Virtual power plant optimization scheduling method considering controllable charging and discharging of electric vehicle

Abstract

The invention discloses a virtual power plant optimal scheduling method considering controllable charging and discharging of an electric vehicle. The method comprises the following steps: establishing a VPP power grid interaction architecture and a VPP operation mode; classifying the states of the electric vehicles after grid connection; classifying the electric vehicles according to the state vector of the single electric vehicle after the single electric vehicle enters the network; the controllability of the EVG is calculated; the invention discloses a VPP two-stage robust optimization scheduling model. According to the invention, state classification is carried out on a single electric vehicle according to a state vector when the single electric vehicle accesses the network, and a controllability calculation method of an electric vehicle group is accurately given according to a mincoski addition idea; operation constraint and coordination control are comprehensively considered, and a two-stage robust optimization problem is decomposed into a main problem and a sub-problem for iterative solution; a controllable capacity curve of the EVG, an optimal scheduling scheme of the VPP and an optimal capacity allocation scheme of the ESS under different controllable proportionality coefficients are given. And comparing the performances of the two-stage robust optimization method, the single-stage robust optimization method and the deterministic optimization method under different prediction deviation coefficients.

Description

Technical Field

[0001] The present invention relates to a scheduling method, and in particular to a virtual power plant optimization scheduling method taking into account controllable charging and discharging of electric vehicles. Background Art

[0002] In recent years, renewable energy has been vigorously promoted due to its great advantages in alleviating energy shortages and reducing carbon emissions. However, due to the large scale and dispersion of distributed energy and the high uncertainty of output power, the stable operation of the power grid will be affected in the absence of effective coordination and remote control. The emergence of virtual power plants (VPPs) can solve this problem well. VPPs can realize the aggregation of distributed energy resources (DERs) such as distributed power sources, energy storage systems (ESS), wind power (WPP), photovoltaic generators (PV), electric vehicles (EVs) in different regions through advanced information and communication technologies and coordinated optimization, and participate in the coordinated management of power grid operation as a whole as a power plant. There is a lack of methods to optimize the scheduling of virtual power plants by considering the uncertainty of electric vehicles and renewable energy in VPPs and controllability.

[0003] Technical solution of the prior art 1: In response to the economic dispatch problem of VPP, Zhang et al. proposed a master-slave game model, taking VPP as an e-commerce retailer to participate in the charging management of electric vehicles, realizing the energy complementarity between DER and electric vehicles, and improving the overall economy. Yang et al. and Sun et al. constructed a bidding model for VPPs including electric vehicles and WPPs to participate in the power market, realizing resource optimization scheduling considering various uncertainties. Ju et al. constructed a day-ahead robust optimization dispatch model for VPPs including electric vehicles, WPPs, gas turbines and other DERs, improving the operating efficiency of the system. Wang et al. regarded loads such as electric vehicles and air-conditioning loads as generalized energy storage and realized the three-stage coordinated optimization dispatch of VPPs. Although the above literature has made some progress in the research of dispatch methods and optimization strategies containing VPPs, the modeling of electric vehicle groups (EVGs) in VPPs is too idealized, and the impact of the inherent constraints between multiple types of electric vehicles on the feasibility and economy of VPP dispatch schemes is not considered. In response to this problem, Zhan et al. responded to the accuracy of operator decisions by evaluating the controllable capacity of EVGs, but they did not consider the impact of the number of electric vehicles of various types and the changes in the parameters of electric vehicle entry states on the controllable capacity of EVGs, so the dispatch scheme obtained may still be difficult to implement. Based on the above literature, considering that large-scale and multi-type electric vehicles will participate in grid operation through VPP in the future, it is of great significance to accurately calculate the controlled capacity of EVG and introduce it into the dispatch optimization of VPP.

