Source load storage micro-grid management method and system for cluster-containing electric vehicles in worst scene

Abstract

The invention discloses a method and a system for managing a source-load storage micro-grid containing cluster electric vehicles for the worst scene, and relates to the field of power system optimization scheduling, and the method comprises the following steps: S1, in the source-load storage micro-grid containing the cluster electric vehicles, carrying out optimization scheduling on the source-load storage micro-grid containing the cluster electric vehicles; establishing an operation constraint and coordination control model comprising an electric vehicle cluster, an energy storage unit, a gas turbine unit and a photovoltaic unit; s2, a two-stage three-layer robust optimization model of a min-max-min structure is established, decision variables are divided into a first-stage variable x and a second-stage variable y, and the two-stage three-layer robust optimization model takes the minimum daily operation cost of the microgrid as an optimization target and takes the operation constraints and the coordination control model in the step S1 as constraints; and S3, adopting a column and constraint generation algorithm to decompose the original problem in the step S2 into a main problem and a sub-problem for iterative solution. According to the invention, Camp is adopted; solving the established model according to a G algorithm idea to obtain an optimal day-ahead economic dispatching scheme of the micro-grid in the worst scene; the conservative property of the micro-grid optimization scheme can be flexibly adjusted by changing the uncertainty robust adjustment coefficient, and reasonable selection between the operation cost and the operation risk by a micro-grid dispatcher is facilitated.

Description

Technical Field

[0001] The present invention relates to the field of power system optimization and dispatching, and in particular to a source-load-storage microgrid management method and system containing cluster electric vehicles for the worst scenario. Background Art

[0002] With the proposal of the "dual carbon" goal, the development of distributed new energy technologies, mainly wind power and photovoltaics, has become the core of research in the energy field. With the increasing penetration of new energy in the power grid, its volatility and uncertainty have increased the difficulty of power grid operation and dispatch, and the flexibility of the power system faces severe challenges. Microgrid, as an integrated system containing a variety of distributed power sources, energy storage systems and loads, can achieve a high degree of autonomy and is an effective way to achieve a high proportion of new energy access to the power grid and economic operation of the system.

[0003] In addition, with the increase in the number of electric vehicles (EVs) in my country and the development of battery technology, EVs as distributed energy storage units can be combined with vehicle-to-grid (V2G) technology to reasonably arrange charging and discharging plans to achieve "peak shaving and valley filling" of the power grid, making EVs one of the most promising demand-side resources in the future. However, EVs have the characteristics of small capacity and large scale, and are scattered at the bottom of the system structure, resulting in their relative "insulation" from the market's electricity price signals and insufficient ability to participate in optimization scheduling. When the number of EVs is large, if each EV is accurately modeled, the optimization solution will fall into the "dimensionality disaster", making the solution too slow or even impossible to solve. Therefore, it is necessary to study a flexibility characterization method for clustered EVs to participate in demand response.

[0004] At the same time, the greater randomness of new energy makes the operation of microgrids have strong uncertainty factors. In addition, only by fully considering the uncertainty of wind power and photovoltaic unit output can a more economical and reliable dispatching plan be formulated. The two-stage robust optimization method has a good performance in dealing with the volatility of wind and photovoltaic output. It establishes a robust optimization model that replaces random variables with uncertain sets and finds the dispatching plan of microgrids in the "worst" scenario, which can make the dispatching plan of microgrids more meaningful and in line with the actual situation. Summary of the invention

[0005] The present invention aims to provide a source-load-storage microgrid management method and system containing cluster electric vehicles for the worst scenario, taking into account the uncertainty of photovoltaic and load, and transforming the two-stage problem into a main problem and sub-problems in mixed integer linear form for solution. In addition, a robust adjustment coefficient is added to the two-stage robust model, which allows microgrid dispatchers to flexibly choose the degree of conservatism of the dispatching scheme. By comparing the charging and discharging power sources of cluster electric vehicles under different time-of-use electricity price incentives; the impact of introducing new time-of-use electricity prices with high price differences, high mean and spikes on the overall operating economy of the microgrid is analyzed.

[0006] The present invention adopts the following technical solution: a source-load-storage microgrid management method containing cluster electric vehicles for the worst scenario, comprising:

[0007] Step S1, in a source-load-storage microgrid containing a cluster of electric vehicles, establishing an operation constraint and coordinated control model including an electric vehicle cluster, energy storage, a gas turbine unit, and a photovoltaic unit;

[0008] Step S2, establish a two-stage three-layer robust optimization model with a min-max-min structure, divide the decision variables into the first-stage variable x and the second-stage variable y, and the two-stage three-layer robust optimization model takes the minimum daily operation cost of the microgrid as the optimization goal, and takes the operation constraints and coordinated control model in step S1 as constraints;

[0009] Step S3, using the column and constraint generation algorithm, decomposes the original problem in step S2 into a main problem and sub-problems and solves them iteratively.

