coefficient, which should be large enough to ensure that new energy is preferentially consumed, and curtailment only occurs when the constraint condition cannot be met; is the maximum output of the jth new energy generator unit at time period t; is the average value of the node load curve, the expression of which is:

[0044]

[0045] It is important to note that P CDL,k,t ∈(0,1), therefore F in the L2 regularization term k (P CDL,k ) on the order of 10 -2 ~10 -4 Between, and in the objective function C i C j The order of magnitude of both costs is at least 10. 6 Therefore, as long as λ is not too large, the regularization term is very small compared to the cost term, and the optimization solution of the model always revolves around minimizing the operating cost of controllable units and the cost of renewable energy curtailment. At the same time, the value of λ cannot be too small, otherwise the regularization term will not work, thus failing to guarantee the universality of the model: when the line capacity is infinitely large, the obtained node load guideline is consistent; in the actual calculation guideline, the selection of λ coefficient in the range of 1 to 1000 is reasonable and will not affect the model solution results.

[0046] c Constraints

[0047] ①Direction constraint

[0048] For the load baseline of each node, the constraint of equation (12) should be satisfied to ensure that when users carry out DR based on the load baseline, only the electricity consumption is transferred in time, and the electricity consumption in the total time period does not increase or decrease.

[0049]

[0050] ② Power balance constraints

[0051]

[0052] Where: N C P represents the number of rigid, non-adjustable loads within the system. C,m,t This represents the value of the m-th rigid load during time period t.

[0053] ③ Line capacity constraints

[0054]

[0055] Where: π l,t Let be the power flowing through the l-th line during time period t; N represents the upper limit of transmission power for the l-th line; L Number of system lines;

[0056] In DC power flow mode, based on the power transmission distribution factor, the power flowing through the line can be obtained by using the injected and outflow power of each node;

[0057]

[0058] H l-i H l-j H l-k H l-m respectively represent the PTDF of controllable generator group i, new energy generator group j, adjustable load k, rigid load m to line l; it should be noted that, P Δ,k,t is a random variable, P Δ,k,t ∈X k .

[0059] Further, the specific steps of S2 in the above node load profile method considering demand response uncertainty are:

[0060] The above calculation model of node load profile is a robust optimization model composed of decision variables P G,i,t , P R,j,t , P CDL,k,t and random variable P Δ,k,t (P Δ,k,t ∈X k ); this model cannot be directly solved and needs to be converted into a deterministic equivalent model; considering the worst disturbance of the random variable, i.e. the maximum line power caused by the deviation of adjustable load relative to the node load profile, and then eliminating the random variable in the line capacity constraint; let ψ l,t be the line power generated by the decision variable and the rigid load, as shown in equation (16):

[0061]

[0062] Therefore, the line capacity constraint (14) can be converted into equation (17):

[0063]

[0064] Considering the directionality of line power flow, the minimum and maximum line power caused by the random variable satisfy the relationship:

[0065]

[0066] Substituting equation (18) into equation (17) obtains the equivalent inequality of line capacity constraint:

[0067]

[0068] Among them, the maximum value of the left end of the inequality constraint can be obtained by solving the following optimization problem:

[0069]

[0070] Therefore, the robust model of the node load curve can be converted into a determination equivalent model, which is a quadratic programming problem.

[0071] Further, the specific steps of S3 in the node load curve method considering the demand response uncertainty are:

[0072] a) the power department issues a unified load curve to the whole system, and the unified load curve can be obtained by setting the upper limit of the line transmission power in the model of the present scheme to infinity;

[0073] b) the load aggregator of each node area evaluates the adjustable load response capability of the area according to the unified load curve, including the local deviation and the overall deviation of the adjustable load tracking curve, and then reports the adjustable load response range to the power department;

[0074] c) the power department calculates the node load curve according to the adjustable load response range data reported by the load aggregator and the relevant parameters of the whole network, and issues it to each node area;

[0075] d) the user participating in the DR adjusts the power consumption mode under the guidance of the load curve of the node where the user is located, so that the degree of approximation of the load curve to the node load curve is within the response range reported by the load aggregator; after that, the DR implementer compensates the user according to the similarity degree of the actual power consumption curve of the user to the node load curve and the response range reported in advance.