[0004] Technical solution of the second prior art: Since the fluctuation of WPP and PV output will affect the stability of VPP operation, considering the controllability of electric vehicles, in the study of VPP economic dispatch problem, how to effectively deal with the impact of uncertainty of renewable energy output on the stable operation of the system is also very important. Talari et al. and Tajeddini et al. deal with the uncertainty of wind and light through random programming and simulate uncertain parameters through finite scenarios, but random programming cannot truly reflect the variable conditions in certain scenarios. Li et al. established a VPP model based on the conditional risk value theory, which reduced the system risk, but there is still decision-making risk. Zhang et al. introduced the wind abandonment penalty coefficient in the upper and lower objective functions of the VPP dispatch model, but essentially did not deal with the fluctuation problem of WPP and PV output. Compared with the above methods, Wang et al. used a random robust optimization method to deal with the randomness of wind power and photovoltaic output, but the calculated VPP dispatch scheme was too conservative. Xu et al. [established a robust optimization model with a minimum-maximum-minimum structure to deal with the uncertainty of wind and photovoltaic output, but could not adjust the conservatism of the VPP dispatch scheme. Liu et al. further established a two-stage robust optimization method, replacing random variables with uncertain sets and introducing adjustment parameters to adjust the robustness of the scheduling scheme.

[0005] In order to improve the level of new energy consumption, it is necessary to give more consideration to the role of electric vehicles in the complementary and mutual assistance of electric energy between distribution network areas. Electric vehicles have high flexibility and rapid response capabilities, and can absorb excess new energy in the system through effective scheduling. However, the above-mentioned optimization scheduling methods of the prior art cannot effectively solve the uncertainty of electric vehicles and renewable energy. Summary of the invention

[0006] The goal of this invention is to deal with the uncertainty of electric vehicles and renewable energy in a virtual power plant (VPP), and propose a two-stage robust virtual power plant optimization scheduling method considering the controllability of electric vehicle groups (EVGs). The technical scheme is as follows:

[0007] A virtual power plant optimization scheduling method considering controllable charging and discharging of electric vehicles, the method comprising the following steps:

[0008] Step (1): Clarify the VPP grid interaction architecture and VPP operation model. VPP plans the purchase and sale of electricity based on various factors and submits it to ISO. In the day-ahead energy market, the electricity that exceeds the load will not be settled and will be fined, and the electricity that is below the load will be sold at the purchase price. In the real-time market, VPP reduces power deviation by adjusting internal resources. Uncontrollable excess power generation will not be cleared and will be fined more heavily, and the missing power will be sold at the real-time market purchase price.

[0009] Step (2): Classify the state of electric vehicles after they are connected to the grid. Before giving the calculation method of EVG controllability, electric vehicles should be classified according to the state vector of a single electric vehicle after it enters the network.

[0010] Step (3): Calculation of EVG controllability The variables of the electric vehicle are processed by summing the variables using the Minkowski method to obtain the controllability of the electric generator.

[0011] Step (4): VPP two-stage robust optimization scheduling model. In order to deal with the risks brought to the system by WT and PV output uncertainty, this paper adopts a two-stage robust optimization method to solve the optimal scheduling scheme of VPP.

[0012] Beneficial Effects

[0013] (1) The state of a single electric vehicle is classified according to its state vector when it enters the grid, and then the controllability calculation method of the electric vehicle group is accurately given based on the Minkowski addition idea.

[0014] (2) The operation constraints and coordinated control of CGT, ESS, EVG, WPP and PV are comprehensively considered. Based on the column and constraint generation algorithm (C&CG), the two-stage robust optimization problem is decomposed into a main problem and sub-problems for iterative solution.