[0010] Furthermore, the components of the coordinated control model include a micro gas turbine power model, and the formula is as follows:

[0011] C PG,t =k G P G,t Δt

[0012] Among them, k G P is the comprehensive cost factor considering the fuel cost and maintenance cost of the gas turbine; G,t is the output power of the gas turbine during period t; Δt is the scheduling step length; C PG,t is the unit power generation cost of the gas turbine in period t.

[0013] Furthermore, the components of the coordinated control model include a charging and discharging power and capacity model of the energy storage battery ESS, which is expressed as follows:

[0014]

[0015] Among them, W b is the capacity of the energy storage battery; subscript b represents microgrid park b; subscript t represents time period t; T is the total time period in the dispatching cycle; is the energy level of the energy storage battery in park b during period t, is the energy level of the energy storage battery in park b at time t-1; and are the charging power and discharging power of ESS in park b during period t; η ess,ch With η ess,dis They are the charging efficiency and discharging efficiency of ESS respectively; and They are charging power limit and discharging power limit respectively; A variable indicating whether the energy storage battery is in a charging state, 1 if it is, and 0 if it is not; A variable indicating whether the energy storage battery is in a discharge state, 1 if it is, and 0 if it is not; and They are the minimum state of charge and maximum state of charge of ESS respectively; and They represent the charge states at time periods t=0 and t=T in the ESS scheduling period respectively.

[0016] Furthermore, the components of the coordinated control model include a power energy boundary model of the cluster electric vehicles in time shift, which is expressed as follows:

[0017]

[0018] Among them, E i is the battery capacity of the i-th electric vehicle; is the energy level of the battery of the i-th electric vehicle at time t and time t-1; is the power demand level of the i-th electric vehicle; SOC EV,min and SOC EV,max They are the upper and lower limits of battery capacity respectively; Respectively represent the charging state and discharging state of the i-th electric vehicle, which are 0-1 variables; 1 indicates that the i-th electric vehicle is in the charging state, otherwise, it is not; 1 indicates that the i-th electric vehicle is in the discharging state, otherwise, it is not in the discharging state; and are the battery charging power and discharging power of electric vehicle i in period t respectively; P i EV,ch,max and P i EV,dis,max They are respectively the battery charging power limit and the discharging power limit; η EV,ch With η EV,dis They are the battery charging efficiency and discharging efficiency respectively; They represent the charging state and discharging state of the i-th electric vehicle in park b respectively; Indicates that the i-th electric car is at t a The energy level of the battery during the period.

[0019] Furthermore, the components of the coordinated control model include uncertainty models of photovoltaic and load, which are expressed as follows:

[0020] u=[u PV,t u L,t ] T ,t=1,2…T

[0021]

[0022] Where u is the output of photovoltaic and load in the uncertainty model; u PV,t and u L,t are photovoltaic output and conventional load power respectively; ν PV,t Indicates whether the uncertainty of photovoltaic power in this period reaches the boundary of the fluctuation range, if yes, it is 1, otherwise, it is 0; ν L,t Indicates whether the uncertainty of the load in this period reaches the boundary of the fluctuation range, if yes, it is 1, if not, it is 0; and are the maximum deviation of photovoltaic fluctuation and the maximum deviation of load power fluctuation respectively; Γ PV is the robust regulation coefficient of photovoltaics; Γ L is the robust regulation coefficient of the load; and are the predicted power of photovoltaic output and conventional load respectively; T is a scheduling cycle.

[0023] Furthermore, the components of the coordinated control model include a shiftable load model, and the dispatching cost of the shiftable load is:

[0024]

[0025] Among them, K tr The unit compensation cost for the translatable load; is the expected electricity demand of residents in period t; P tr,t is the actual dispatch power of residents in period t; the absolute value represents the difference between the expected power demand and the actual dispatch power of the shiftable load, Δt is the minimum dispatch time scale, C tr,t represents the dispatch cost of the shiftable load;

[0026] Introduce auxiliary variable P tr1,t and P tr2,t And add constraints to convert the dispatch cost of the shiftable load into a linear form:

[0027]

[0028] P tr1,t≥0,P tr2,t ≥0

[0029] C tr,t =K tr [P tr1,t +P tr2,t ]

[0030] Among them, P tr1,t and P tr2,t are auxiliary variables, which represent that the residents have two dispatch powers greater than or equal to 0 in time period t. In the constraint relationship, the sum of the two is equivalent to the absolute value of the difference between the expected electricity demand and the actual dispatch power of the shiftable load;

[0031] The power consumption characteristics satisfied by the movable load are expressed as:

[0032]

[0033] Among them, D tr is the total power demand during the period when the load can be shifted to participate in the microgrid dispatch; P tr,t The actual dispatch power for residents in period t; and They represent the minimum and maximum electricity demands of the user in period t respectively; Δt represents the minimum scheduling period.