[0076] Compared with the prior art, the present application has the following advantages:

[0077] The present application proposes a node load curve method considering the uncertainty of demand response, which issues a load curve to each adjustable load of each node as a guiding target for the participation of the DR of each node load; considering the uncertainty of the adjustable load response, a robust and safe node load curve calculation model is established; based on the node load curve calculated by the proposed model, large-scale DR is carried out, and when the adjustable load response DR is within the range reported in advance, the power grid does not have line power overrun, which guarantees the security of the power flow of the power grid; that is, the method can guarantee that the line power flow of the system meets the capacity constraint for any actual response result of the adjustable load within the agreed DR response range; it can be used for carrying out large-scale DR, and the economy and safety of the system are taken into account. BRIEF DESCRIPTION OF DRAWINGS

[0078] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed for the description of the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort on the basis of these drawings.

[0079] Figure 1 A schematic diagram of the node load curve system of the present application;

[0080] Figure 2 A flowchart of the demand response based on the node load curve of the present application;

[0081] Figure 3 A 9-node system topology diagram of the present application;

[0082] Figure 4 A schematic diagram of the node load curve of the present application;

[0083] Figure 5 A power schematic diagram of the line 5-6 for developing DR of the present application; DETAILED DESCRIPTION

[0084] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings of the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the present application.

[0085] The present application provides a technical solution: a node load curve method considering the uncertainty of demand response, comprising the following steps:

[0086] S1: constructing a node load curve calculation model

[0087] First, the uncertainty of large-scale adjustable load response is studied and modeled, and the uncertainty set of the response deviation of the adjustable load compared with the curve is constructed, and then the objective function of the model and the related constraint conditions are defined;

[0088] a uncertainty set

[0089] Define the uncertainty deviation P of the adjustable load relative to the node load curve Δ,k,t :

[0090] P DR,k,t =A k (P CDL,k,t +P Δ,k,t )

[0091] t=1,2,L,T;k=1,2,L,ND (1)

[0092] P DR,k,t is the actual response value of the kth adjustable load at time period t; A k is the total sum of the kth adjustable load in T time periods, calculated by equation (2); P CDL,k,t is the value of the kth node load curve at time period t; P Δ,k,t is the deviation of the kth adjustable load at time period t relative to the node load curve; T is the total number of time periods; N D is the total number of adjustable loads;

[0093]

[0094] wherein: is the day-ahead forecast value of the kth adjustable load at time period t;

[0095] The deviation is constrained according to the response capability of the adjustable load to determine the fluctuation range of the deviation. Specifically, the range constraint includes two levels of local and overall;

[0096] At the local level, the deviation of each time period has an upper and lower limit:

[0097]

[0098] wherein: P Δ,k,t and respectively represent the lower and upper limits of the response deviation of the kth adjustable load at time period t;

[0099] At the overall level, the line response degree of the adjustable load is defined, which is calculated based on the actual power consumption curve of the user after participating in the DR, and the line response degree is:

[0100]

[0101] wherein: E k is the line response degree of the kth adjustable load, and ε is a given constant;

[0102] Before the DR is carried out, the actual line response degree of the adjustable load cannot be obtained, so the line response degree of the adjustable load is regulated to be above a certain threshold:

[0103]

[0104] wherein: is the line response degree threshold of the kth adjustable load;

[0105] Thus, the fluctuation range of the adjustable load deviation constitutes the following uncertain set:

[0106] Xk = {P Δ,k | formula (3), formula (5)} (6)

[0107] In the formula, X k is an uncertain set composed of the kth adjustable load response deviation; P Δ,k is a time series vector of the kth adjustable load response deviation;

[0108] b objective function

[0109] By carrying out DR, the adjustable load participates in system regulation, so as to achieve the purposes of promoting system economic operation and improving new energy consumption; the closer the curve shape of the adjustable load is to the node load baseline, the lower the operation cost of the controllable unit of the system is, and the higher the consumption of new energy is; therefore, the curve shape of the adjustable load corresponding to the node load baseline should make the system operation cost and the abandoned energy cost minimum; in addition, in order to ensure the consistency of the node load baseline and the uniqueness of the model solution under the condition that the line capacity is infinite, a regularization term represented by the difference degree of the node load baseline is introduced in the objective function; therefore, the objective function of the model is to minimize the sum of the controllable unit operation cost, the abandoned energy cost of the new energy and the regularization term, as shown in formula (7):

[0110]

[0111] In the formula, N G is the number of controllable generators; C i is the operation cost of the ith controllable generator; P G,i,t is the active power output of the ith controllable generator at the t period; N R is the number of new energy generators; C j is the abandoned energy cost of the jth new energy generator; P R,j,t is the output of the jth new energy generator at the t period; λ is a regularization coefficient; F k is the difference degree of the kth node load baseline; P CDL,k is a time series vector of the kth node load baseline; the decision variables include P G,i,t , P R,j,t , P CDL,k,t ;

