Multi-price demand response dispatch system for peak shifting based on gradient algorithm

Abstract

The embodiment of the present invention discloses a multi-price demand response pricing system for peak shifting based on a gradient algorithm, the system comprising: an electricity price-load optimization module, used to determine the optimal equipment load and the optimal electricity cost based on user equipment information and a group of electricity prices determined by a multi-pricing formulation module; a power grid dispatching module, used to determine the optimal power generation cost based on the total load of the power grid and the basic information of the power grid; a multi-pricing formulation module, used to calculate a group of electricity prices based on the optimal equipment load and the optimal electricity cost, and the optimal power generation cost, a partial derivative link, used to establish the optimal equipment load and the optimal electricity cost, the optimal power generation cost, the partial derivative of the group of electricity prices, and solve the final optimal electricity price combination through a gradient algorithm. The present invention establishes the relationship between the optimal value and the input variable, and calculates its partial derivative, and finally quickly solves the final optimal electricity price combination through the gradient method, which can effectively achieve peak shaving and valley filling and improve the level of social energy efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to a multi-electricity price demand response dispatching system oriented to peak shifting phenomenon based on a gradient algorithm. Background Art

[0002] Smart grid is the future development direction of power system construction in various countries, and demand response is one of the main topics of smart grid development, which plays a role in the fields of new energy power generation, resource allocation optimization, scientific load management, etc., so that demand-side management can play a greater role. In China, demand-side management is one of the important development directions of the energy and power industry. The essence of demand-side response is to achieve peak shaving and valley filling, improve the safety and stability of power system operation, and improve the overall operation efficiency of society through electricity prices or incentives for users to change their own electricity consumption methods and actively participate in the energy interaction of power grid operation.

[0003] Demand response (DR) is the abbreviation of power demand response. It refers to the short-term behavior of power users changing their inherent habitual power consumption mode after receiving the direct compensation notice of induced load reduction or power price increase signal issued by the power supplier when the wholesale power market price rises or the system reliability is threatened, so as to reduce or shift the power load in a certain period of time to respond to power supply, thereby ensuring the stability of the power grid and suppressing the rise of power prices. It is one of the solutions for demand side management (DSM). Demand response can be divided into two categories: the first category is time-based demand response; the second category is incentive-based demand response. Incentive-based demand response works according to time, and through advanced contract rules, the consumption behavior of customers will change temporarily. In response to load changes, power companies give corresponding compensation or penalties. Time-based demand response is also called price-based demand response, which means that users adjust their power demand responsively according to the received price signal, including time-of-use electricity price response, real-time electricity price response, Jianfeng electricity price response, etc.

[0004] At present, most cities in China adopt peak-valley electricity prices, and large industrial electricity users adopt peak-valley electricity prices. The peak hours are 8:00-12:00 and 17:00-21:00; the flat hours are 12:00-17:00 and 21:00-24:00; the valley hours are 0:00-8:00. Figure 1As shown. For cities with a large proportion of flexible large users (referring to users with large electricity consumption and whose electricity load is flexibly shifted with the low electricity price period, generally large industrial users), a "peak shifting phenomenon" will occur. Flexible large users will avoid starting production equipment during peak electricity consumption, and thus give priority to starting equipment during the low electricity price period. However, the length of the low electricity period often cannot meet production needs, so the equipment will continue to be started during the flat electricity price period. When most of the large user loads in the city are transferred to the low electricity price period, the peak shifting phenomenon will occur. That is, the peak electricity consumption always moves with the low electricity price period, weakening the peak shaving and valley filling effect. Summary of the invention

[0005] In order to solve the problem that under the traditional time-of-use electricity price model, the time-of-use electricity price for all users in the region is the same, resulting in the weakening of the "peak shaving and valley filling" effect, a multi-electricity price demand response scheduling mechanism based on a gradient algorithm is provided for the peak shifting phenomenon in large industrial cities.

[0006] The embodiment of the present invention discloses a multi-price demand response scheduling system for peak shifting phenomenon based on a gradient algorithm, including:

[0007] The electricity price-load optimization module is used to determine the optimal equipment load and the optimal electricity cost based on the user equipment information of the auxiliary user group and the optimal multi-price curve group determined by the multi-price setting module;

[0008] The grid dispatching module is used to determine the optimal power generation cost based on the total grid load and basic grid information;

[0009] A multi-electricity price formulation module, used to calculate an optimal multi-electricity price curve group based on the optimal equipment load, the optimal electricity cost, and the optimal power generation cost;

[0010] The partial derivative link is used to establish the partial derivative of the optimal equipment load and the optimal electricity cost, the optimal power generation cost, and the optimal multi-price curve group, and solve the final optimal electricity price combination through the gradient algorithm based on the partial derivative.

[0011] As an optional implementation, the multi-electricity price formulation module includes a first objective function unit, a first indicator constraint unit and a first algorithm optimization unit; the electricity price-load optimization module includes a second objective function unit, a second indicator constraint unit and a second algorithm optimization unit; the power grid dispatching module includes a third objective function unit, a third indicator constraint unit and a third algorithm optimization unit; wherein:

[0012] The second objective function unit is used to construct a second objective function model based on the optimal multi-electricity price curve group output by the multi-electricity price formulation module; the second indicator constraint unit is used to construct a second indicator constraint model according to the user equipment information; the second algorithm optimization unit is used to calculate the optimal equipment load and the optimal electricity cost according to the second objective function model and the second indicator constraint model;

[0013] The third objective function unit is used to construct a third objective function model; the third indicator constraint unit is used to construct a third indicator constraint model according to the total load of the power grid and the basic information of the power grid; the third algorithm optimization unit is used to calculate the optimal power generation cost according to the third objective function model and the third indicator constraint model;

[0014] The first objective function unit is used to construct a first objective function model based on the optimal equipment load, the optimal electricity cost, and the optimal power generation cost; the first indicator constraint unit is used to construct a first indicator constraint model according to the optimal equipment load and the optimal electricity cost; the first algorithm optimization unit is used to calculate multiple electricity price groups according to the first objective function model and the first indicator constraint model;

[0015] The partial derivative link is used to establish the partial derivative of the first objective function model for the optimal power generation cost; establish the partial derivative of the optimal power generation cost for the optimal startup state input to the power grid dispatching module; establish the partial derivative of the optimal startup state for the multiple electricity price groups input to the electricity price-load optimization module; establish the partial derivative of the first objective function model for the optimal electricity consumption cost; establish the partial derivative of the optimal electricity consumption cost for the multiple electricity price groups input to the electricity price-load optimization module; establish the partial derivative of the first objective function model for the multiple electricity price groups input to the electricity price-load optimization module; finally, based on the partial derivative of the first objective function model for the multiple electricity price groups, the final optimal electricity price combination is solved by the gradient method.