[0034] Furthermore, in step S2, the deterministic problem is:

[0035]

[0036] Among them, C PG,t is the unit power generation cost of the gas turbine in period t; C EV,t is the charging and discharging cost of electric vehicles in period t; C tr,t is the dispatching cost of the shiftable load in period t, C ESS,t is the charging and discharging operating cost of the energy storage during period t; C M,t is the interaction power cost between the microgrid and the distribution network in period t;

[0037] The pre-scheduling phase problem is described as:

[0038]

[0039] Among them, c T x is the operating cost of the first stage, including the charging and discharging costs of cluster electric vehicles; d Ty is the operating cost of the second stage, including the gas turbine power generation cost, the shiftable load scheduling cost, the energy storage facility operation cost and the microgrid distribution network interaction power cost; x is the decision variable of the outer min problem; u and y are the decision variables of the inner max-min problem, c is the coefficient column vector of the cluster electric vehicle charging and discharging cost function in the first stage; d is the coefficient column vector of the operating cost function in the second stage; Q, R, F, G, D, K, E, L, Y are the coefficient matrices of each constraint variable; m, g, f, h are constant column vectors.

[0040] Furthermore, the step S3 comprises:

[0041] Step S31, applying the idea of ​​zero-sum game to the pre-scheduling phase problem in step S2, using the column and constraint generation algorithm to solve it, decoupling the original problem into a main problem and sub-problems for iteration;

[0042] Among them, the main problem after decomposition of the original problem is:

[0043]

[0044] Where l is the number of iterations; θ is the intermediate variable introduced in the problem decomposition process; y l is the auxiliary variable related to the subproblem introduced at the lth iteration; The worst scenario of photovoltaic and load obtained by solving the sub-problem at the lth iteration;

[0045] Among them, the sub-problems obtained after the original problem is decomposed are:

[0046]

[0047] Among them, x * The first-stage variables obtained by solving the main problem; V represents the coefficient matrix of the cluster electric vehicle charging and discharging constraint variables;

[0048] Using KKT or strong duality theory, the subproblem is transformed into a max problem.

[0049] Furthermore, the step S3 comprises:

[0050] Step S32, use the big M method and introduce continuous auxiliary variables to linearize the sub-problems. Through the above transformation, the main problem and sub-problems decomposed by the two-stage three-layer robust optimization model based on the worst scenario are mixed integer linear programming. After continuous iterative solution, an economic dispatch plan that optimizes the operating cost of the microgrid under the worst scenario is obtained.

[0051] The present invention also provides a source-load-storage microgrid management system containing cluster electric vehicles for the worst scenarios, including a distribution network, photovoltaics, a micro gas turbine unit, energy storage, various demand response flexible loads, cluster electric vehicles and a microgrid central controller. The distribution network is power-connected to the photovoltaics, the micro gas turbine unit, the energy storage, the various demand response flexible loads and the cluster electric vehicles; the microgrid central controller is communication-connected to the photovoltaics, the micro gas turbine unit, the energy storage, the various demand response flexible loads and the cluster electric vehicles, and the microgrid central controller executes the above-mentioned source-load-storage microgrid management method containing cluster electric vehicles for the worst scenarios.

[0052] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:

[0053] The present invention provides a source-load-storage microgrid management method and system containing cluster electric vehicles for the worst scenario. The uncertainty of renewable energy output and load is taken into account, and the constructed model is solved by CC&G algorithm idea to obtain the optimal day-ahead economic dispatch plan of the microgrid under the worst scenario. By changing the robust adjustment coefficient of uncertainty, the conservatism of the microgrid optimization plan can be flexibly adjusted, which is conducive to the microgrid dispatcher to make a reasonable choice between operation cost and operation risk. Moreover, the advantage of the two-stage robust optimization method over the deterministic optimization method becomes more significant as the prediction error increases, which shows the importance of considering uncertainty when performing optimization dispatching, and the good applicability of robustness in solving such problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 The present invention is a flow chart of a source-load-storage microgrid management method containing cluster electric vehicles for the worst scenario.

[0055] Figure 2 for Figure 1 Flow chart of step S3 in FIG.

[0056] Figure 3 This is a structural block diagram of a source-load-storage microgrid management system containing cluster electric vehicles for the worst scenario. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the application is further elaborated below, and the described embodiment is only a part of the embodiment involved in the present invention. All non-innovative embodiments of other researchers in this field on this embodiment belong to the protection scope of the present invention. At the same time, for the step numbering in the embodiment of the present invention, it is only set for the convenience of explanation, and the order between the steps is not limited in any way. The execution order of each step in the embodiment can be adaptively adjusted according to the understanding of those skilled in the art.

[0058] The present invention provides a source-load-storage microgrid management method containing cluster electric vehicles for the worst scenario, using a two-stage robust microgrid economic optimization scheduling solution method to manage the source-load-storage microgrid containing cluster electric vehicles, such as Figure 1 As shown, the following steps are included:

[0059] Step S1, in a source-load-storage microgrid containing a cluster of electric vehicles, an operation constraint and coordination control model including an electric vehicle cluster, energy storage, a gas turbine unit and a photovoltaic unit is established.

[0060] More specifically, the components of the coordinated control model include: a micro gas turbine power model, a power energy boundary model of cluster electric vehicles with time shifting, a charging and discharging power and capacity model of the energy storage battery ESS, an uncertainty model of photovoltaics and loads, and a demand response flexible load (shiftable load) model.