[0112] C i , C j , F k are respectively expressed as:

[0113] C i (P G,i,t ) = a i (P G,i,t ) 2 +b i PG,i,t +c i

[0114] t = 1, 2, L, T; i = 1, 2, L, N G (8)

[0115]

[0116]

[0117] in which: a i ,b i ,c i is the coefficient of the cost curve of the ith controllable generator set; c R is the penalty coefficient for energy curtailment, which should be large enough to ensure that new energy is preferentially consumed, and energy curtailment only occurs when the constraint condition cannot be met; is the maximum output of the jth new energy generator set at the tth time period; is the average value of the node load profile, and the expression is:

[0118]

[0119] It should be noted that P CDL,k,t ∈(0, 1), so F k (P CDL,k ) in the L2 regularization term is of the order of 10 -2 ~ 10 -4 , while the magnitudes of the two cost terms C i and C j in the objective function are at least 10 6 ; therefore, as long as λ is not taken to be particularly large, the regularization term is very small compared to the cost term, and the optimization of the model is always focused on minimizing the controllable generator set operation cost and the new energy curtailment cost; at the same time, λ cannot be taken to be too small, otherwise the regularization term will not work, and thus the universality of the model cannot be guaranteed: when the line capacity is large, the node load profile obtained is consistent; in actual calculation of the profile, it is reasonable to select the λ coefficient in the range of 1 ~ 1000, and it will not affect the solution of the model;

[0120] c constraint condition

[0121] ① Profile constraint

[0122] For the load profile of each node, the constraint of formula (12) should be met to ensure that when users carry out DR based on the load profile, only the electricity consumption in time is shifted, and the electricity consumption in the total time period is not increased or decreased;

[0123]

[0124] ②Power balance constraints

[0125]

[0126] where N C is the number of rigid non-controllable loads in the system; P C,m,t is the value of the mth rigid load at time period t;

[0127] ③Line capacity constraints

[0128]

[0129] where π l,t is the power flowing through the lth line at time period t; P is the upper limit of transmission power of the lth line; N L is the number of lines in the system;

[0130] In the DC power flow mode, the power flowing through the line can be obtained based on the power transmission distribution factor (PTDF) using the injection and outflow power of each node.

[0131]

[0132] where H l-i , H l-j , H l-k , H l-m respectively represent the PTDF of the controllable generator group i, the new energy generator group j, the adjustable load k, and the rigid load m to the line l; it should be noted that P Δ,k,t is a random variable, P Δ,k,t ∈ X k .

[0133] S2: Model transformation and solution;

[0134] The above calculation model of the node load profile is a robust optimization model composed of decision variables P G,i,t , P R,j,t , P CDL,k,t and random variables P Δ,k,t (P Δ,k,t ∈ X k ); this model cannot be directly solved and needs to be transformed into a deterministic equivalent model; considering the worst disturbance of the random variable, i.e., the maximum line power caused by the deviation of the adjustable load relative to the node load profile, the random variable in the line capacity constraint is eliminated; let ψ l,t be the line power generated by the decision variable and the rigid load, as shown in equation (16):

[0135]

[0136] The line capacity constraint (14) can be transformed into equation (17):

[0137]

[0138] Considering the directionality of line power flow, the minimum and maximum values of line power caused by random variables satisfy the relationship:

[0139]

[0140] Substituting equation (18) into equation (17) gives the equivalent inequality of the line capacity constraint:

[0141]

[0142] The maximum value of the left end of the inequality constraint can be obtained by solving the following optimization problem:

[0143]

[0144] Therefore, the robust model about the node load profile can be transformed into a deterministic equivalent model, which is a quadratic programming problem. After substituting the results of optimization problem (20) into equation (19), it can be solved by commercial software such as Gurobi, Cplex, etc.

[0145] S3: DR based on node load profile.

[0146] a) The power department issues a unified load profile to the whole system, which can be obtained by setting the upper limit of line transmission power in the model of this scheme to infinity;

[0147] b) The load aggregators in each node area evaluate the adjustable load response capability of the local area according to the unified load profile, including the local deviation and overall deviation of the adjustable load tracking profile, and then report the response range of the adjustable load to the power department;

[0148] c) The power department calculates the node load profile according to the adjustable load response range data reported by the load aggregators and the relevant parameters of the whole network, and issues it to each node area;

[0149] d) The users participating in DR adjust their power consumption mode under the guidance of the load profile of their own node, so that the degree of approximation of the load curve to the node load profile is within the response range reported by the load aggregator; After that, the DR implementer incentivizes and compensates the users according to the similarity between the actual power consumption curve and the node load profile and the response range reported in advance.