[0016] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0017] The present invention constructs an electricity price-load optimization module, a power grid dispatching module and a multi-price setting module for the auxiliary user group. The three form a complex nested optimization problem, which can be solved by a genetic algorithm, but the solution speed is relatively slow. This patent provides a solution method for the nested optimization problem. Through model equivalence, each sub-optimization problem in the nested optimization problem is converted into a mathematical problem of direct calculation by establishing a mathematical optimization model. The relationship between the optimal value and the input variable can be established, and its partial derivative can be calculated, and finally quickly solved by the gradient method. The electricity price can eventually effectively achieve peak shaving and valley filling, improve the level of social energy efficiency, and bring more profits to the auxiliary user group and the power grid company, achieving a win-win situation. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 It is a structural principle diagram of a multi-electricity price demand response scheduling system for peak shifting phenomenon based on a gradient algorithm disclosed in an embodiment of the present invention;

[0020] Figure 2 It is a structural principle diagram of a multi-electricity price setting module disclosed in an embodiment of the present invention;

[0021] Figure 3 It is a structural principle diagram of the electricity price-load optimization module disclosed in an embodiment of the present invention;

[0022] Figure 4 is a structural principle diagram of a power grid dispatching module disclosed in an embodiment of the present invention;

[0023] Figure 5 It is a flow chart of the partial guidance link disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0024] 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.

[0025] It should be noted that the terms "first", "second", "third", "fourth", etc. in the specification and claims of the present invention are used to distinguish different objects rather than to describe a specific order. The terms "including" and "having" in the embodiments of the present invention and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device including a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] Example

[0027] See also Figure 1 , Figure 1 1 is a schematic diagram of the structure of a multi-price demand response dispatch system for peak shifting based on a gradient algorithm disclosed in an embodiment of the present invention. Figure 1 As shown, the multi-price demand response dispatching system for peak shifting based on the gradient algorithm includes: a multi-price setting module 1, a price-load optimization module 2, a power grid dispatching module 3, and an auxiliary user group 4; wherein: the multi-price setting module 1 is connected to the price-load optimization module 2; the multi-price setting module 1 is connected to the power grid dispatching module 3; the multi-price setting module 1 is connected to the auxiliary user group 4; the price-load optimization module 2 is connected to the auxiliary user group 4. The auxiliary user group specifically refers to a high-component flexible user group that participates in the mechanism proposed in this patent after screening and completing the benefit distribution negotiation with the power grid. Their characteristics include huge electricity consumption, and a large amount of load can be flexibly transferred with the low electricity price.

[0028] The main functions of each part are: the electricity price-load optimization module 2 is used to determine the optimal equipment load and the optimal electricity cost based on the user equipment information of the auxiliary user group and a set of electricity prices determined by the multi-electricity price formulation module. The power grid dispatching module 3 is used to determine the optimal power generation cost based on the total load of the power grid and the basic information of the power grid; the multi-electricity price formulation module 1 is used to calculate a set of electricity prices (i.e., a multi-electricity price group, also called the initial optimal multi-electricity price curve group) based on the optimal equipment load, the optimal electricity cost, and the optimal power generation cost using a genetic algorithm, and then establish the relationship between the optimal value and the input variable through a partial derivative link based on the group of electricity prices and the optimal equipment load, the optimal electricity cost, and the optimal power generation cost, and calculate its partial derivative, and finally quickly solve it through the gradient method to obtain the final optimal electricity price combination, and send the final optimal electricity price combination to the auxiliary user group.

[0029] like Figure 2 As shown, the multi-electricity price formulation module 1 includes a first objective function unit 11, a first indicator constraint unit 12 and a first optimization algorithm unit 13. Among them: one input end of the first objective function unit 11 is connected to the output end of the electricity price-load optimization module 2, and the other input end of the first objective function unit 11 is connected to the output end of the power grid dispatching module 3, which is used to construct a first objective function model based on the optimal equipment load and the optimal electricity cost, and the optimal power generation cost. The input end of the first indicator constraint unit 12 is connected to the output end of the electricity price-load optimization module 2, which is used to construct a first indicator constraint model based on the optimal equipment load and the optimal electricity cost. The two input ends of the first optimization algorithm unit 13 are respectively connected to the output end of the first objective function unit 11 and the output end of the first indicator constraint unit 12, which is used to calculate a set of electricity prices based on the first objective function model and the first indicator constraint model using a genetic algorithm.

[0030] like Figure 3As shown, the electricity price-load optimization module 2 includes a second objective function unit 21, a second indicator constraint unit 22 and a second algorithm optimization unit 23. Wherein: the input end of the second objective function unit 21 is connected to the output end of the multi-electricity price formulation module 1, and is used to construct a second objective function model based on the optimal multi-electricity price curve group output by the multi-electricity price formulation module. The second indicator constraint unit is used to construct a second indicator constraint model according to the user equipment information. The two input ends of the second algorithm optimization unit 23 are respectively connected to the output end of the second objective function unit 21 and the output end of the second indicator constraint unit 22, and are used to calculate the optimal equipment load and the optimal electricity cost according to the second objective function model and the second indicator constraint model.

[0031] like Figure 4 As shown, the power grid dispatching module 3 includes a third objective function unit 31, a third indicator constraint unit 32 and a third algorithm optimization unit 33. Among them: the third objective function unit is used to construct a third objective function model. The input end of the third indicator constraint unit 32 is connected to the output end of the electricity price-load optimization module 2, and is used to construct a third indicator constraint model according to the total load of the power grid and the basic information of the power grid; the two input ends of the third algorithm optimization unit 33 are respectively connected to the output end of the third objective function unit 31 and the output end of the third indicator constraint unit 32, and are used to calculate the optimal power generation cost according to the third objective function model and the third indicator constraint model.

[0032] In the above scheme, for the electricity price-load optimization module 2, it receives the electricity price group output by the multi-electricity price setting module, establishes a second objective function unit 21 and a second indicator constraint unit 22 to establish a model according to the equipment parameters, and calculates and outputs the optimal compliance behavior and optimal electricity cost of each auxiliary user group 4 through the second algorithm optimization unit 23.

[0033] In the above scheme, the power grid dispatching module 3 receives the optimal load behavior output by the electricity price-load optimization module 2 and calculates the total load of the entire network. According to the grid information and unit information, the third objective function unit 31 and the third indicator constraint unit 32 are established and the optimal power generation cost is calculated through the third algorithm optimization unit 33.

[0034] In the above scheme, for the multi-electricity price formulation module 1, the optimal electricity cost output by the electricity price-load optimization module 2 and the optimal power generation cost output by the power grid dispatching module 3 are received. A first objective function unit 11 and a first indicator constraint unit 12 are established, and the optimal electricity price combination is calculated through the first optimization algorithm unit 13.