[0061] Step S2, establish a two-stage three-layer robust optimization model with a min-max-min structure, divide the decision variables into the first-stage variables x and the second-stage variables y, and the two-stage three-layer robust optimization model takes the minimization of the microgrid daily operation cost as the optimization goal, and takes the various operation constraints and coordinated control models in step S1 as constraints.

[0062] Step S3, using the column and constraint generation algorithm, decomposes the original problem in step S2 into a main problem and sub-problems and solves them iteratively.

[0063] Among them, the sub-problems in max-min form are converted into max form through KKT conditions or strong duality theory, and then solved using the big M method to obtain the optimal strategy for economic dispatch of the source-load-storage microgrid containing cluster electric vehicles for the worst scenario.

[0064] Furthermore, the power model of the micro gas turbine in step S1 is as follows:

[0065] C PG,t =k G P G,t Δt

[0066] Among them, k GP is the comprehensive cost factor considering the fuel cost and maintenance cost of the gas turbine; G,t is the output power of the gas turbine during period t; Δt is the scheduling step length, which can be taken as lh in a specific embodiment; C PG,t is the unit power generation cost of the gas turbine in period t.

[0067] Considering that the response speed of the micro gas turbine power output is much less than 1h, only the output power constraint of the gas turbine is considered, and its climbing constraint and start-stop constraint are not considered. Its output power constraint is

[0068]

[0069] Where: and is the minimum and maximum output power of the gas turbine, P G,t is the output power of the gas turbine during period t.

[0070] Furthermore, the charging and discharging power and capacity model of the energy storage battery ESS in step S1 is expressed as follows:

[0071]

[0072] Among them, W b is the capacity of the energy storage battery; subscript b represents microgrid park b; subscript t represents time period t; T is the total time period in the dispatching cycle; is the energy level of the energy storage battery in park b during period t, is the energy level of the energy storage battery in park b at time t-1; and are the charging power and discharging power of ESS in park b during period t; η ess,ch With η ess,dis They are the charging efficiency and discharging efficiency of ESS respectively; and They are charging power limit and discharging power limit respectively; A variable indicating whether the energy storage battery is in a charging state, 1 if it is, and 0 if it is not; A variable indicating whether the energy storage battery is in a discharge state, 1 if it is, and 0 if it is not; and They are the minimum state of charge and maximum state of charge of ESS respectively; and They respectively represent the charge status at t=0 and t=T in the ESS scheduling cycle. The capacity of the energy storage equipment must remain consistent to ensure the continuity of scheduling.

[0073] Furthermore, the power energy boundary model of the cluster electric vehicles in step S1 is expressed as follows:

[0074]

[0075] Among them, E i is the battery capacity of the i-th electric vehicle; is the energy level of the battery of the i-th electric vehicle at time t and time t-1; is the power demand level of the i-th electric vehicle; SOC EV,min and SOC EV,max They are the upper and lower limits of battery capacity respectively; Respectively represent the charging state and discharging state of the i-th electric vehicle, which are 0-1 variables; 1 indicates that the i-th electric vehicle is in the charging state, otherwise, it is not; 1 indicates that the i-th electric vehicle is in the discharging state, otherwise, it is not in the discharging state; and are the battery charging power and discharging power of electric vehicle i during period t; P i EV,ch,max and P i EV,dis,max They are respectively the battery charging power limit and the discharging power limit; η EV,ch With η EV,dis They are the battery charging efficiency and discharging efficiency respectively; They represent the charging state and discharging state of the i-th electric vehicle in park b respectively; Indicates that the i-th electric car is at t a The energy level of the battery during the period.

[0076] Furthermore, the uncertainty model of photovoltaic and load in step S1 is expressed as follows:

[0077] u=[u PV,t u L,t ] T ,t=1,2…T

[0078]

[0079] Where u is the output of photovoltaic and load in the uncertainty model; u PV,t and u L,t are photovoltaic output and conventional load power respectively; ν PV,t Indicates whether the uncertainty of photovoltaic power in this period reaches the boundary of the fluctuation range, if yes, it is 1, otherwise, it is 0; ν L,t Indicates whether the uncertainty of the load in this period reaches the boundary of the fluctuation range, if yes, it is 1, if not, it is 0; and are the maximum deviation of photovoltaic fluctuation and the maximum deviation of load power fluctuation respectively; Γ PVis the robust regulation coefficient of photovoltaics; Γ L is the robust regulation coefficient of the load; the robust regulation coefficient indicates the total number of time periods in which the uncertain parameter can reach the boundary of the fluctuation range at most in the total scheduling period, and its value is 1, 2…T; and are the predicted power of photovoltaic output and conventional load respectively; T is a scheduling cycle. Among them, the robust adjustment coefficient can be used to adjust the conservativeness of the two-stage robust model. Decision makers can adjust the parameters according to their own risk preferences. The larger the robust adjustment coefficient, the more conservative the robust model.