[0150] Take a specific system as an example to illustrate:

[0151] The system topology is as follows Figure 3As shown, the DR center needs to calculate the node load curve according to the controllable generator set parameters, new energy data, non-DR load data, total amount of DR participating load, system network parameters and DR load response range reported by the load aggregator, and the result is as shown in FIG. 4. Figure 4 As shown. After receiving the load curve, the DR load of each node adjusts its power consumption curve to follow the node load curve and ensures that the actual value of the response is within the response range reported by the load aggregator. Taking a line in the system as an example for analysis, Figure 5 The transmission power of line 5-6 during the DR process is given, and the transmission power is the average value of 100 scenarios that meet the DR load response range. Under the four adjustable load response degrees, the line power curve in the whole period is below the upper limit of the transmission power, and there is no power overrun in the system. It shows that the node load curve of the method has robust safety, and the degree of DR load response curve fluctuates within a certain range under the guidance of the corresponding node load curve, and there is no line power overrun, which guarantees the power flow safety of the system.

[0152] In the description of the present specification, the description of the terms "one embodiment", "example", "specific example" and the like means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0153] The preferred embodiments of the application disclosed above are only used to help explain the application. The preferred embodiments do not describe all the details and limit the application to the specific embodiments described. Obviously, many modifications and changes can be made according to the content of the present specification. The present specification selects and describes these embodiments in order to better explain the principles and practical applications of the application, so that those skilled in the art can well understand and utilize the application. The application is limited only by the claims and their entire scope and equivalents.

