Virtual power plant adjustable resource optimization scheduling method based on mixed integer programming

Abstract

The invention discloses a virtual power plant adjustable resource optimization scheduling method based on mixed integer programming. The method comprises the following steps: step 1, establishing a virtual power plant adjustable resource model; step 2, establishing an objective function; and 3, solving the model. According to the method, firstly, characteristic analysis is carried out on adjustable resources, through energy consumption behavior and energy consumption characteristic analysis, the type of the adjustable resources is determined, the operation state of the adjustable resources, a load regulation and control method and the like are considered, a model is constructed based on deep learning, virtual power plant resource potential is modeled, and an adjustable range is determined; a virtual power plant adjustable resource optimization scheduling model based on mixed integer programming is established on the basis of factors such as the adjustment capability, the responsible time period, the response duration and the adjustment economy of adjustable resources by taking the lowest overall adjustment cost of a virtual power plant participating in an electric energy market and a peak regulation market as a target; and carrying out optimization solution on the established model by adopting a branch and bound method.

Description

Technical Field

[0001] The present invention relates to the technical field of computer-aided design, and in particular to a method for optimizing and scheduling adjustable resources of a virtual power plant based on mixed integer programming. Background Art

[0002] Virtual power plant technology, a key solution for effectively aggregating and deploying large-scale flexible resources and revolutionizing the problem of insufficient flexibility in power systems, offers digital and intelligent orchestration of these resources, enabling the aggregation, storage, supply, and utilization of energy from a vast array of distributed, flexible resources. Optimizing the scheduling of these resources is a key issue within virtual power plants.

[0003] Currently, traditional centralized control models struggle to adapt to the dynamic response characteristics of massive heterogeneous resources, resulting in a lack of a refined hierarchical mechanism for regulating resources such as distributed photovoltaics, energy storage equipment, and interruptible loads. Furthermore, information silos exist during resource aggregation, and there is a lack of a unified digital twin mapping and collaborative optimization mechanism between subsystems, resulting in time delays between regulation instructions and actual responses. Furthermore, existing dispatch models often use static optimization algorithms that are unable to dynamically adapt to multi-dimensional, real-time changes in power market time-of-use electricity prices, demand response incentive policies, and other factors, resulting in insufficient cost-effectiveness and flexibility in regulation strategies. These technical shortcomings often lead to virtual power plants facing problems such as rising regulation costs and low resource utilization when responding to scenarios such as peak load regulation and new energy consumption, making it difficult to meet the needs of new power systems for rapid and precise regulation of flexible resources.

[0004] Therefore, how to ensure energy demand while achieving accurate and efficient regulation of resources, reflecting the adjustment order and adjustment amount of various resources when responding to different demand requirements, and reducing the adjustment cost in practice has become a technical problem that technical personnel in this field urgently need to solve. Summary of the Invention

[0005] In view of the above-mentioned defects of the prior art, the present invention provides a method for optimizing and scheduling adjustable resources of a virtual power plant based on mixed integer programming. The purpose of the method is to ensure energy demand while achieving accurate and efficient regulation of resources. It can reflect the adjustment sequence and adjustment amount of various resources when responding to different demand requirements, and can reduce the adjustment cost in practice.

[0006] To achieve the above objectives, the present invention discloses a method for optimizing and scheduling adjustable resources of a virtual power plant based on mixed integer programming, comprising the following steps:

[0007] Step 1: Establish a virtual power plant adjustable resource model;

[0008] Step 2: Establish the objective function;

[0009] Step 3: Solve the model.