[0080] Furthermore, the demand response flexible load (shiftable load) model in step S1 takes into account shiftable loads in addition to cluster electric vehicles. Shiftable loads have certain flexibility in the process of microgrid economic dispatch, and residents can change their electricity plans according to incentives to achieve the purpose of serving the microgrid. The adjustment of electricity plans is affected by residents' electricity demand, and the microgrid needs to give users certain compensation. Among them, the dispatch cost of shiftable loads is:

[0081]

[0082] Among them, K tr The unit compensation cost for the translatable load; is the expected electricity demand of residents in period t; P tr,t is the actual dispatch power of residents in period t; the absolute value represents the difference between the expected power demand and the actual dispatch power of the shiftable load, Δt is the minimum dispatch time scale, C tr,t Represents the dispatch cost of the shiftable load.

[0083] Furthermore, we introduce the auxiliary variable P tr1,t and P tr2,t Adding constraints can convert the dispatching cost of the shiftable load into a linear form:

[0084]

[0085] P tr1,t ≥0,P tr2,t ≥0

[0086] C tr,t =K tr [P tr1,t +P tr2,t ]

[0087] Among them, P tr1,t and P tr2,tare auxiliary variables, which respectively represent that the residents have two dispatch powers greater than or equal to 0 in time period t. In the constraint relationship, the sum of the two is equivalent to the absolute value of the difference between the expected electricity demand and the actual dispatch power of the shiftable load.

[0088] At the same time, the power consumption characteristics that the movable load needs to meet can be expressed as:

[0089]

[0090] Among them, D tr is the total power demand during the period when the load can be shifted to participate in the microgrid dispatch; P tr,t The actual dispatch power for residents in period t; and They represent the minimum and maximum electricity demands of the user in period t respectively; Δt represents the minimum scheduling period.

[0091] Furthermore, the two-stage three-layer robust optimization model established in step S2 aims to minimize the daily operation cost of the microgrid. When the photovoltaic output and the power of conventional load are predicted values, the deterministic problem is:

[0092]

[0093] Among them, C PG,t is the unit power generation cost of the gas turbine in period t; C EV,t is the charging and discharging cost of electric vehicles in period t; C tr,t is the dispatching cost of the shiftable load in period t, C ESS,t is the charging and discharging operating cost of the energy storage during period t; C M,t is the interaction power cost between the microgrid and the distribution network in period t.

[0094] More specifically, consider establishing a two-stage robust virtual power plant economic dispatch model to solve the dispatching scheme that makes the system operate at the lowest cost when photovoltaic, wind power and load are in the worst scenario. The pre-dispatching stage of the two-stage three-layer robust optimization model is a min-max-min model, which divides the decision variables into first-stage variables x and second-stage variables y. The first-stage variable x is made before the uncertain information is known, and the constraints are independent of the uncertain variables. The second-stage variable y is made after the uncertain information is known, and is also called the compensation variable. The core of the two-stage three-layer robust optimization model lies in the solution of the pre-dispatching stage. It is written in a compact form in the form of a matrix. The problem of the pre-dispatching stage can be described as:

[0095]

[0096]

[0097] Among them, cT x is the operating cost of the first stage, including the charging and discharging costs of cluster electric vehicles; d T y is the operating cost of the second stage, including the gas turbine power generation cost, the shiftable load dispatching cost, the energy storage facility operating cost and the microgrid distribution network interactive power cost; x is the decision variable of the outer min problem; u and y are the decision variables of the inner max-min problem. The purpose is to find the worst scenario that maximizes the minimum operating cost of the second stage based on the results of the first stage variable solution.

[0098] Among them, c is the coefficient column vector of the charging and discharging cost function of the cluster electric vehicles in the first stage; d is the coefficient column vector of the operating cost function in the second stage; Q, R, F, G, D, K, E, L, and Y are the coefficient matrices of each constraint variable; and m, g, f, and h are constant column vectors.

[0099] Objective function of the pre-scheduling phase The constraints considered include: gas turbine output constraints, energy storage facility charging and discharging and capacity constraints, electric vehicle charging and discharging power and capacity constraints, and microgrid distribution network interaction power constraints.

[0100] Furthermore, if Figure 2 As shown, step S3 includes the following sub-steps:

[0101] In step S31, the pre-scheduling phase problem in step S2 is solved by applying the idea of ​​zero-sum game using the column and constraint generation algorithm (C&CG), and the original problem is decoupled into a main problem and sub-problems for iteration.

[0102] Among them, the main problem (MP) obtained after the original problem is decomposed is:

[0103]

[0104] Where l is the number of iterations; θ is the intermediate variable introduced in the problem decomposition process; y l is the auxiliary variable related to the subproblem introduced at the lth iteration; is the worst scenario of photovoltaic and load obtained by solving the subproblem at the lth iteration, where is a known quantity.

[0105] Among them, the sub-problems (SP) obtained after the original problem is decomposed are:

[0106]

[0107] Among them, x * The first-stage variables obtained by solving the main problem are known variables here; V represents the coefficient matrix of the cluster electric vehicle charging and discharging constraint variables.