[0015] (3) The controllable capacity curve of EVG, the optimal dispatching scheme of VPP and the optimal capacity allocation scheme of ESS under different controllable proportional coefficients are given. In addition, the performance of two-stage robust optimization method, single-stage robust optimization method and deterministic optimization method under different prediction deviation coefficients are compared. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a schematic diagram of the optimization scheduling framework of the present invention;

[0017] Figure 2 A schematic diagram of the interaction framework between the virtual power plant of the present invention, electric vehicles and the power grid;

[0018] Figure 3 It is a schematic diagram of the schedulable time period of the charging and discharging behavior of an electric vehicle under a fully controlled state of the present invention;

[0019] Figure 4 A schematic diagram of the schedulable time periods for charging and discharging behaviors of electric vehicles in a semi-controlled state of the present invention;

[0020] Figure 5 A schematic diagram of the schedulable time periods for charging and discharging behaviors of an electric vehicle in an out-of-control state of the present invention;

[0021] Figure 6 It is the charge and discharge power boundary curve diagram of EVG under different controlled proportional coefficients;

[0022] Figure 7 This is the VPP operating profit curve under EVG controlled capacity. DETAILED DESCRIPTION

[0023] A virtual power plant optimization scheduling method considering controllable charging and discharging of electric vehicles, the method comprising the following steps:

[0024] Step (1): Establish the VPP grid interaction architecture and VPP operation model. VPP plans the purchase and sale of electricity based on various factors and submits it to ISO. In the day-ahead energy market, the electricity that exceeds the load will not be settled and will be fined, and the electricity that is below the load will be sold at the purchase price. In the real-time market, VPP reduces power deviation by adjusting internal resources. The uncontrollable excess power generation will not be cleared and will be fined more heavily, and the missing power will be sold at the real-time market purchase price.

[0025] The interaction architecture of VPP, EVG and power grid is as follows:

[0026] The VPP grid interaction architecture mainly consists of two layers: independent system operator (ISO) and virtual power plant. The virtual power plant aggregates distributed energy resources (DERs) such as distributed power sources, energy storage systems (ESS), wind power (WPP), photovoltaic generators (PV), and electric vehicles (EV) in different regions into a whole, and directly participates in the coordinated management of the power grid through transactions with ISO.

[0027] The operating modes of VPP and EVG are:

[0028] Assuming that VPP participates in the day-ahead energy market and the real-time power balance market as a price receiver, the operation mode of VPP accepting ISO dispatch and participating in the power market is as follows: before the end of the energy market on the same day (12:00 am), the operator needs to coordinate internal resources according to the controllable capacity curve of EVG, the predicted output of wind power and photovoltaic units, and the load demand curve issued by ISO, and submit the power purchase and sale plan and dispatch plan for each period of the next day to ISO. In the day-ahead energy market, ISO will not settle and impose a fine on the surplus power of VPP that exceeds the load demand, and the deficit power that is lower than the load demand will be sold at the power purchase price in the energy market. In the subsequent real-time market, in order to cope with the output fluctuations of renewable energy units, VPP will give priority to adjusting internal controllable resources to reduce power deviation. However, for the excess power generation that VPP still cannot suppress, ISO will not clear it and impose a heavier fine than the energy market, and sell the missing power at the real-time market power purchase price. In addition, after each dispatch cycle, ISO should conduct a deviation assessment on the actual power generation and power consumption of VPP to improve the VPP's level of absorption of wind and solar energy resources.

[0029] Step (2): Classify the state of electric vehicles after they are connected to the grid. Before giving the calculation method of EVG controllability, electric vehicles should be classified according to the state vector of a single electric vehicle after it enters the network.

[0030] The state vector of electric vehicles after entering the network can be divided into the following three categories:

[0031] 1) Category I National Electric Vehicle

[0032] Electric vehicles in Class I state have enough time to access the grid and discharge the remaining power on the grid to the minimum power, and then charge to the off-grid demand before leaving the grid. VPP has enough time to flexibly arrange the charging and discharging behavior of these electric vehicles. Then assume that the definition domain of the electric vehicle network access time is shown in formula (1).

[0033]

[0034] In the formula, Refers to the network access time of the jth EV in EVG

[0035] The charging time required for electric vehicles to meet off-grid demand is shown in formula (2):

[0036]

[0037] In the formula, It refers to the SOC of the jth electric vehicle when it is off-grid. Refers to the minimum value of the electric vehicle SOC, P c The upper limit of charging and discharging power of electric vehicles.