[0010] Preferably, step 1 is as follows:

[0011] Step 1.1: Modeling the adjustable capacity of building air conditioning, specifically:

[0012] The energy consumption model of the cooling side is constructed based on the cooling side energy consumption and the freezing side energy consumption of the cooling station's chiller, cooling tower, cooling water pump, and chilled water pump. Specifically, it is as follows:

[0013] P t cws =P t ch +P t ct +P t cwp ;

[0014] Among them, P t cws P is the energy consumption of the cooling side during period t; t ch P is the energy consumption of the cooling machine during period t; t ct P is the energy consumption of the cooling tower during period t; t cwp is the energy consumption of the chilled water pump during period t;

[0015] Establishing a chiller adjustable capacity model:

[0016]

[0017] Among them, T t cws -T t chws T is the temperature difference of the chilled water during the period t; t chws T is the chilled water outlet temperature during period t; t cws Cooling water outlet temperature; is the cooling machine load during period t; a0, a1, a2, a3, a4, and a5 are all empirical coefficients, obtained by Python learning based on the historical operating data of the cooling machine;

[0018] Establishing a cooling tower adjustable capacity model:

[0019]

[0020] Among them, T t cwris the cooling water inlet temperature during period t; b0, b1, b2, b3, b4, and b5 are empirical coefficients obtained by Python learning based on historical operating data of the cooling tower;

[0021] Cooling tower frequency The cooling tower energy consumption P during period t t ct Model:

[0022]

[0023] Among them, b6 is the empirical coefficient, which is obtained by Python learning based on the historical operation data of the cooling tower;

[0024] Establishing the cooling water pump adjustable capacity model:

[0025]

[0026] Where, ΔT t cw is the temperature difference between the inlet and outlet of the cooling water; c0, c1, c2, c3, c4, and c5 are all empirical coefficients, which are obtained by Python learning based on the historical operation data of the cooling water pump;

[0027] Establish cooling water pump frequency Energy consumption of cooling water pump t cwp Digital twin model of:

[0028]

[0029] Among them, c6 is the empirical coefficient, which is obtained by Python learning based on the historical operating data of the cooling water pump;

[0030] Establishing the chilled water pump adjustable capacity model:

[0031]

[0032] Among them, d0, d1, d2, d3, d4 and d5 are empirical coefficients, which are obtained by Python learning based on the historical operating data of the chilled water pump;

[0033] Establishing chilled water pump frequency Energy consumption of chilled water pump t chwp Digital twin model of:

[0034]

[0035] Among them, d6 is the empirical coefficient, which is obtained by Python learning based on the historical operating data of the chilled water pump;

[0036] Step 1.2: Model the adjustable capacity of the charging station, as follows:

[0037]

[0038] P t cs =N·P e (t);

[0039] Among them, E e (t) is the total energy accumulated at the charging station at time t; E e (t-1) is the total energy accumulated by the charging station at time t-1; P e (t) is the charging power of the electric vehicle at time t; Δt is the total charging time at time t after adjustment; E e,min (t) is the lower limit of the energy accumulated by the charging station at time t; E e,max (t) is the upper limit of energy accumulated by the charging station at time t; P t cs is the total charging power of multiple vehicles at time t; N is the number of vehicles charging at the charging station at time t; P e,min (t) is the minimum value of the car charging power at time t; P e,max (t) is the maximum value of the car charging power at time t;

[0040] Step 1.3: Modeling the adjustable capacity of energy storage. Based on the operating cost function of energy storage in the cost of energy storage and the response model based on incentive demand response, the modeling is as follows:

[0041]

[0042] in, is the operating cost of energy storage during period t; is the response cost function based on incentive demand response in period t; α es is the operating cost coefficient of energy storage; P t es is the charging and discharging power of the energy storage during period t, which is positive during charging and negative during discharging; is the adjustment time of energy storage in period t; α ib is the operating cost of energy storage during period t; β ib is the linear coefficient of the incentive-based demand response compensation amount; P t ib is the output based on incentive demand response during period t; The price that the output user can accept for adjustment during period t;

[0043] The energy storage constraints include storage capacity constraints and output upper and lower limits, specifically:

[0044]

[0045] The rated power of energy storage charging and discharging;

[0046] The basic power and capacity constraints of electricity are:

[0047]

[0048] in, is the initial charge of the energy storage during period t; The maximum capacity for charging energy storage; P t esc is the energy storage charging power; P t esf is the energy storage discharge power; η 充 is the energy storage charging efficiency; η 放 is the energy storage discharge efficiency.