[0108] Among them, the subproblem is an NP-hard two-level optimization problem. When the first-stage variable x and the uncertain variable u are given, the inner layer min is a linear deterministic optimization problem. Using KKT or strong duality theory, it is converted into a max problem, which can be merged with the outer layer max problem to become a single-layer convex semidefinite optimization problem.

[0109] Step S32, use the big M method and introduce continuous auxiliary variables to linearize the sub-problems. Through the above transformation, the main problem and sub-problems decomposed by the two-stage three-layer robust optimization model based on the worst scenario are mixed integer linear programming (MILP). After continuous iterative solution, an economic dispatch plan that optimizes the operating cost of the microgrid under the worst scenario is obtained.

[0110] Therefore, the purpose of the two-stage robust optimization for the worst-case scenario is to obtain a set of first-stage solutions in the pre-scheduling process. Given this set of solutions, when the uncertain parameters take any value in the interval, the system constraints can be satisfied through the second-stage solutions, and the total operating cost of the system is the lowest in the worst-case scenario.

[0111] The present invention also provides a source-load-storage microgrid management system containing cluster electric vehicles for the worst scenarios, including: a distribution network 1, a public connection point 2, photovoltaics 3, a micro gas turbine unit 4, energy storage 5, various demand response flexible loads 6, cluster electric vehicles 7 and a microgrid central controller 8. The distribution network 1 is connected to the photovoltaics 3, the micro gas turbine unit 4, the energy storage 5, various demand response flexible loads 6, and the cluster electric vehicles 7 through the public connection point 2 to form a power flow; the microgrid central controller 8 is connected to the photovoltaics 3, the micro gas turbine unit 4, the energy storage 5, various demand response flexible loads 6, and the cluster electric vehicles 7 to form an information flow.

[0112] Among them, the microgrid central controller 8 uses a source-load-storage microgrid management method containing cluster electric vehicles for the worst scenario of the present invention. In the optimization problem stage, the objective function is determined to be the lowest daily operating cost of the microgrid, and the constraints include various equipment operation constraints; in the pre-dispatching stage, the minimum operating cost of the microgrid in the worst scenario is calculated; in the re-dispatching stage, the specific second-stage variables are solved by the decision information of the first stage and the real data of the known uncertain parameters, and the minimum re-dispatching cost is calculated, and finally a microgrid dispatching plan that comprehensively considers economy and reliability is obtained.

[0113] The following is a specific embodiment of the present invention to introduce a source-load-storage microgrid management method containing cluster electric vehicles for the worst scenario.

[0114] It should be noted that the scenario of this specific embodiment is as follows: in a source-load-storage microgrid containing cluster electric vehicles on a typical summer day in a certain city, the Jiangsu residential electricity tiered electricity price is used as the day-ahead transaction electricity price for power exchange between the distribution network and the load microgrid.

[0115] In this specific embodiment, the microgrid parameters are shown in Table 1 below. Taking 1 hour as a scheduling period, the optimal scheduling method with T=24 hours as the scheduling period is studied.

[0116] Table 1 Parameters of each unit of microgrid

[0117]

[0118]

[0119] In this specific embodiment, the process of implementing the source-load-storage microgrid management method of the present invention for the worst scenario containing cluster electric vehicles is as follows:

[0120] Based on the three typical scenarios 1, 2, and 3 obtained by reducing the scenarios based on the historical data of photovoltaic and load, it is first given that 100 electric vehicles participate in the economic dispatch of the microgrid, and the robust adjustment coefficients are 3, 6, 12, and 24, and the maximum fluctuation deviation is 15%. The robust adjustment coefficients of conventional loads are 6, 12, 24, and 24, and the maximum fluctuation deviation is 10%. The actual value of the uncertain information obtained by re-dispatching is the worst scenario. Based on the worst scenario, under different prediction errors, deterministic optimization and two-stage robust optimization are used to economically dispatch the microgrid, and the advantages and characteristics of the two-stage robust optimization and deterministic optimization methods in the optimization dispatch process are analyzed and compared to verify the effectiveness and economy of the proposed method. Among them, the three typical scenarios obtained after scenario reduction are specifically manifested in the different power of photovoltaic and load, and the main feature is the different weather. Scenario 1 is a representative scene of sunny days, scenario 2 is a representative scene of cloudy days, and scenario 3 is a representative scene of rainy days.