[0038] The time required for an electric vehicle to discharge from the remaining power of the grid to the minimum power value is shown in formula (3).

[0039]

[0040] In the formula, It refers to the SOC when the jth electric vehicle is connected to the network.

[0041] The grid connection time of electric vehicles can be expressed as formula (4).

[0042]

[0043] In the formula It refers to the off-grid time of the jth electric vehicle in EVG.

[0044] In summary, the grid-connected time of electric vehicles in Class I state should satisfy formula (5).

[0045]

[0046] like Figure 3 As shown, VPP can be included in Any time period, with constant power P c Dispatching the charging and discharging behavior of electric vehicles in Class I state.

[0047] 2) Class II electric vehicles

[0048] Due to the restrictions of grid connection time and off-grid demand, electric vehicles in Class II state do not have enough time to fully discharge to the minimum power. Therefore, VPP can only give priority to allowing them to discharge as much as possible and switch such electric vehicles from the discharge state to the charging state within a certain period of time until the off-grid time. In this state, the moment when the electric vehicle switches from the discharge state to the charging state is shown in formula (6).

[0049]

[0050] In the formula, Refers to the charging efficiency of electric vehicles. Refers to the discharge efficiency of electric vehicles.

[0051] The shortest charging time required for an electric vehicle to meet its travel needs is shown in formula (7).

[0052]

[0053] In conclusion, the grid-connected time of electric vehicles in the second state should satisfy formula (8).

[0054]

[0055] like Figure 4 As shown, VPP can be Select an arbitrary time period within the time frame, and schedule the discharge of electric vehicles in the second state with a constant power Pc, and These electric vehicles are charged off-grid at a constant power Pc.

[0056] 3) Electric vehicles in Category III

[0057] The grid connection time of electric vehicles in Category III state is too short to even meet their own basic charging needs, so VPP cannot schedule the charging and discharging behavior of such electric vehicles. In this state, after the electric vehicle is connected to the grid, VPP must charge it immediately to meet its charging needs to the maximum extent.

[0058] In conclusion, the grid connection time of electric vehicles in Class III state should satisfy equation (9).

[0059]

[0060] like Figure 5 As shown, in Within the scope, VPP cannot freely dispatch the charging and discharging of electric vehicles in Category III status, which need to be charged at a constant power Pc until they are off the grid after being connected to the grid.

[0061] Step (3): Calculation of EVG controllability The variables of the electric vehicle are processed by summing the variables using the Minkowski method to obtain the controllability of the electric generator.

[0062] Minkowski method

[0063] Minkowski summation is a kind of addition applicable to Euclidean space, and its expression is shown in formula (10). Its physical essence is the expansion set of multiple spaces. After calculating the Minkowski sum of two variable spaces, the envelope of the variable sum can be obtained. Based on this idea, the electric vehicle group can be aggregated into a generalized energy storage device.

[0064]

[0065] Where: A and B are two variable spaces; a and b are elements of A and B respectively; represents the Minkowski sum of A and B. The premise of Minkowski summation is that the two variable spaces have the same domain. However, due to the differences in the grid-connected time of individual electric vehicles, their domain T EVn is heterogeneous. Therefore, it is necessary to extend it to the same scheduling time set T, thereby obtaining the generalized energy storage model shown in formula (11):

[0066]

[0067] Based on the EV classification results, the Minkowski method is summed for all EV variables to obtain the decision envelope space consisting of all possible charging and discharging behaviors of EVG. This envelope space is the controllability of the electric generator. Details on the Minkowski addition calculation principle. After the above processing, it is possible to avoid modeling each EV separately. This method not only considers the network access state vector of a single EV, retains the constraint relationship between a single EV and EVG, but also accurately calculates the controllability of EVG.

[0068] On this basis, the charging and discharging power and electric power boundary values ​​of electric vehicles under three states are given below, as shown in equations (10)-(15).