[0049] More preferably, in step 2, assuming that the virtual power plant is set as the recipient of market prices when the model is constructed, the joint bidding model of the virtual power plant's adjustable resources in the day-ahead market and the real-time market is as follows:

[0050]

[0051] Among them, C t is the total cost of regulating the adjustable load during period t; is the adjustment cost of the adjustable load of the air conditioner during period t; is the adjustment cost of the adjustable load of the charging station during period t; is the regulation cost of the energy storage adjustable load during period t;

[0052] P t yy =P t ah +P t cs +P t es ;

[0053] Among them, P t yy is the load invited during period t; P t ah is the output of air conditioning load based on incentive demand response during period t; P t cs P is the output of the charging station based on incentive demand response during period t; t es The output of energy storage based on incentive demand response during period t;

[0054] Air conditioning optimization scheduling modeling:

[0055]

[0056] 0≤P t ah ≤(P t cws +P t chwp );

[0057] in, The price that the output user can accept for adjustment during period t; Adjust the time of air conditioning;

[0058] Charging station optimization scheduling modeling:

[0059]

[0060] P e,min (t)≤P t cs ≤P e,max (t)

[0061] in, The price that the output user can accept for adjustment during period t; The adjustment time for the charging station;

[0062] Energy storage optimization scheduling modeling:

[0063]

[0064] in, The price that the output user can accept for adjustment during period t; It is the adjustment time of energy storage.

[0065] More preferably, step 3 uses a branch-and-bound method to solve the established model, as follows:

[0066] Step 3.1: Input initial data;

[0067] Step 3.2: Input algorithm parameters.

[0068] Step 3.3: Solve the relaxation problem for the original problem A

[0069] like If there is no feasible solution, then there is no feasible solution, so stop looking for a solution;

[0070] like There is an optimal solution and it satisfies the integer constraint from 0 to 1, that is, it is also the optimal solution of A. It is also the upper and lower bounds of the optimal target value of the original problem A, that is, Stop solving;

[0071] Step 3.4, branch;

[0072] If the relaxation of the original problem A The optimal solution If the integer constraint is not met, then any non-integer variable is selected. Using the rounding function Construct two additional constraints

[0073]

[0074] is not greater than The largest integer;

[0075] Add constraints to the original problem A and We get two sub-problems A1 and A2:

[0076] minf(x1,x2)

[0077] sth(x1,x2)=0

[0078]

[0079] 0≤x j %≤0;

[0080] minf(x1,x2)

[0081] sth(x1,x2)=0

[0082]

[0083] 1≤x j %≤1;

[0084] The feasible solution set of A1 and A2 is also the feasible solution set of the original problem A. By decomposing the original problem A, the feasible solution set of A1 and A2 is eliminated in the relaxation problem. All feasible solutions whose components are in the range (0,1);

[0085] Step 3.5, bounding, specifically: using the branch and bound method to obtain a sub-problem A of the same level as the original problem A n Decomposition of a set, n = 1, 2, 3, ..., m;

[0086] The same level subproblems refer to each subproblem A n These are all subproblems obtained by A after the same number of branches;

[0087] If the current upper and lower bounds of each step are and f , each subproblem An The upper and lower bounds are and f n , then the current upper bound δ and lower bound φ are:

[0088]

[0089] φ=min( f1 、 f2 、…、 f m );

[0090] Step 3.6, compare and prune;

[0091] If there are too many discrete variables, the sub-problems to be branched will be too large, which will increase the number of branches and sub-problems, and reduce the calculation speed, so pruning can be performed;

[0092] A n There is no feasible solution;

[0093] A n The optimal solution of meets the integer constraint;

[0094] A n The optimal value of

[0095] By comparison, if the subproblem is not pruned, return to step 3.4;

[0096] When all subproblems have been pruned, that is, there are no subproblems to be processed, and the feasible solution that reaches the current upper bound f is the optimal solution to the original problem;

[0097] Step 3.7: Output the optimal solution as the optimized adjustable capacity interval.