[0035] In the above scheme, high-component flexible users who meet the requirements of participating in the electricity price mechanism proposed by this patent are first screened out as auxiliary users. The auxiliary user group mainly plays the role of peak load regulation. For a multi-price demand response scheduling mechanism for the peak shift phenomenon in large industrial cities: First, the multi-price formulation module outputs a set of electricity prices, which is input into the second objective function unit 21. The second indicator constraint unit 22 is responsible for collecting user equipment information, such as equipment power, total equipment startup time, whether the equipment is transferable, and other information. And the optimal load behavior and the optimal electricity cost are obtained through the second algorithm optimization unit 23. Among them, the optimal electricity consumption behavior is input into the third indicator constraint unit 32. The third indicator constraint unit 32 collects grid information and unit information at the same time, and combines the third objective function unit 31 to calculate the optimal unit power generation power combination and power generation cost through the third algorithm optimization unit 33. For the multi-electricity price setting module, it receives the user electricity cost output by the electricity price-load optimization module 2 and the power generation cost output by the power grid dispatching module, and outputs another set of electricity prices to the electricity price-load optimization module 2 through the genetic algorithm. This is repeated until a set of electricity prices is obtained. After the auxiliary users (users in the auxiliary user group, hereinafter referred to as auxiliary users or users) respond, their electricity costs can be reduced to the agreed value. At the same time, peak shaving and valley filling can be achieved to the greatest extent, so that the power generation cost of the power grid is minimized.

[0036] In the above scheme, the partial derivative link establishes the partial derivative of the first objective function model to the optimal power generation cost; establishes the partial derivative of the optimal power generation cost to the optimal startup state input to the power grid dispatching module; establishes the partial derivative of the optimal startup state to the multiple electricity price groups input to the electricity price-load optimization module; establishes the partial derivative of the first objective function model to the optimal electricity consumption cost; establishes the partial derivative of the optimal electricity consumption cost to the multiple electricity price groups input to the electricity price-load optimization module; establishes the partial derivative of the first objective function model to the multiple electricity price groups input to the electricity price-load optimization module; finally, according to the partial derivative of the first objective function model to the multiple electricity price groups, the final optimal electricity price combination is solved by the gradient method.

[0037] Specifically, the second objective function unit is used to construct a second objective function model based on the optimal equipment load, the optimal electricity cost, and the optimal power generation cost, including:

[0038] Construct the second objective function model:

[0039]

[0040] in:

[0041]

[0042] Where: Bill and Obj 1(MS) is the total electricity cost of all auxiliary user groups. Since each user takes the minimum electricity cost as the optimization goal of load behavior, the minimum electricity cost of each user is equivalent to the objective function of minimizing the electricity cost of all users. NU is the number of users in the auxiliary user group; NT is the minimum number of optimized time periods. If a day is divided into 24 hours, then NT = 24; NDev is the number of transferable devices of all users in the auxiliary user group; e x =[1,…,1] represents a row vector whose row x column is all 1. It can be understood that e NU is a row vector with 1 row and NU columns all set to 1, is a column vector with all 1s in NT rows and columns; P is the price curve group of the auxiliary user group, also called the price group or price. MS is the startup state of the transferable device. DP represents the device power, dp i represents the power of the i-th device. UL is the total load of non-transferable devices. The i-th row and j-th column in SCA is 1, indicating that the j-th device in the MS belongs to the i-th user, otherwise it is 0; p k,j is the electricity price of the kth user in the jth period; ms i,j Indicates the startup status of the i-th device in the j-th period. If it is 1, the device is started, otherwise, the device is shut down; dp i Indicates the power of the i-th device; ul k,j It represents the total load of the non-transferable equipment of the k-th user in the j-th period.

[0043] The second indicator constraint unit is used to construct a second indicator constraint model according to the user equipment information, including:

[0044] The total startup time is used as the second constraint function of the second indicator constraint model.

[0045] The daily power consumption of each device (the total daily power consumption time of each device) must be equal to the time required to meet the user's certain production task:

[0046]

[0047] Where: ms i,j is the power consumption time of the i-th device in the j-th period, wt i is the total time required for the i-th device to start up;

[0048] Use the startup state as the second constraint function of the second indicator constraint model:

[0049] ms i,j =1or0(4)

[0050] In the formula: means that the startup status of the i-th device in the j-th period can only be startup or shutdown;

[0051] Use the non-transferable constraint as the second and third constraint functions of the second indicator constraint model:

[0052] Assuming that the i-th device needs to be started all the time or started at a specific time, the i-th device is non-transferable; the load of the i-th device is taken as a constant, that is, UL in formula (1), and does not participate in the optimization process.

[0053] The third objective function unit is used to construct a third objective function model, including:

[0054] The third objective function model is constructed based on the minimum total power generation cost:

[0055]

[0056] in:

[0057] In the formula: Cost is the power generation cost; NG is the number of generator sets; a i , b i 、c i is the coal consumption curve parameter of the i-th unit; g i,j Obj represents the power generation of the i-th unit in the j-th period, 2 (G) is the minimum total power generation cost;

[0058] The third indicator constraint unit is used to construct a third indicator constraint model according to the total load of the power grid and the basic information of the power grid, including:

[0059] The power balance constraint is used as the third constraint function of the third indicator constraint model, as shown in formula (6):

[0060]

[0061] in:

[0062]

[0063] Where: ND is the total load of the whole network; D is the total load of the whole network, d i,j represents the power consumption of the i-th load in the j-th period; BL is the total load of other users in the load D minus the auxiliary user group, i.e., the base load; bl i,j represents the base load size of the i-th total load in the j-th period; ML is the total load of the transferable equipment of each auxiliary user, ml i,j represents the total load of the transferable equipment of the i-th auxiliary user in the j-th period; the i-th row and j-th column of KM represent the j-th auxiliary user belonging to the i-th total load;

[0064] The maximum / minimum power generation constraints of the unit are used as the third-second constraint function of the third indicator constraint model, as shown in formula (8):

[0065] g min,i ≤g i,j ≤g max,i (8)

[0066] in:

[0067] Where: g min,i , g max,i They represent the minimum and maximum power generation of the i-th unit, respectively, and g i,j represents the power generation of the i-th unit in the j-th period;

[0068] The unit climbing power constraint is used as the third constraint function of the third indicator constraint model, as shown in formula (9):

[0069]

[0070] in:

[0071] Where: R max,i is the maximum climbing power of the i-th unit;

[0072] The line flow constraint is used as the third and fourth constraint functions of the third index constraint model, as shown in formulas (10) and (11):

[0073] -pl max,i ≤pl i,j ≤pl max,i (10)

[0074] PL = SF × (KG × G - KD × D)

[0075] SF=XB×KL T ×(KL×XB×KL T )- 1

[0076] in:

[0077]

[0078] Where: pl i,j represents the power flow through the i-th line in the j-th period; pl max,i represents the maximum flow allowed to flow through the i-th line; PL is the line flow matrix; SF is the transfer factor matrix; XB is the admittance matrix, xb i represents the admittance of the ith line; KL is the line association matrix; KG is the generator group association matrix; KD is the load association matrix.