[0069] The charging power boundary of electric vehicles in Class I state is:

[0070]

[0071] The value limit of the discharge power is the same as the value limit of the charge power.

[0072] The charging power boundary of electric vehicles in the second state is

[0073]

[0074] The discharge power boundary of electric vehicles in Class II state is

[0075]

[0076] The charging power boundary of the third type of electric vehicle is the same as that of the first type of electric vehicle and is not given here. The discharge power boundary of this type of electric vehicle is

[0077]

[0078] In any state, the boundary of the electric energy value during the charging and discharging process of an electric vehicle is expressed as

[0079]

[0080]

[0081] Since the Minkowski summation method is based on the premise that the domains of all variables are the same, it is necessary to extend the domains of all EV variables to the entire scheduling period, as shown in Equation (16).

[0082] T={t|t=1,2,…,24} (16)

[0083] Then, the mathematical modeling variables of EVG can be obtained, as shown in formula (17).

[0084]

[0085] is the charging power of the jth electric vehicle in EVG, It refers to the discharge power of the jth electric vehicle in EVG.

[0086] The constraint boundary of the feasible region of electric vehicle charging and discharging power is shown in formula (18).

[0087]

[0088] Step (4): VPP two-stage robust optimization scheduling model. In order to deal with the risks brought to the system by WT and PV output uncertainty, a two-stage robust optimization method is used to solve the optimal scheduling scheme of VPP.

[0089] 1) Objective function

[0090] Based on the electricity price information released by ISO, load demand curve, and the predicted output of WPP and PV, VPP takes maximizing operating income as the optimization goal, integrates internal resources, and provides a day-ahead dispatching plan. The income of VPP can be expressed as formula (20):

[0091]

[0092] In the formula, R CGT,t Refers to the operating income of CGT at time t, R W,t refers to WPP’s operating income at time t, R PV,t Refers to the operating income of PV at time t, R EV,t refers to the operating income of EV at time t, R ESS,t Refers to the operating income of ESS at time t, ΔR t It refers to the cost of purchasing and disposing electricity at the VPP in the DAM at time t;

[0093] 2) Constraints

[0094] ①CGT constraints

[0095]

[0096] In the formula, The minimum output of the i-th CGT at time t is: The maximum output of the i-th CGT at time t, U CGT,i,t It is a binary variable, indicating whether CGT i is running at time t. If yes, it takes 1, otherwise it takes 0

[0097] ②EVG constraints

[0098]

[0099]

[0100]

[0101]

[0102] Formulas (21)-(22) are the upper and lower limits of the electric vehicle charging and discharging power; Formulas (23)-(24) represent the constraint relationship between the electric vehicle power and the charging and discharging power;

[0103] ③ Energy storage system constraints

[0104]

[0105] ④Power balance constraints

[0106]

[0107] Example

[0108] A virtual power plant optimization scheduling method taking into account the controllability of electric vehicles, the method comprising the following steps:

[0109] Step (1): Clarify the VPP grid interaction architecture and VPP operation model. VPP plans the purchase and sale of electricity based on various factors and submits it to ISO. In the day-ahead energy market, the electricity that exceeds the load will not be settled and will be fined, and the electricity that is below the load will be sold at the purchase price. In the real-time market, VPP reduces power deviation by adjusting internal resources. Uncontrollable excess power generation will not be cleared and will be fined more heavily, and the missing power will be sold at the real-time market purchase price.

[0110] In order to verify the effectiveness of the model and solution algorithm proposed in this paper, a VPP system is taken as an example. The scheduling period of the VPP studied in this paper is 1 day, with a total of 24 periods, each period is 1 hour. Considering that the EVG includes a photovoltaic unit with a maximum power of 4.5MW, a wind turbine with a maximum power of 6.5MW, three gas turbines with maximum powers of 5.67, 4.5 and 3.515MW, a storage device with a maximum capacity of 2.5, and 1,500 electric vehicles in the VPP, the forecast deviation of the WPP and photovoltaic power output is 20%. In addition, in order to prevent VPP from arbitrage from the electricity market, the purchase price of VPP is 1.2 times the sales price; the penalty coefficient for power abandonment is 1.3.