[0121] The implementation results are shown in Table 2 below. It can be seen from the operating cost comparison results in Table 2 that within the scheduling period, when the prediction accuracy is low, the method provided by the present invention is more economical than the total operating cost of deterministic optimization. Although the day-ahead scheduling scheme obtained by the robust optimization method has a higher day-ahead operating cost (￥18009.4), because it takes into account the uncertainty of load power and photovoltaic output, the balancing cost (￥1143.5) caused by the prediction error in the real-time market is lower than the balancing cost (￥5117.4) of the deterministic optimization method. With the increase of the prediction error (the prediction error value is 1.5), the balancing cost (￥12220.5) of the deterministic optimization method increases significantly. Although the balancing cost corresponding to the day-ahead scheduling scheme obtained by the robust optimization method has also increased, its total operating cost (￥23058.2) is still lower than that of the deterministic optimization method, and the effect of reducing the operating cost is more obvious. During the scheduling cycle, when the prediction accuracy is high (the prediction error is 0.5), the scheduling scheme obtained by the deterministic optimization method is closer to the actual operation due to the small deviation between the prediction curve and the actual curve, thereby reducing the corresponding balancing cost. At this time, the robust optimization method is too "conservative" because the day-ahead scheduling scheme it formulates is too conservative, and there is surplus power sold at a lower price in the real-time market, which makes the total operating cost higher than the deterministic optimization method. In summary, the advantages of the two-stage robust optimization method over the deterministic optimization method become more significant as the prediction error increases, which shows the importance of considering uncertainty when optimizing scheduling, and the good applicability of the two-stage robust optimization in solving such problems.

[0122] Table 2 Performance comparison of optimization methods under different prediction errors

[0123]

[0124]

[0125] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A source-load-storage microgrid management method containing cluster electric vehicles for the worst scenario, characterized in that: include, Step S1, in a source-load-storage microgrid containing a cluster of electric vehicles, establishing an operation constraint and coordinated control model including an electric vehicle cluster, energy storage, a gas turbine unit, and a photovoltaic unit; Step S2, establish a two-stage three-layer robust optimization model with a min-max-min structure, divide the decision variables into the first-stage variable x and the second-stage variable y, and the two-stage three-layer robust optimization model takes the minimum daily operation cost of the microgrid as the optimization goal, and takes the operation constraints and coordinated control model in step S1 as constraints; Step S3, using the column and constraint generation algorithm, decomposes the original problem in step S2 into a main problem and sub-problems and solves them iteratively.

2. A source-load-storage microgrid management method for the worst scenario containing cluster electric vehicles as claimed in claim 1, characterized in that: The components of the coordinated control model include a micro gas turbine power model, which is formulated as follows: C PG,t =k G P G,t Δt Among them, k G P is the comprehensive cost factor considering the fuel cost and maintenance cost of the gas turbine; G,t is the output power of the gas turbine during period t; Δt is the scheduling step length; C PG,t is the unit power generation cost of the gas turbine in period t.

3. A source-load-storage microgrid management method for the worst scenario containing cluster electric vehicles as claimed in claim 2, characterized in that: The components of the coordinated control model include the charging and discharging power and capacity model of the energy storage battery ESS, which is expressed as follows: Among them, W b is the capacity of the energy storage battery; subscript b represents microgrid park b; subscript t represents time period t; T is the total time period in the dispatching cycle; is the energy level of the energy storage battery in park b during period t, is the energy level of the energy storage battery in park b at time t-1; and are the charging power and discharging power of ESS in park b during period t; η ess,ch With η ess,dis They are the charging efficiency and discharging efficiency of ESS respectively; and They are charging power limit and discharging power limit respectively; A variable indicating whether the energy storage battery is in a charging state, 1 if it is, and 0 if it is not; A variable indicating whether the energy storage battery is in a discharge state, 1 if it is, and 0 if it is not; and They are the minimum state of charge and maximum state of charge of ESS respectively; and They represent the charge states at time periods t=0 and t=T in the ESS scheduling period respectively.

4. A source-load-storage microgrid management method for the worst scenario containing cluster electric vehicles as claimed in claim 3, characterized in that: The components of the coordinated control model include a power energy boundary model of the cluster electric vehicles time shift, which is expressed as follows: Among them, E i is the battery capacity of the i-th electric vehicle; is the energy level of the battery of the i-th electric vehicle at time t and time t-1; is the power demand level of the i-th electric vehicle; SOC EV,min and SOC EV,max They are the upper and lower limits of battery capacity respectively; Respectively represent the charging state and discharging state of the i-th electric vehicle, which are 0-1 variables; 1 indicates that the i-th electric vehicle is in the charging state, otherwise, it is not; 1 indicates that the i-th electric vehicle is in the discharging state, otherwise, it is not in the discharging state; and are the battery charging power and discharging power of electric vehicle i during period t respectively; and They are respectively the battery charging power limit and the discharging power limit; η EV,ch With η EV,dis They are the battery charging efficiency and discharging efficiency respectively; They represent the charging state and discharging state of the i-th electric vehicle in park b respectively; Indicates that the i-th electric car is at t a The energy level of the battery during the period.

5. A source-load-storage microgrid management method for the worst scenario containing cluster electric vehicles as claimed in claim 4, characterized in that: The components of the coordinated control model include uncertainty models of photovoltaic and load, which are expressed as follows: u=[u PV,t u L,t ] T ,t=1,2…T Where u is the output of photovoltaic and load in the uncertainty model; u PV,t and u L,t are photovoltaic output and conventional load power respectively; ν PV,t Indicates whether the uncertainty of photovoltaic power in this period reaches the boundary of the fluctuation range, if yes, it is 1, otherwise, it is 0; ν L,t Indicates whether the uncertainty of the load in this period reaches the boundary of the fluctuation range, if yes, it is 1, if not, it is 0; and are the maximum deviation of photovoltaic fluctuation and the maximum deviation of load power fluctuation respectively; Γ PV is the robust regulation coefficient of photovoltaics; Γ L is the robust regulation coefficient of the load; and are the predicted power of photovoltaic output and conventional load respectively; T is a scheduling cycle.