[0098] Beneficial effects of the present invention:

[0099] The present invention is aimed at diversified virtual power plant resources. The regulation characteristics of resources such as air conditioners, charging stations, and energy storage are taken into account in the model. By analyzing energy consumption behavior and energy consumption characteristics, the model is constructed, and the operating status of adjustable resources and load control methods are fully considered. While ensuring energy demand, accurate and efficient resource regulation is achieved.

[0100] The present invention takes into account factors such as the adjustment capability, responsive period, response time, and adjustment economy of adjustable resources in the model, and can reflect the adjustment sequence and adjustment amount of various resources under different demand response requirements, which can reduce the adjustment cost in practice.

[0101] The concept, specific structure and technical effects of the present invention will be further described below in conjunction with the accompanying drawings to fully understand the purpose, characteristics and effects of the present invention. DETAILED DESCRIPTION

[0102] Example

[0103] The method for optimizing the scheduling of adjustable resources of a virtual power plant based on mixed integer programming includes the following steps:

[0104] Step 1: Establish a virtual power plant adjustable resource model;

[0105] Step 2: Establish the objective function;

[0106] Step 3: Solve the model.

[0107] In some embodiments, step 1 is as follows:

[0108] Step 1.1: Modeling the adjustable capacity of building air conditioning, specifically:

[0109] In practical applications, air conditioning and refrigeration systems include chillers, cooling towers, cooling water pumps, and chilled water pumps. The energy consumption of a cooling station mainly consists of four parts: chiller energy, cooling tower energy, cooling water pump energy, and chilled water pump energy.

[0110] The energy consumption model of the cooling side is constructed based on the energy consumption of the cooling machine, cooling tower, cooling water pump, and chilled water pump of the cooling station according to the energy consumption of the cooling side and the energy consumption of the chilled side. Specifically:

[0111] P t cws =P t ch +P t ct +P t cwp ;

[0112] Among them, P t cws P is the energy consumption of the cooling side during period t; t ch P is the energy consumption of the cooling machine during period t; t ct P is the energy consumption of the cooling tower during period t; t cwp is the energy consumption of the chilled water pump during period t;

[0113] Establishing a chiller adjustable capacity model:

[0114]

[0115] Among them, T t cws -T t chws T is the temperature difference of the chilled water during the period t; t chws T is the chilled water outlet temperature during period t; tcws Cooling water outlet temperature; is the cooling machine load during period t; a0, a1, a2, a3, a4, and a5 are all empirical coefficients, obtained by Python learning based on the historical operating data of the cooling machine;

[0116] Establishing a cooling tower adjustable capacity model:

[0117]

[0118] Among them, T t cwr is the cooling water inlet temperature during period t; b0, b1, b2, b3, b4, and b5 are empirical coefficients obtained by Python learning based on historical operating data of the cooling tower;

[0119] Cooling tower frequency The cooling tower energy consumption P during period t t ct Model:

[0120]

[0121] Among them, b6 is the empirical coefficient, which is obtained by Python learning based on the historical operation data of the cooling tower;

[0122] Establishing the cooling water pump adjustable capacity model:

[0123]

[0124] Where, ΔT t cw is the temperature difference between the inlet and outlet of the cooling water; c0, c1, c2, c3, c4, and c5 are all empirical coefficients, which are obtained by Python learning based on the historical operation data of the cooling water pump;

[0125] Establish cooling water pump frequency Energy consumption of cooling water pump t cwp Digital twin model of:

[0126]

[0127] Among them, c6 is the empirical coefficient, which is obtained by Python learning based on the historical operating data of the cooling water pump;

[0128] Establishing the chilled water pump adjustable capacity model:

[0129]

[0130] Among them, d0, d1, d2, d3, d4 and d5 are empirical coefficients, which are obtained by Python learning based on the historical operating data of the chilled water pump;

[0131] Establishing chilled water pump frequency Energy consumption of chilled water pump t chwp Digital twin model of:

[0132]

[0133] Among them, d6 is the empirical coefficient, which is obtained by Python learning based on the historical operating data of the chilled water pump;

[0134] Step 1.2: Model the adjustable capacity of the charging station, as follows:

[0135]

[0136] P t cs =N·P e (t);