[0111] Step (2): Classify the state of electric vehicles after they are connected to the grid. Before giving the calculation method of EVG controllability, electric vehicles should be classified according to the state vector of a single electric vehicle after it enters the network.

[0112] Step (3): Calculation of EVG controllability The variables of the electric vehicle are processed by summing the variables using the Minkowski method to obtain the controllability of the electric generator.

[0113] Here we assume that the vector The ratio of EVs in the first, second, and third states in the EVG is called the controlled ratio factor. This section analyzes the controllability of the EVG at several ratios ζ = [0.6, 0.3, 0.1], [0.4, 0.5, 0.1], [0.2, 0.7, 0.1], [0.4, 0.3, 0.3], [0.2, 0.5, 0.3], [0.2, 0.3, 0.5].

[0114] exist Figure 6 In, when When constant, as The increase and The difference between the EVG charge and discharge power boundary values ​​in the same time period will increase as the power consumption decreases. This is because when electric vehicles in Class I state enter the power grid, the EVG charging power boundary and discharge power boundary increase in parallel, but as electric vehicles in Class II and Class III states gradually enter the power grid, and these electric vehicles enter the power grid relatively late and leave the power grid early, this makes it impossible for VPP to fully dispatch their discharge behavior, which ultimately causes the EVG discharge power boundary to drop faster than the charging power boundary in the same time period.

[0115] when is a constant, as The increase and As the value decreases, the peak values ​​of the charging power boundary and the discharging power boundary of EVG in the same time period become larger and the peak time becomes closer. This is because the more electric vehicles in the Class I state, the more electric vehicles can participate in the VPP charging and discharging, and the closer the change trend of the EVG charging and discharging power boundary is. constant, Increase and Compared with the case where the power consumption of EVG is reduced, the difference between the charge and discharge power boundaries of EVG changes less during the same period. This is because EVs in Class II can still provide partial discharge power and can still accept VPP discharge scheduling, while EVs in Class III cannot accept discharge scheduling.

[0116] Step (4): VPP two-stage robust optimization scheduling model. In order to deal with the risks brought to the system by WT and PV output uncertainty, this paper adopts a two-stage robust optimization method to solve the optimal scheduling scheme of VPP.

[0117] because Assume that the step size of the three parameters is 0.1. The 66 combinations of EVG controlled capacity are used to further study the impact of VPP operating income. The simulation results are as follows Figure 7 As shown. Figure 7 In, when When increasing from 0 to 1, the overall VPP return rate decreases in a "step-like" manner. If c increases by 0.1, the 66 groups of controlled scaling factors can be divided into 11 groups. For each group of controlled scaling factors, when When it remains unchanged, The reduction and As the number of EVG increases, the revenue of VPP tends to decrease. This is because compared with the second type of electric vehicles, the first type of electric vehicles can provide more discharge power and more flexible discharge time, which allows VPP to increase revenue by dispatching these electric vehicles during the "peak price" period. Therefore, the more the first type of electric vehicles in the EVG, the more significant the increase in VPP. However, it should be noted that when When the number of Class III EVs increases, the upper and lower limits of VPP operating income will drop rapidly, which means that when there are too many Class III EVs in the EVG, the operating income of VPP will be significantly reduced. Therefore, VPP should minimize the dispatch of such EVs to maintain stable income.