[0079] The first objective function unit is used to construct a first objective function model based on the optimal equipment load, the optimal electricity cost, and the optimal power generation cost, including:

[0080] The first objective function model is constructed based on the minimum variable cost of the power grid:

[0081]

[0082] Formula (12) indicates that part of the benefits brought by the reduction of power generation costs by peak shaving and valley filling is distributed to auxiliary user groups to meet their fee reduction standards, and the rest belongs to the power grid.

[0083] The first indicator constraint unit is used to construct a first indicator constraint model according to the optimal equipment load and the optimal electricity cost, including:

[0084] The percentage constraint of the reduction of user electricity charges based on the auxiliary user group (after each "auxiliary user" participates in the new mechanism, their electricity charges are reduced by an agreed percentage compared with the electricity charges under the original electricity price) is used as the first constraint function of the first indicator constraint model, as shown in formula (13):

[0085]

[0086] Where: bill i represents the minimum electricity cost of the user in the i-th auxiliary user group under the new electricity price; bill init,i represents the minimum electricity cost of the user in the i-th auxiliary user group under the original electricity price; δ i is the percentage of fee reduction agreed between the users of the i-th auxiliary user group and the grid, and is called bill init,i ·δ i is the target electricity fee for users in the i-th auxiliary user group;

[0087] The first and second constraint functions of the first indicator constraint model based on the maximum / minimum value constraint of electricity price are as shown in formula (14):

[0088] p min ≤p i,j ≤p max (14)

[0089] Where: p min is the minimum value allowed for the electricity price, p max is the maximum value allowed for the electricity price, p i,j is the electricity price of the user in the i-th auxiliary user group during the j-th period.

[0090] According to formula (6), formula (7) and formula (12), the third algorithm optimization unit, the second algorithm optimization unit and the first algorithm optimization unit will constitute a complex nested optimization problem, which is difficult to solve quickly using the traditional gradient method. Figure 5 As shown, the present invention can solve the nested optimization problem by using the gradient method by establishing the partial derivative of the optimal target value and the input parameters.

[0091] Specifically, the partial derivative of the optimal power generation cost to the optimal startup state input to the power grid dispatching module is established, including:

[0092] Step A1: Establish the constraint equations:

[0093]

[0094] In the formula, Q(G)=0 is the equation of the equality constraint transformed from the inequality constraint, and H(G) is the equation corresponding to the equality constraint equation, where:

[0095]

[0096] Among them, Q 1 (g), Q 2 (g), Q 3 (g), Q 4 (g), Q 5 (g), Q 6 (g) are, in order, the upper constraint of the unit, the upper constraint of the unit ramp power, the upper constraint of the power flow, the lower constraint of the unit, the lower constraint of the unit ramp power and the lower constraint of the power flow;

[0097] Step A2: Transform the unit upper limit constraint:

[0098]

[0099] Among them, A 1 is the coefficient matrix of the unknown matrix G, I is a matrix with dimensions NG rows and NG columns; B 1 is a constant matrix with dimensions NG rows and NT columns;

[0100] Step A3: Convert the upper limit constraint of the unit climbing power:

[0101]

[0102] in

[0103] In formula (17), A 2 is the coefficient matrix of the unknown matrix G, I is a matrix with dimensions NG rows and NG columns; B 2 is a constant matrix with dimensions NG rows and NT columns;

[0104] Step A4: Convert the line power flow upper limit constraint:

[0105]

[0106] Among them, A 3 is the coefficient matrix of the unknown matrix G; B 3 is a constant matrix with dimensions NG rows and NT columns;

[0107] Step A5: Transform the lower limit constraint of the unit:

[0108]

[0109] Among them, A 4 is the coefficient matrix of the unknown matrix G, I is a matrix with dimensions NG rows and NG columns; B 4 is a constant matrix with dimensions NG rows and NT columns;

[0110] Step A6: Convert the lower limit constraint of the unit climbing power:

[0111]

[0112] Among them, A 5 is the coefficient matrix of the unknown matrix G, I is a matrix with dimensions NG rows and NG columns; B 5 is a constant matrix with dimensions NG rows and NT columns;

[0113] Step A7: Convert the line power flow lower limit constraint:

[0114]

[0115] Among them, A 6 is the coefficient matrix of the unknown matrix G; B 6 is a constant matrix with dimensions NG rows and NT columns;

[0116] make:

[0117]

[0118] Among them, the dimension of coefficient matrix A is 4NG+2NL rows and NG columns, and the dimension of constant matrix B is 4NG+2NL rows and NT columns;

[0119] Step A8: Extract and solve the equation:

[0120] For the solution of each column of G, since there are NG generators, that is, there are NG unknowns, NG equations need to be solved; and the equation system is The equation for the jth column is, Therefore, to obtain the solution of the jth column of G, we need to extract (NG-1) equations from the inequality constraint equation Q(g); that is, we need to extract (NG-1) rows from A and (NG-1) elements from the jth column of B; the row where the (NG-1) elements extracted from the jth column of B are located corresponds to the row where the (NG-1) row elements extracted from A are located; therefore, for each combination of extracted equations in a column, we have Column combination; for the j-th column k-th column combination, it extracts (NG-1) rows from A and forms a coefficient matrix A(j,k) with NG rows and NG columns together with the coefficient terms of the equality constraint H(g); extracts (NG-1) elements from the j-th column of B and forms a constant matrix B(j,k) with NG rows and 1 columns together with the constant terms of the equality constraint H(g); at this time, the j-th column of G j The solution is:

[0121] G j =A(j,k) -1 ×B(j,k) (23)

[0122] Where: G j represents the j-th column of G;

[0123] And B has NT columns, so there are a total of total combination; for each total combination, calculate the solution of the column combination in the total combination separately, that is, G = [G 1 , G 2 , ..., G NT ]; therefore, there are NP possible solutions to G;

[0124] Step 9: Introduce the penalty function:

[0125] First, the three auxiliary penalty functions in formula (24) are introduced; among them, the parameter α is as large as possible, and the parameters β and γ are adjusted according to the actual situation; the auxiliary penalty function f is mainly used in the patent to deal with inequality constraints, and the inequality constraints are added to the objective function in the form of penalties to transform it into unconstrained optimization; 1 ≤x≤c 2 The method for handling inequality constraints of the form is as follows: 1 ≤x≤c 2 The inequality constraint of the form is split into x≤c 2 、c 1 ≤x two inequalities, where x is a variable, c 1 、c 2 is a constant; then, with the help of function f, x≤c 2 、c 1 ≤x is written as f(k·(xc 2 ))、f(-k·(x+c 1)) and added to the objective function, where k is a parameter and its size is adjusted according to the actual application:

[0126]

[0127] Therefore, converting formulas (8), (9) and (10) into penalty items corresponds to the following formula (25) as shown below; adding the penalty items in formula (5) and formula (25) to obtain formula (26), the constrained optimization can be converted into unconstrained optimization, that is, if a constraint is broken, the corresponding penalty item will be very large, if no constraint is broken, then l g1 ~l g6 are approximately equal to 0;

[0128]

[0129] PObj 2 =Obj 2 +l g1 +l g2 +l g3 +l g4 +l g5 +l g6 (26)

[0130] Step A10: Introduce the minimum approximate differentiable function:

[0131]

[0132] Where: X = [x 1 ,x 2 ,…x N ] 1×N

[0133] Formula (27) is an approximate function for screening the minimum value, that is, the input is the X array, and the output is the approximate value of the minimum value in X; the larger the α parameter is, the closer the approximate value is to the true value;

[0134] Step A11: Optimal value solution:

[0135] Substituting the solution of NP species G obtained in step A8 into formula (26), we get NP PObj 2 The value of is shown in formula (28); then substitute formula (28) into formula (27) to obtain PObj 2 The minimum value of PObj is shown in formula (29); 2 When the minimum value is obtained, all variables are within the constraints. g1 ~l g6 is equal to 0, so at this time PObj 2,min That is the required Obj 2 The minimum value that satisfies all constraints:

[0136] PObj 2 ={pobj 2 (1),pobj 2 (2),...pobj 2 (NP)} (28)

[0137] PObj 2,min =f min (PObj 2 ) (29)

[0138] Step A12: Obj 2 The minimum value of the optimal starting state MS is used to obtain the partial derivative:

[0139] From formula (23), we can see that since A(j, k) and B(j, k) contain the total load D, and D is obtained through the startup state MS, as shown in formula (7), G can be calculated by MS; and formulas (28) and (29) are calculated by G, so Obj 2 The minimum value of PObj 2,min It can be directly calculated from the startup state MS, that is, there is a mathematical expression f MS Make PObj 2,min =f MS (MS), so we can get

[0140] Establish the partial derivative of the optimal startup state for the multi-price group input to the price-load optimization module, including:

[0141] Step B1: Enable the state upper limit constraint for transition:

[0142]

[0143] Where Q 7 (g), Q 7 (g) are the upper and lower constraints of the device startup state MS;

[0144]

[0145] In formula (31), A 7 is the coefficient matrix of the unknown MS matrix, I is a matrix with dimensions of NT rows and NT columns; B 7 is a constant matrix with all 1s and dimensions of NT rows and NU columns;

[0146] Step B2: Transform the lower limit constraint of the startup state:

[0147]

[0148] In formula (32), A 8 is the coefficient matrix of the unknown MS matrix, I is a matrix with dimensions of NT rows and NT columns; B 8 is a constant matrix of all zeros, with dimensions of NT rows and NU columns;

[0149]

[0150] In formula (33), the dimension of coefficient matrix A is 2NT rows and NT columns, and the dimension of constant matrix B is 2NT rows and NU columns;

[0151] Step B3: Extract and solve the equation:

[0152] For MS T The solution of each column has NT time periods, that is, there are NT unknowns, so NT equations need to be solved; and the equation system is The equation for the jth column is, Therefore, to obtain the solution of the jth column, we need to extract (NT-1) equations from the inequality constraint equation Q(ms); that is, we need to extract (NT-1) rows from A and (NT-1) elements from the jth column of B as the constant matrix of the equation; the row where the (NT-1) elements extracted from the jth column of B are located corresponds to the row where the (NT-1) row elements extracted from A are located; therefore, for each combination of extracted equations in a column, we have Column combination; for the j-th column k-th column combination, it extracts (NT-1) rows from A and forms a coefficient matrix A(j,k) with NT rows and NT columns together with the coefficient terms of the equality constraint H(ms); extracts (NT-1) elements from the j-th column of B and forms a constant matrix B(j,k) with NT rows and 1 column together with the constant terms of the equality constraint H(ms); at this time MS T The MS of the jth column Tj The solution is:

[0153] MS Tj =A(j,k) -1 ×B(j,k) (34)

[0154] Where: MS Tj Indicates MS T The jth column of

[0155] And B has NU columns, so there are a total of total combination; for each total combination, the solution of the column combination in the total combination is calculated separately, that is, MS T =[MS T1 , M.S. T2 ,……,MS TNU ]; therefore MS TThere are also NP possible solutions.

[0156] Step B4: Introducing the penalty function

[0157] Equation (4) is converted into penalty items, which correspond to the items in formula (35) shown below. By adding the penalty items in formula (1) and formula (35), we get formula (36), which can transform the constrained optimization into unconstrained optimization. That is, if a constraint is violated, the corresponding penalty item will be very large. If no constraint is violated, that is, ms is not greater than 1 or less than 0, then l dr1 ~l dr2 are approximately equal to 0;

[0158]

[0159] PObj 1 =Obj 1 +l dr1 +l dr2 (36)

[0160] Step B5: Introduce the approximate differentiable function of the input variable corresponding to the minimum value:

[0161]

[0162] Where: X = {X 1 ,X 2 ,…X N} 1×N , f a is any function;

[0163] Formula (37) is the approximate differentiable function of the input variable corresponding to the minimum value, that is, the input X set, X i represents the i-th solution, and the output is the approximate value of X that can minimize the function f(X) in X. The larger the α parameter is, the closer the approximate value is to the true value.

[0164] Step B6: Optimal value and its corresponding input variable solution:

[0165] The NPs obtained in step B4 were MS T Substituting the solution into formula (36), we get NP PObj 1 The value of is shown in formula (38); then substitute formula (38) into formula (37) to obtain PObj 1 The minimum value of PObj is shown in formula (39); 1 When the minimum value is obtained, all variables are within the constraints. dr1 ~l dr2 is equal to 0, so at this time PObj 1,min That is the required Obj 1At the minimum value that satisfies all constraints; then substitute formula (38) into formula (37) to obtain formula (40), that is, PObj 1,min The corresponding input variable X min :

[0166] PObj 1 ={pobj 1 (MS 1 ),pobj 1 (MS 2 ),...pobj 1 (MS NP )} (38)

[0167] PObj 1,min =f min (PObj 1 ) (39)

[0168] MS min =f minX (PObj 1 (MS)) (40)

[0169] Step B7: Obj 1 The minimum value of and its corresponding MS are used to find the partial derivative of the electricity price group P:

[0170] From formula (39), formula (40) and formula (1), formula (36) and formula (37), we can know that PObj 1,min With MS min It can be directly obtained by mathematical formula from the electricity price group P; therefore, and

[0171] The partial derivative of the first objective function model for the multiple electricity price groups input to the electricity price-load optimization module is established, including:

[0172] Step C1: Inequality constraint conversion:

[0173] Formula (13) and formula (14) are converted into the penalty terms of formula (41) respectively; and the penalty terms of formula (41) are added to formula (12) to obtain formula (42):

[0174]

[0175]

[0176] Step C2: Solving the optimal electricity price group P:

[0177] According to step A12, and the value obtained in step B7 You can get According to step B7, Then PObj 3 The partial derivative of the electricity price group is shown in formula (43); Can be obtained directly;

[0178]

[0179] Step C3: Iterate and solve the optimal electricity price group P by gradient method, as shown in formula (44):

[0180]

[0181] Among them, λ is the iteration step size.