[0137] Among them, E e (t) is the total energy accumulated at the charging station at time t; E e (t-1) is the total energy accumulated by the charging station at time t-1; P e (t) is the charging power of the electric vehicle at time t; Δt is the total charging time at time t after adjustment; E e,min (t) is the lower limit of the energy accumulated by the charging station at time t; E e,max (t) is the upper limit of energy accumulated by the charging station at time t; P t cs is the total charging power of multiple vehicles at time t; N is the number of vehicles charging at the charging station at time t; P e,min (t) is the minimum value of the car charging power at time t; P e,max (t) is the maximum value of the car charging power at time t;

[0138] In practical applications, the mathematical model for evaluating the adjustable capability of electric vehicles is as follows:

[0139]

[0140] Among them, E n is the size of demand; E c The upper limit of the rechargeable energy during the stay time;

[0141] If E n <E c , then the electric vehicle has adjustable capability, which is used to model the adjustable capability of the charging station.

[0142] Step 1.3: Modeling the adjustable capacity of energy storage. Based on the operating cost function of energy storage in the cost of energy storage and the response model based on incentive demand response, the modeling is as follows:

[0143]

[0144] in, is the operating cost of energy storage during period t; is the response cost function based on incentive demand response in period t; α es is the operating cost coefficient of energy storage; P t es is the charging and discharging power of the energy storage during period t, which is positive during charging and negative during discharging; is the adjustment time of energy storage in period t; α ib is the operating cost of energy storage during period t; β ib is the linear coefficient of the incentive-based demand response compensation amount; P t ib is the output based on incentive demand response during period t; The price that the output user can accept for adjustment during period t;

[0145] The energy storage constraints include storage capacity constraints and output upper and lower limits, specifically:

[0146]

[0147] The rated power of energy storage charging and discharging;

[0148] The basic power and capacity constraints of electricity are:

[0149]

[0150] in, is the initial charge of the energy storage during period t; The maximum capacity for charging energy storage; P t esc is the energy storage charging power; P t esf is the energy storage discharge power; η 充 is the energy storage charging efficiency; η 放 is the energy storage discharge efficiency.

[0151] In some embodiments, in step 2, assuming that the virtual power plant is set as the recipient of market prices when the model is constructed, the joint bidding model of the virtual power plant's adjustable resources in the day-ahead market and the real-time market is as follows:

[0152] In practical applications, a mixed integer programming model is established with the economic indicators of the system as the objective function. Therefore, it is assumed that the virtual power plant is set as the recipient of the market price when the model is constructed, and the impact of the virtual power plant's participation in the electricity market bidding on the market electricity price is not considered.

[0153]

[0154] Among them, C t is the total cost of regulating the adjustable load during period t; is the adjustment cost of the adjustable load of the air conditioner during period t; is the adjustment cost of the adjustable load of the charging station during period t; is the regulation cost of the energy storage adjustable load during period t;

[0155] P t yy =P t ah +P t cs +P t es ;

[0156] Among them, P t yy is the load invited during period t; P t ah is the output of air conditioning load based on incentive demand response during period t; P t cs P is the output of the charging station based on incentive demand response during period t; t es The output of energy storage based on incentive demand response during period t;

[0157] Air conditioning optimization scheduling modeling. In practical applications, the air conditioning adjustment cost mainly refers to the air conditioning adjustment cost function. The specific response cost function is as follows:

[0158]

[0159] 0≤P t ah ≤(P t cws +P t chwp );

[0160] in, The price that the output user can accept for adjustment during period t; Adjust the time of air conditioning;

[0161] Charging station optimization scheduling modeling. In practical applications, the adjustment cost of the charging station mainly refers to the adjustment cost function of the charging station. The specific response cost function is as follows:

[0162]

[0163] P e,min (t)≤P t cs ≤P e,max (t)

[0164] in, The price that the output user can accept for adjustment during period t; The adjustment time for the charging station;

[0165] Energy storage optimization scheduling modeling. In practical applications, the regulation cost of energy storage mainly refers to the regulation cost function of energy storage. The specific response cost function is as follows:

[0166]

[0167] in, The price that the output user can accept for adjustment during period t; It is the adjustment time of energy storage.