6. A source-load-storage microgrid management method for the worst scenario containing cluster electric vehicles as claimed in claim 5, characterized in that: The components of the coordinated control model include a shiftable load model, and the dispatching cost of the shiftable load is: Among them, K tr The unit compensation cost for the translatable load; is the expected electricity demand of residents in period t; P tr,t is the actual dispatch power of residents in period t; the absolute value represents the difference between the expected power demand and the actual dispatch power of the shiftable load, Δt is the minimum dispatch time scale, C tr,t represents the dispatch cost of the shiftable load; Introduce auxiliary variable P tr1,t and P tr2,t And add constraints to convert the dispatch cost of the shiftable load into a linear form: P tr1,t ≥0,P tr2,t ≥0 C tr,t =K tr [P tr1,t +P tr2,t ] Among them, P tr1,t and P tr2,t are auxiliary variables, which represent that the residents have two dispatch powers greater than or equal to 0 in time period t. In the constraint relationship, the sum of the two is equivalent to the absolute value of the difference between the expected electricity demand and the actual dispatch power of the shiftable load; The power consumption characteristics satisfied by the movable load are expressed as: Among them, D tr is the total power demand during the period when the load can be shifted to participate in the microgrid dispatch; P tr,t The actual dispatch power for residents in period t; and They represent the minimum and maximum electricity demands of the user in period t respectively; Δt represents the minimum scheduling period.

7. A source-load-storage microgrid management method for the worst-case scenario containing clustered electric vehicles as claimed in claim 6, characterized in that: In step S2, the deterministic problem is: Among them, C PG,t is the unit power generation cost of the gas turbine in period t; C EV,t is the charging and discharging cost of electric vehicles in period t; C tr,t is the dispatching cost of the shiftable load in period t, C ESS,t C is the charging and discharging operating cost of the energy storage during period t; M,t is the interaction power cost between the microgrid and the distribution network in period t; The pre-scheduling phase problem is described as: Among them, c T x is the operating cost of the first stage, including the charging and discharging costs of cluster electric vehicles; d T y is the operating cost of the second stage, including the gas turbine power generation cost, the shiftable load scheduling cost, the energy storage facility operation cost and the microgrid distribution network interaction power cost; x is the decision variable of the outer min problem; u and y are the decision variables of the inner max-min problem, c is the coefficient column vector of the cluster electric vehicle charging and discharging cost function in the first stage; d is the coefficient column vector of the operating cost function in the second stage; Q, R, F, G, D, K, E, L, Y are the coefficient matrices of each constraint variable; m, g, f, h are constant column vectors.

8. A source-load-storage microgrid management method for the worst-case scenario containing clustered electric vehicles as claimed in claim 7, characterized in that: The step S3 comprises: Step S31, applying the idea of ​​zero-sum game to the pre-scheduling phase problem in step S2, using the column and constraint generation algorithm to solve it, decoupling the original problem into a main problem and sub-problems for iteration; Among them, the main problem after decomposition of the original problem is: Where l is the number of iterations; θ is the intermediate variable introduced in the problem decomposition process; y l is the auxiliary variable related to the subproblem introduced at the lth iteration; The worst scenario of photovoltaic and load obtained by solving the sub-problem at the lth iteration; Among them, the sub-problems obtained after the original problem is decomposed are: Among them, x * The first-stage variables obtained by solving the main problem; V represents the coefficient matrix of the cluster electric vehicle charging and discharging constraint variables; Using KKT or strong duality theory, the subproblem is transformed into a max problem.

9. A source-load-storage microgrid management method for the worst scenario containing cluster electric vehicles as claimed in claim 8, characterized in that: The step S3 comprises: Step S32, use the big M method and introduce continuous auxiliary variables to linearize the sub-problems. Through the above transformation, the main problem and sub-problems decomposed by the two-stage three-layer robust optimization model based on the worst scenario are mixed integer linear programming. After continuous iterative solution, an economic dispatch plan that optimizes the operating cost of the microgrid under the worst scenario is obtained.

10. A source-load-storage microgrid management system containing cluster electric vehicles for the worst scenario, characterized in that: It includes a distribution network, photovoltaics, micro gas turbine units, energy storage, various types of demand response flexible loads, cluster electric vehicles and a microgrid central controller. The distribution network is power-connected to the photovoltaics, the micro gas turbine units, the energy storage, the various types of demand response flexible loads and the cluster electric vehicles; the microgrid central controller is communication-connected to the photovoltaics, the micro gas turbine units, the energy storage, the various types of demand response flexible loads and the cluster electric vehicles. The microgrid central controller executes a source-load-storage microgrid management method containing cluster electric vehicles for the worst scenario as described in any one of claims 1 to 9.

Images