[0182] The above is a detailed introduction to the multi-electricity price demand response scheduling system for peak shifting phenomenon based on the gradient algorithm disclosed in the embodiment of the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for general technical personnel in this field, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A multi-price demand response dispatching system for peak shifting based on a gradient algorithm, characterized in that: include: The electricity price-load optimization module is used to determine the optimal equipment load and the optimal electricity cost based on the user equipment information of the auxiliary user group and the optimal multi-price curve group determined by the multi-price setting module; The grid dispatching module is used to determine the optimal power generation cost based on the total grid load and basic grid information; A multi-electricity price formulation module, used to calculate an optimal multi-electricity price curve group based on the optimal equipment load, the optimal electricity cost, and the optimal power generation cost; A partial derivative link is used to establish the partial derivatives of the optimal equipment load and the optimal electricity cost, the optimal power generation cost, and the optimal multi-price curve group, and solve the final optimal electricity price combination through the gradient algorithm based on the partial derivative; The multi-price setting module includes a first objective function unit, a first indicator constraint unit and a first algorithm optimization unit; the price-load optimization module includes a second objective function unit, a second indicator constraint unit and a second algorithm optimization unit; the power grid dispatching module includes a third objective function unit, a third indicator constraint unit and a third algorithm optimization unit; wherein: The second objective function unit is used to construct a second objective function model based on the optimal multi-electricity price curve group output by the multi-electricity price formulation module; the second indicator constraint unit is used to construct a second indicator constraint model according to the user equipment information; the second algorithm optimization unit is used to calculate the optimal equipment load and the optimal electricity cost according to the second objective function model and the second indicator constraint model; The third objective function unit is used to construct a third objective function model; the third indicator constraint unit is used to construct a third indicator constraint model according to the total load of the power grid and the basic information of the power grid; the third algorithm optimization unit is used to calculate the optimal power generation cost according to the third objective function model and the third indicator constraint model; The first objective function unit is used to construct a first objective function model based on the optimal equipment load, the optimal electricity cost, and the optimal power generation cost; the first indicator constraint unit is used to construct a first indicator constraint model according to the optimal equipment load and the optimal electricity cost; the first algorithm optimization unit is used to calculate multiple electricity price groups according to the first objective function model and the first indicator constraint model; The partial derivative link is used to establish the partial derivative of the first objective function model for the optimal power generation cost; establish the partial derivative of the optimal power generation cost for the optimal startup state input to the power grid dispatching module; establish the partial derivative of the optimal startup state for the multiple electricity price groups input to the electricity price-load optimization module; establish the partial derivative of the first objective function model for the optimal electricity consumption cost; establish the partial derivative of the optimal electricity consumption cost for the multiple electricity price groups input to the electricity price-load optimization module; establish the partial derivative of the first objective function model for the multiple electricity price groups input to the electricity price-load optimization module; finally, based on the partial derivative of the first objective function model for the multiple electricity price groups, the final optimal electricity price combination is solved by the gradient method.

2. The multi-price demand response scheduling system for peak shifting based on the gradient algorithm according to claim 1 is characterized in that: The second objective function unit is used to construct a second objective function model based on the optimal equipment load, the optimal electricity cost, and the optimal power generation cost, including: Construct the second objective function model: in: Where: Bill and Obj1(MS) are the total electricity charges of all auxiliary user groups; NU is the number of users in the auxiliary user group; NT is the minimum number of optimized time periods; NDev is the number of devices that can be transferred by all users in the auxiliary user group; e x =[1,…,1] represents a row vector with all 1s in row x column; P is the electricity price curve group of the auxiliary user group, p k,j is the electricity price of the kth user in the jth period; MS is the startup state of the transferable device, where ms i,j Indicates the startup status of the i-th device in the j-th period. If it is 1, the device is started, otherwise, the device is shut down; DP represents the device power, dp i represents the power of the i-th device; UL is the total load of the non-transferable equipment, ul k,j represents the total load of the non-transferable devices of the kth user in the jth period; the kth row and jth column in the SCA is 1, indicating that the jth device in the MS belongs to the kth user, otherwise it is 0; The second indicator constraint unit is used to construct a second indicator constraint model according to the user equipment information, including: Use the total startup time as the second constraint function of the second indicator constraint model: The total daily power consumption of each device is equal to the time required by the user for a certain production task: Where: ms i,j is the power consumption time of the i-th device in the j-th period, wt i is the total time required for the i-th device to start up; Use the startup state as the second constraint function of the second indicator constraint model: ms i,j =1 or 0 (4) In the formula: means that the startup status of the i-th device in the j-th period can only be startup or shutdown; Use the non-transferable constraint as the second and third constraint functions of the second indicator constraint model: Assuming that the i-th device needs to be started all the time or started at a specific time, the i-th device is non-transferable; the load of the i-th device is taken as a constant, that is, UL in formula (1), and does not participate in the optimization process.

3. The multi-price demand response scheduling system for peak shifting based on the gradient algorithm according to claim 2 is characterized in that: The third objective function unit is used to construct a third objective function model, including: The third objective function model is constructed based on the minimum total power generation cost: In the formula: Cost is the power generation cost; NG is the number of generator sets; a i 、b i 、c i is the coal consumption curve parameter of the i-th unit; g i,j represents the power generation of the i-th unit in the j-th period, and Obj2(G) is the minimum total power generation cost; The third indicator constraint unit is used to construct a third indicator constraint model according to the total load of the power grid and the basic information of the power grid, including: The power balance constraint is used as the third constraint function of the third indicator constraint model, as shown in formula (6): in: Where: ND is the total load of the whole network; D is the total load of the whole network, d i,j represents the power consumption of the i-th load in the j-th period; BL is the total load of other users in the load D minus the auxiliary user group, i.e., the base load; bl i,j represents the base load size of the i-th total load in the j-th period; ML is the total load of the transferable equipment of each auxiliary user, ml i,j represents the total load of the transferable equipment of the i-th auxiliary user in the j-th period; the i-th row and j-th column of KM represent the j-th auxiliary user belonging to the i-th total load; The maximum / minimum power generation constraints of the unit are used as the third-second constraint function of the third indicator constraint model, as shown in formula (8): g min,i ≤g i,j ≤g max,i (8) in: Where: g min,i , g max,i They represent the minimum and maximum power generation of the i-th unit, respectively, and g i,j represents the power generation of the i-th unit in the j-th period; The unit climbing power constraint is used as the third constraint function of the third indicator constraint model, as shown in formula (9): in: Where: R max,i is the maximum climbing power of the i-th unit; The line flow constraint is used as the third and fourth constraint functions of the third index constraint model, as shown in formulas (10) and (11): -pl max,i ≤pl i,j ≤pl max,i (10) Where: pl i,j represents the power flow through the i-th line in the j-th period; pl max,i represents the maximum flow allowed to flow through the i-th line; PL is the line flow matrix; SF is the transfer factor matrix; XB is the admittance matrix, xb i represents the admittance of the ith line; KL is the line association matrix; KG is the generator group association matrix; KD is the load association matrix.