[0168] In some embodiments, step 3 solves the established model using a branch-and-bound method, as follows:

[0169] Step 3.1: Input initial data;

[0170] Step 3.2: Input algorithm parameters.

[0171] Step 3.3: Solve the relaxation problem for the original problem A

[0172] like If there is no feasible solution, then there is no feasible solution, so stop looking for a solution;

[0173] like There is an optimal solution and it satisfies the integer constraint from 0 to 1, that is, it is also the optimal solution of A. It is also the upper and lower bounds of the optimal target value of the original problem A, that is, Stop solving;

[0174] Step 3.4, branch;

[0175] If the relaxation of the original problem A The optimal solution If the integer constraint is not met, then any non-integer variable is selected. Using the rounding function Construct two additional constraints

[0176]

[0177] is not greater than The largest integer;

[0178] In practical applications, for 0 to 1 integer programming problems, since the values ​​of the relaxed discrete variables are all in the interval [0,1], when branching, each variable to be branched is simply decomposed into and There are two cases, that is, the values ​​are 0 and 1 respectively.

[0179] Add constraints to the original problem A and We get two sub-problems A1 and A2:

[0180] minf(x1,x2)

[0181] sth(x1,x2)=0

[0182]

[0183]

[0184] 0≤x j %≤0;

[0185] minf(x1,x2)

[0186] sth(x1,x2)=0

[0187]

[0188] 1≤x j %≤1;

[0189] The feasible solution set of A1 and A2 is also the feasible solution set of the original problem A. By decomposing the original problem A, the feasible solution set of A1 and A2 is eliminated in the relaxation problem. All feasible solutions whose components are in the range (0,1);

[0190] Step 3.5, bounding, specifically: using the branch and bound method to obtain a sub-problem A of the same level as the original problem A n Decomposition of a set, n = 1, 2, 3, ..., m;

[0191] The same level subproblems refer to each subproblem A n These are all subproblems obtained by A after the same number of branches;

[0192] If the current upper and lower bounds of each step are and f, each subproblem A n The upper and lower bounds are and f n , then the current upper bound δ and lower bound φ are:

[0193]

[0194] φ=min( f1 、 f2 、…、 f m );

[0195] In practical applications, for the previous step and f ,have f ≤φ, which shows that the iterative calculation of the branch and bound method makes the upper bound of the optimal objective value of the original problem smaller and smaller, while the lower bound becomes larger and larger.

[0196] Step 3.6, compare and prune;

[0197] If there are too many discrete variables, the sub-problems to be branched will be too large, which will increase the number of branches and sub-problems, and reduce the calculation speed, so pruning can be performed;

[0198] A n There is no feasible solution;

[0199] A n The optimal solution of meets the integer constraint;

[0200] A n The optimal value of

[0201] By comparison, if the subproblem is not pruned, return to step 3.4;

[0202] When all subproblems have been pruned, that is, there are no subproblems to be processed, and the feasible solution that reaches the current upper bound f is the optimal solution to the original problem;

[0203] Step 3.7: Output the optimal solution as the optimized adjustable capacity interval.

[0204] In practical applications, the present invention first analyzes the characteristics of adjustable resources, determines the type of adjustable resources of the invention through analysis of energy consumption behavior and energy consumption characteristics, considers the operating status of adjustable resources, load control methods, etc., builds a model based on deep learning, surveys the resource potential of virtual power plants, and determines the adjustable range; with the goal of minimizing the overall regulation cost of virtual power plants participating in external electricity energy markets and peak-shaving markets, based on factors such as the regulation capability, responsive period, response time, and regulation economy of adjustable resources, a virtual power plant adjustable resource optimization scheduling model based on mixed integer programming is established, and the established model is optimized and solved using the branch and bound method.

[0205] The above describes in detail the preferred embodiments of the present invention. It should be understood that those skilled in the art can make numerous modifications and variations based on the concepts of the present invention without inventive effort. Therefore, any technical solutions that can be derived by those skilled in the art through logical analysis, reasoning, or limited experimentation based on the concepts of the present invention and the prior art should be within the scope of protection defined by the claims.