Claims

1. A method for node load curtailment considering demand response uncertainty, characterized in that, The method comprises the following steps: S1: constructing a node load baseline calculation model; S2: model transformation and solution; S3: carrying out DR based on the node load baseline; The specific steps of S1 are as follows: First, the uncertainty of large-scale adjustable load response is researched and modeled, the response deviation uncertainty set of the adjustable load compared with the baseline is constructed, and then the objective function of the model is defined; a) Uncertainty set Definition of the uncertainty deviation P of the adjustable load with respect to the node load reference line Δ,k,t : P DR,k,t = A k (P CDL,k,t + P Δ,k,t ) t = 1, 2,..., T; k = 1, 2,..., N D (1) where: P DR,k,t is the actual response value of the kth adjustable load at time period t; A k is the total of the kth adjustable load over T time periods, calculated from equation (2); P CDL,k,t is the value of the kth node load profile at time period t; P Δ,k,t is the deviation of the kth adjustable load at time period t from the node load profile; T is the total number of time periods; N D is the total number of adjustable loads; In the formula: is the day-ahead forecast value of the kth adjustable load at the t period; According to the response ability of the adjustable load, the deviation is constrained to determine the fluctuation range of the deviation; specifically, it includes two levels of range constraints, local and overall; On the local level, the upper and lower limits of the deviation of each time period exist: t = 1, 2,..., T; k = 1, 2,..., N D (3) wherein: P Δ,k,t and respectively represent the lower and upper limits of the response deviation of the kth adjustable load at the t period. On the overall level, the baseline response degree of the adjustable load is defined, which is calculated based on the actual power consumption curve of the user after participating in the DR, and the baseline response degree is: where: E k is the slope of the kth adjustable load, and ε is a given constant. Before carrying out the DR, the actual baseline response degree of the adjustable load cannot be obtained, so it is stipulated that the baseline response degree of the adjustable load is above the threshold: wherein: is the threshold of the quasi-linear response of the kth adjustable load; Therefore, the fluctuation range of the deviation of the adjustable load constitutes the following uncertainty set: X k = {P Δ,k | Formula (3), Formula (5)} (6) wherein: X k is an uncertain set of kth adjustable load response biases; P Δ,k is a time series vector of kth adjustable load response biases; b) Objective function Through the DR, the adjustable load participates in system regulation; in order to ensure the consistency of the node load baseline and the uniqueness of the model solution when the line capacity is infinite, a regularization term represented by the node load baseline difference degree is introduced into the objective function; therefore, the objective function of the model is to minimize the sum of the controllable unit operation cost, the energy abandonment cost of new energy and the regularization term, as shown in formula (7): In the formula: N G is the number of controllable generator sets; C i is the operating cost of the ith controllable generator set; P G,i,t is the active power output of the ith controllable generator set at the t time period; N R is the number of new energy generator sets; C j is the energy abandonment cost of the jth new energy generator set; P R,j,t is the output of the jth new energy generator set at the t time period; λ is a regularization coefficient; F k is the difference degree of the kth node load curve; P CDL,k is the time series vector of the kth node load curve; decision variables include P G,i,t , P R,j,t , P CDL,k,t ; C i , C j , F k The expressions of the above are respectively: C i (P G,i,t )=a i (P G,i,t ) 2 +b i P G,i,t +c i t = 1, 2,..., T; i = 1, 2,..., N G (8) t = 1,2,...,T; j = 1,2,...,N R (9) k = 1, 2,..., N D (10) wherein: a i b i c i is the coefficient of the cost curve of the ith controllable generator set; c R is the penalty coefficient of abandoned energy; is the maximum output of the jth new energy generator set at the t period; is the average value of the node load curve, and the expression is: P CDL,k,t ∈(0,1), F k in L2 regularization term CDL,k ) is of the order of 10 -2 ~ 10 -4 , C i , C j two cost terms in the objective function are at least of the order of 10 6 ; when the line capacity is large, the obtained node load profile is consistent; in the actual calculation of the profile, the λ coefficient is selected in the range of 1-1000. 2.The method of claim 1, wherein, The above S1 also defines the constraint conditions of the model: c) Constraint conditions ①Baseline constraint For the load baseline of each node, it should satisfy the constraint of formula (12) to ensure that when the user carries out the DR based on the load baseline, only the electricity consumption in time is transferred, and the electricity consumption in the total period is not increased or decreased; ②Power balance constraint t=1,2,…,T (13) where: N C is the number of rigid non-adjustable loads within the system; P C,m,t is the value of the mth rigid load at time period t; ③Line capacity constraint t = 1, 2,..., T; l = 1, 2,..., N L (14) where: π l,t Pil(t) is the power flowing through the ilth line at time t; Pil is the upper limit of transmission power for the ilth line; N L is the number of system lines; In the direct current flow mode, based on the power transmission distribution factor, the power flowing through the line can be obtained by using the injection and outflow power of each node; t = 1,2,...,T; l = 1,2,...,N L (15) where H l-i , H l-j , H l-k , H l-m denote the PTDF of controllable generator group i, new energy generator group j, adjustable load k, rigid load m to line l, respectively; P Δ,k,t is a random variable, P Δ,k,t ∈X k . 3.The method of claim 1, wherein, The specific steps of S2 are as follows: The calculation model of the node load curve is composed of decision variables P G,i,t , P R,j,t , P CDL,k,t and random variables P Δ,k,t (robust optimization model; the model cannot be directly solved and needs to be converted into a deterministic equivalent model; the worst disturbance of the random variable is considered, i.e., the maximum line power caused by the deviation of the adjustable load relative to the node load curve, and then the random variable in the line capacity constraint is eliminated; let ψ l,t be the line power generated by the decision variable and the rigid load, as shown in equation (16): t = 1,2,...,T; l = 1,2,...,N L (16) Therefore, the line capacity constraint (14) can be converted into formula (17): t = 1,2,...,T; l = 1,2,...,N L (17) Considering the directionality of the line flow, the minimum and maximum values of the line power caused by the random variable satisfy the relationship: Substituting formula (18) into formula (17) obtains the equivalent inequality of the line capacity constraint: t = 1,2,...,T; l = 1,2,...,N L (19) Wherein, the maximum value of the left end of the inequality constraint can be obtained by solving the following optimization problem: Therefore, the robust model about the node load baseline can be transformed into a deterministic equivalent model, which is a quadratic programming problem. After substituting the result obtained by solving the optimization problem (20) into formula (19), it can be solved by commercial software Gurobi, Cplex. 4.The method of claim 1, wherein, The specific steps of S3 are as follows: a) The power department issues a unified load baseline to the whole system, and the unified load baseline can be obtained by setting the upper limit of the line transmission power in the model to infinity; b) The load aggregator of each node area evaluates the response ability of the adjustable load in the area according to the unified load baseline, including the local deviation and overall deviation of the adjustable load tracking the baseline, and then reports the response range of the adjustable load to the power department; c) The power department calculates the node load curve according to the adjustable load response range data reported by the load aggregator and the relevant parameters of the whole network, and publishes it to each node area; d) The users participating in the DR adjust their power consumption mode under the guidance of the load curve of their own node, so that the degree of approximation of the load curve to the node load curve is within the response range reported by the load aggregator. After that, the DR implementer compensates the users according to the similarity between the actual power consumption curve and the node load curve and the response range reported in advance.

Images