4. The multi-price demand response scheduling system for peak shifting based on the gradient algorithm according to claim 3 is characterized in that: The first objective function unit is used to construct a first objective function model based on the optimal equipment load, the optimal electricity cost, and the optimal power generation cost, including: The first objective function model is constructed based on the minimum variable cost of the power grid: Formula (12) indicates that part of the benefits brought by the reduction of power generation costs by peak shaving and valley filling is distributed to auxiliary user groups to meet their fee reduction standards, and the rest belongs to the power grid.

5. The multi-price demand response scheduling system for peak shifting based on the gradient algorithm according to claim 4 is characterized in that: The first indicator constraint unit is used to construct a first indicator constraint model according to the optimal equipment load and the optimal electricity cost, including: The percentage constraint of the user electricity fee reduction based on the auxiliary user group is used as the first constraint function of the first indicator constraint model, as shown in formula (13): Where: bill i represents the minimum electricity cost of the user in the i-th auxiliary user group under the new electricity price; bill init,i represents the minimum electricity cost of the user in the i-th auxiliary user group under the original electricity price; δ i is the percentage of fee reduction agreed between the users of the i-th auxiliary user group and the grid, and is called bill init,i ·δ i is the target electricity fee for users in the i-th auxiliary user group; The first and second constraint functions of the first indicator constraint model based on the maximum / minimum value constraint of electricity price are as shown in formula (14): p min ≤p i,j ≤p max (14) Where: p min is the minimum value allowed for the electricity price, p max is the maximum value allowed for the electricity price, p i,j is the electricity price of the user in the i-th auxiliary user group during the j-th period.