Claims

1. A virtual power plant adjustable resource optimization scheduling method based on mixed integer programming; characterized in that: The process includes the following steps: Step 1: Establish a virtual power plant adjustable resource model; Step 2: Establish the objective function; Step 3: Solve the model.

2. The method for optimizing and scheduling adjustable resources of a virtual power plant based on mixed integer programming according to claim 1, characterized in that: Step 1 is as follows: Step 1.1: Modeling the adjustable capacity of building air conditioning, specifically: The energy consumption model of the cooling side is constructed based on the cooling side energy consumption and the freezing side energy consumption of the cooling station's chiller, cooling tower, cooling water pump, and chilled water pump. Specifically, it is as follows: P t cws =P t ch +P t ct +P t cwp ; Among them, P t cws P is the energy consumption of the cooling side during period t; t ch P is the energy consumption of the cooling machine during period t; t ct P is the energy consumption of the cooling tower during period t; t cwp is the energy consumption of the chilled water pump during period t; Establishing a chiller adjustable capacity model: Among them, T t cws -T t chws T is the temperature difference of the chilled water during the period t; t chws T is the chilled water outlet temperature during period t; t cws Cooling water outlet temperature; is the cooling machine load during period t; a0, a1, a2, a3, a4, and a5 are all empirical coefficients, obtained by Python learning based on the historical operating data of the cooling machine; Establishing a cooling tower adjustable capacity model: Among them, T t cwr is the cooling water inlet temperature during period t; b0, b1, b2, b3, b4, and b5 are empirical coefficients obtained by Python learning based on historical operating data of the cooling tower; Cooling tower frequency The cooling tower energy consumption P during period t t ct Model: Among them, b6 is the empirical coefficient, which is obtained by Python learning based on the historical operation data of the cooling tower; Establishing the cooling water pump adjustable capacity model: Where, ΔT t cw is the temperature difference between the inlet and outlet of the cooling water; c0, c1, c2, c3, c4, and c5 are all empirical coefficients, which are obtained by Python learning based on the historical operation data of the cooling water pump; Establish cooling water pump frequency Energy consumption of cooling water pump t cwp Digital twin model of: Among them, c6 is the empirical coefficient, which is obtained by Python learning based on the historical operating data of the cooling water pump; Establishing the chilled water pump adjustable capacity model: Among them, d0, d1, d2, d3, d4 and d5 are empirical coefficients, which are obtained by Python learning based on the historical operating data of the chilled water pump; Establishing chilled water pump frequency Energy consumption of chilled water pump t chwp Digital twin model of: Among them, d6 is the empirical coefficient, which is obtained by Python learning based on the historical operating data of the chilled water pump; Step 1.2: Model the adjustable capacity of the charging station, as follows: P t cs =N·P e (t); Among them, E e (t) is the total energy accumulated at the charging station at time t; E e (t-1) is the total energy accumulated by the charging station at time t-1; P e (t) is the charging power of the electric vehicle at time t; Δt is the total charging time at time t after adjustment; E e,min (t) is the lower limit of the energy accumulated by the charging station at time t; E e,max (t) is the upper limit of energy accumulated by the charging station at time t; P t cs is the total charging power of multiple vehicles at time t; N is the number of vehicles charging at the charging station at time t; P e,min (t) is the minimum value of the car charging power at time t; P e,max (t) is the maximum value of the car charging power at time t; Step 1.3: Modeling the adjustable capacity of energy storage. Based on the operating cost function of energy storage in the cost of energy storage and the response model based on incentive demand response, the modeling is as follows: in, is the operating cost of energy storage during period t; is the response cost function based on incentive demand response in period t; α es is the operating cost coefficient of energy storage; P t es is the charging and discharging power of the energy storage during period t, which is positive during charging and negative during discharging; is the adjustment time of energy storage in period t; α ib is the operating cost of energy storage during period t; β ib is the linear coefficient of the incentive-based demand response compensation amount; P t ib is the output based on incentive demand response during period t; The price that the output user can accept for adjustment during period t; The energy storage constraints include storage capacity constraints and output upper and lower limits, specifically: The rated power of energy storage charging and discharging; The basic power and capacity constraints of electricity are: in, is the initial charge of the energy storage during period t; The maximum capacity for charging energy storage; P t esc is the energy storage charging power; P t esf is the energy storage discharge power; η 充 is the energy storage charging efficiency; η 放 is the energy storage discharge efficiency.

3. The method for optimizing and scheduling adjustable resources of a virtual power plant based on mixed integer programming according to claim 2, characterized in that: In step 2, assuming that the virtual power plant is set as the recipient of market prices when the model is constructed, the joint bidding model of the virtual power plant's adjustable resources in the day-ahead market and the real-time market is as follows: Among them, C t is the total cost of regulating the adjustable load during period t; is the adjustment cost of the adjustable load of the air conditioner during period t; is the adjustment cost of the adjustable load of the charging station during period t; is the regulation cost of the energy storage adjustable load during period t; P t yy =P t ah +P t cs +P t es ; Among them, P t yy is the load invited during period t; P t ah is the output of air conditioning load based on incentive demand response during period t; P t cs P is the output of the charging station based on incentive demand response during period t; t es The output of energy storage based on incentive demand response during period t; Air conditioning optimization scheduling modeling: 0≤P t ah ≤(P t cws +P t chwp ); in, The price that the output user can accept for adjustment during period t; Adjust the time of air conditioning; Charging station optimization scheduling modeling: P e,min (t)≤P t cs ≤P e,max (t) in, The price that the output user can accept for adjustment during period t; The adjustment time for the charging station; Energy storage optimization scheduling modeling: in, The price that the output user can accept for adjustment during period t; It is the adjustment time of energy storage.

4. The method for optimizing and scheduling adjustable resources of a virtual power plant based on mixed integer programming according to claim 3 is characterized in that: Step 3 uses the branch and bound method to solve the established model, as follows: Step 3.1: Input initial data; Step 3.2: Input algorithm parameters. Step 3.3: Solve the relaxation problem for the original problem A like If there is no feasible solution, then there is no feasible solution, so stop looking for a solution; like There is an optimal solution and it satisfies the integer constraint from 0 to 1, that is, it is also the optimal solution of A. It is also the upper and lower bounds of the optimal target value of the original problem A, that is, Stop solving; Step 3.4, branch; If the relaxation of the original problem A The optimal solution If the integer constraint is not met, then any non-integer variable is selected. Using the rounding function Construct two additional constraints is not greater than The largest integer; Add constraints to the original problem A and We get two sub-problems A1 and A2: minf(x1,x2) sth(x1,x2)=0 0≤x j ％≤0； minf(x1,x2) sth(x1,x2)=0 1≤x j ％≤1； The feasible solution set of A1 and A2 is also the feasible solution set of the original problem A. By decomposing the original problem A, the feasible solution set of A1 and A2 is eliminated in the relaxation problem. All feasible solutions whose components are in the range (0,1); Step 3.5, bounding, specifically: using the branch and bound method to obtain a sub-problem A of the same level as the original problem A n Decomposition of a set, n = 1, 2, 3, ..., m; The same level subproblems refer to each subproblem A n These are all subproblems obtained by A after the same number of branches; If the current upper and lower bounds of each step are and f , each subproblem A n The upper and lower bounds are and f n , then the current upper bound δ and lower bound φ are: φ=min( f1 、 f2 、…、 f m ); Step 3.6, compare and prune; If there are too many discrete variables, the sub-problems to be branched will be too large, which will increase the number of branches and sub-problems, and reduce the calculation speed, so pruning can be performed; A n There is no feasible solution; A n The optimal solution of meets the integer constraint; A n The optimal value of By comparison, if the subproblem is not pruned, return to step 3.4; When all subproblems have been pruned, there are no more subproblems to be processed and the current upper bound is reached The feasible solution of is the optimal solution to the original problem; Step 3.7: Output the optimal solution as the optimized adjustable capacity interval.

Images