6. The multi-price demand response scheduling system for peak shifting based on gradient algorithm according to claim 5, characterized in that: Establish the partial derivative of the optimal power generation cost to the optimal startup state input to the grid dispatch module, including: Step A1: Establish the constraint equations: In the formula, Q(G)=0 is the equation of the equality constraint transformed from the inequality constraint, and H(G)=0 is the equation corresponding to the equality constraint equation, where: Among them, Q1(g), Q2(g), Q3(g), Q4(g), Q5(g), and Q6(g) are the upper constraint of the unit, the upper limit constraint of the unit ramp power, the upper constraint of the power flow, the lower constraint of the unit, the lower limit constraint of the unit ramp power, and the lower constraint of the power flow. Step A2: Transform the unit upper limit constraint: Among them, A1 is the coefficient matrix of the unknown number matrix G, I is a matrix with dimensions of NG rows and NG columns; B1 is a constant matrix with dimensions of NG rows and NT columns; Step A3: Convert the upper limit constraint of the unit climbing power: in In formula (17), A2 is the coefficient matrix of the unknown matrix G, I is a matrix with dimensions of NG rows and NG columns; B2 is a constant matrix with dimensions of NG rows and NT columns; Step A4: Convert the line power flow upper limit constraint: Among them, A3 is the coefficient matrix of the unknown number matrix G; B3 is the constant matrix, whose dimension is NG rows and NT columns; Step A5: Transform the lower limit constraint of the unit: Among them, A4 is the coefficient matrix of the unknown number matrix G, I is a matrix with dimensions of NG rows and NG columns; B4 is a constant matrix with dimensions of NG rows and NT columns; Step A6: Convert the lower limit constraint of the unit climbing power: Among them, A5 is the coefficient matrix of the unknown number matrix G, I is a matrix with dimensions of NG rows and NG columns; B5 is a constant matrix with dimensions of NG rows and NT columns; Step A7: Convert the line power flow lower limit constraint: Among them, A6 is the coefficient matrix of the unknown number matrix G; B6 is the constant matrix, whose dimension is NG rows and NT columns; make: Among them, the dimension of coefficient matrix A is 4NG+2NL rows and NG columns, and the dimension of constant matrix B is 4NG+2NL rows and NT columns; Step A8: Extract and solve the equation: For the solution of each column of G, since there are NG generators, that is, there are NG unknowns, NG equations need to be solved; and the equation system is The equation for the jth column is, Therefore, to obtain the solution of the jth column of G, we need to extract (NG-1) equations from the inequality constraint equation Q(g); that is, we need to extract (NG-1) rows from A and (NG-1) elements from the jth column of B; the row where the (NG-1) elements extracted from the jth column of B are located corresponds to the row where the (NG-1) row elements extracted from A are located; therefore, for each combination of extracted equations in a column, we have Column combination; for the j-th column k-th column combination, it extracts (NG-1) rows from A and forms a coefficient matrix A(j,k) with NG rows and NG columns together with the coefficient terms of the equality constraint H(g); extracts (NG-1) elements from the j-th column of B and forms a constant matrix B(j,k) with NG rows and 1 columns together with the constant terms of the equality constraint H(g); at this time, the j-th column of G j The solution is: G j =A(j,k) -1 ×B(j,k) (23) Where: G j represents the j-th column of G; And B has NT columns, so there are a total of total combination; for each total combination, calculate the solution of the column combination in the total combination separately, that is, G = [G 1 , G 2 , ..., G NT ]; therefore, there are NP possible solutions to G; Step 9: Introduce penalty function: First, three auxiliary penalty functions are introduced in formula (24); among them, parameter α is as large as possible, and parameters β and γ are adjusted according to actual conditions; the auxiliary penalty function f is mainly used in the patent to process inequality constraints, and the inequality constraints are added to the objective function in the form of penalties to transform them into unconstrained optimization; the method for processing inequality constraints in the form of c1≤x≤c2 is as follows: the inequality constraint form of c1≤x≤c2 is split into two inequalities x≤c2 and c1≤x, where x is a variable and c1 and c2 are constants; then, with the help of function f, x≤c2 and c1≤x are written as penalty terms in the form of f(k·(x-c2)) and f(-k·(x+c1)), respectively, and added to the objective function, where k is a parameter and the size is adjusted according to the actual application: Therefore, converting formulas (8), (9) and (10) into penalty items corresponds to the following formula (25) as shown below; adding the penalty items in formula (5) and formula (25) to obtain formula (26), the constrained optimization can be converted into unconstrained optimization, that is, if a constraint is broken, the corresponding penalty item will be very large, if no constraint is broken, then l g1 ~l g6 are approximately equal to 0; PObj2=Obj2+l g1 +l g2 +l g3 +l g4 +l g5 +l g6 (26) Step A10: Introduce the minimum approximate differentiable function: Where: X = [x1, x2, ... x N ] 1×N Formula (27) is an approximate function for screening the minimum value, that is, the input is the X array, and the output is the approximate value of the minimum value in X; the larger the α parameter is, the closer the approximate value is to the true value; Step A11: Optimal value solution: Substitute the solution of NP species G obtained in step A8 into formula (26) to obtain the NP values ​​of PObj2, as shown in formula (28); then substitute formula (28) into formula (27) to find the minimum value of PObj2, as shown in formula (29); when PObj2 obtains the minimum value, all variables are not constrained, l g1 ~l g6 is equal to 0, so at this time PObj 2,min That is, the minimum value of Obj2 that satisfies all constraints: PObj2={pobj2(1),pobj2(2),...pobj2(NP)} (28) PObj 2,min =f min (PObj2) (29) Step A12: Take the partial derivative of the minimum value of Obj2 with respect to the optimal startup state MS: From formula (23), we can see that since A(j, k) and B(j, k) contain the total load D, and D is obtained through the startup state MS, as shown in formula (7), G can be calculated by MS; and formulas (28) and (29) are calculated by G, so the minimum value PObj of Obj2 is 2,min It can be directly calculated from the startup state MS, that is, there is a mathematical expression f MS Make PObj 2,min =f MS (MS), so we can get 7. The multi-price demand response dispatching system for peak shifting based on gradient algorithm according to claim 6, characterized in that: Establish the partial derivative of the optimal startup state for the multi-price group input to the price-load optimization module, including: Step B1: Enable the state upper limit constraint for transition: Where Q7(g) and Q7(g) are the upper and lower constraints of the device startup state MS respectively; In formula (31), A7 is the coefficient matrix of the unknown number MS matrix, I is a matrix with dimensions of NT rows and NT columns; B7 is a constant matrix with all values ​​1, with dimensions of NT rows and NU columns; Step B2: Transform the lower limit constraint of the startup state: In formula (32), A8 is the coefficient matrix of the unknown number MS matrix, I is a matrix with dimensions of NT rows and NT columns; B8 is a constant matrix of all zeros, with dimensions of NT rows and NU columns; In formula (33), the dimension of coefficient matrix A is 2NT rows and NT columns, and the dimension of constant matrix B is 2NT rows and NU columns; Step B3: Extract and solve the equation: For MS T The solution of each column has NT time periods, that is, there are NT unknowns, so NT equations need to be solved; and the equation system is The equation for the jth column is, Therefore, to obtain the solution of the jth column, we need to extract (NT-1) equations from the inequality constraint equation Q(ms); that is, we need to extract (NT-1) rows from A and (NT-1) elements from the jth column of B as the constant matrix of the equation; the row where the (NT-1) elements extracted from the jth column of B are located corresponds to the row where the (NT-1) row elements extracted from A are located; therefore, for each combination of extracted equations in a column, we have Column combination; for the j-th column k-th column combination, it extracts (NT-1) rows from A and forms a coefficient matrix A(j,k) with NT rows and NT columns together with the coefficient terms of the equality constraint H(ms); extracts (NT-1) elements from the j-th column of B and forms a constant matrix B(j,k) with NT rows and 1 column together with the constant terms of the equality constraint H(ms); at this time MS T The MS of the jth column Tj The solution is: MS Tj =A(j,k) -1 ×B(j,k) (34) Where: MS Tj Indicates MS T The jth column of And B has NU columns, so there are a total of total combination; for each total combination, the solution of the column combination in the total combination is calculated separately, that is, MS T =[MS T1 , M.S. T2 ,……,MS TNU ]; therefore MS T There are also NP possible solutions. Step B4: Introducing the penalty function Equation (4) is converted into penalty items, which correspond to the items in formula (35) shown below. By adding the penalty items in formula (1) and formula (35), we get formula (36), which can transform the constrained optimization into unconstrained optimization. That is, if a constraint is violated, the corresponding penalty item will be very large. If no constraint is violated, that is, ms is not greater than 1 or less than 0, then l dr1 ~l dr2 are approximately equal to 0; PObj1=Obj1+l dr1 +l dr2 (36) Step B5: Introduce the approximate differentiable function of the input variable corresponding to the minimum value: Where: X={X1,X2,…X N } 1×N , f a is any function; Formula (37) is the approximate differentiable function of the input variable corresponding to the minimum value, that is, the input X set, X i represents the i-th solution, and the output is the approximate value of X that can minimize the function f(X) in X. The larger the α parameter is, the closer the approximate value is to the true value. Step B6: Optimal value and its corresponding input variable solution: The NPs obtained in step B4 were MS T Substitute the solution into formula (36) to obtain NP values ​​of POBj1, as shown in formula (38); then substitute formula (38) into formula (37) to find the minimum value of POBj1, as shown in formula (39); when POBj1 reaches the minimum value, all variables are not constrained, l dr1 ~l dr2 is equal to 0, so at this time PObj 1,min That is, the minimum value of Obj1 that satisfies all constraints; then substitute formula (38) into formula (37) to obtain formula (40), that is, PObj 1,min The corresponding input variable X min : PObj1={pobj1(MS1),pobj1(MS2),...pobj1(MS NP )} (38) PObj 1,min =f min (PObj1) (39) MS min =f minX (PObj1(MS)) (40) Step B7: Calculate the partial derivative of the minimum value of Obj1 and its corresponding MS with respect to the electricity price group P: From formula (39), formula (40) and formula (1), formula (36) and formula (37), we can know that PObj 1,min With MS min It can be directly obtained by mathematical formula from the electricity price group P; therefore, and 8. The multi-price demand response scheduling system for peak shifting based on the gradient algorithm according to claim 7, characterized in that: The partial derivative of the first objective function model for the multiple electricity price groups input to the electricity price-load optimization module is established, including: Step C1: Inequality constraint conversion: Formula (13) and formula (14) are converted into the penalty terms of formula (41) respectively; and the penalty terms of formula (41) are added to formula (12) to obtain formula (42): Step C2: Solving the optimal electricity price group P: According to step A12, and the value obtained in step B7 You can get According to step B7, Then the partial derivative of PObj3 with respect to the electricity price group is as shown in formula (43); Can be obtained directly; Step C3: Iterate and solve the optimal electricity price group P by gradient method, as shown in formula (44): Among them, λ is the iteration step size.

Images