[0118] Table 1 Day-ahead operating benefits of VPP under uncertain adjustment parameters

[0119]

[0120] In order to verify the effectiveness of the scheduling scheme formulated by the two-stage robust optimization method, Table 1 above lists the operating benefits of VPP under the two-stage robust optimization, single-stage robust optimization and deterministic optimization methods. In Table 1, when the uncertainty adjustment parameters are all zero, the VPP operating benefits obtained by the two-stage robust optimization method are the same as those obtained by the deterministic optimization method, while the single-stage robust optimization method obtains the lowest benefits. As the uncertainty parameters increase, the benefits obtained by the robust optimization method become lower and lower, and the benefits obtained by the single-stage robust optimization method are even lower. This is because when the operator formulates the day-ahead plan through the robust optimization method, it takes into account the uncertainty of WPP and PV output, thereby formulating a more conservative plan than the deterministic optimization method. In addition, since the single-stage robust optimization is based on the worst case, the resulting scheduling scheme is more conservative. In contrast, the two-stage robust optimization makes the second-stage decision after the first-stage decision is made and some uncertainties appear, so the scheduling scheme is less conservative and more efficient.

[0121] In order to verify the effectiveness of the model and solution algorithm proposed in this paper, a VPP system is taken as an example. The scheduling period of the VPP studied in this paper is 1 day, with a total of 24 periods, each period is 1 hour. Considering that the EVG includes a photovoltaic unit with a maximum power of 4.5MW, a wind turbine with a maximum power of 6.5MW, three gas turbines with maximum powers of 5.67, 4.5 and 3.515MW, a storage device with a maximum capacity of 2.5, and 1,500 electric vehicles in the VPP, the forecast deviation of the WPP and photovoltaic power output is 20%. In addition, in order to prevent VPP from arbitrage from the electricity market, the purchase price of VPP is 1.2 times the sales price; the penalty coefficient for power abandonment is 1.3.

[0122] In order to verify the effectiveness of the scheduling scheme formulated by the two-stage robust optimization method, the following Table 1 lists the operating benefits of VPP under the two-stage robust optimization, single-stage robust optimization and deterministic optimization methods. In Table 1, when the uncertainty adjustment parameters are all zero, the VPP operating benefits obtained by the two-stage robust optimization method are the same as those obtained by the deterministic optimization method, while the single-stage robust optimization method obtains the lowest benefits. As the uncertainty parameters increase, the benefits obtained by the robust optimization method become lower and lower, and the benefits obtained by the single-stage robust optimization method are even lower. This is because when the operator formulates the day-ahead plan through the robust optimization method, it takes into account the uncertainty of WPP and PV output, thereby formulating a more conservative plan than the deterministic optimization method. In addition, since the single-stage robust optimization is based on the worst case, the resulting scheduling scheme is more conservative. In contrast, the two-stage robust optimization makes the second-stage decision after the first-stage decision is made and some uncertainties appear, so the scheduling scheme is less conservative and more efficient.

[0123] Table 1 Day-ahead operating benefits of VPP under uncertain adjustment parameters

[0124]

[0125]

[0126] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.

Claims

1. A virtual power plant optimization scheduling method considering the controllable charging and discharging of electric vehicles, characterized by: Step (1): Establish the VPP grid interaction architecture and VPP operation model: VPP plans the purchase and sale of electricity based on various factors and submits it to ISO. In the day-ahead energy market, the electricity that exceeds the load will not be settled and will be fined, while the electricity that is below the load will be sold at the purchase price. In the real-time market, VPP reduces power deviation by adjusting internal resources. Uncontrollable excess power generation will not be cleared and will be fined more heavily. At the same time, the missing electricity will be sold at the real-time market purchase price. Step (2): Classify the state of electric vehicles after they are connected to the grid: Classify electric vehicles according to the state vector of a single electric vehicle after it enters the network; Step (3): Calculation of EVG controllability: The variables of the electric vehicle are processed by summing the variables using the Minkowski method to obtain the controllability of the electric generator; Step (4): VPP two-stage robust optimization scheduling model. In order to deal with the risks brought to the system by the uncertainty of WT and PV output, a two-stage robust optimization method is used to solve the optimal scheduling of VPP.

2. The virtual power plant optimization scheduling method considering the controllable charging and discharging of electric vehicles according to claim 1 is characterized by: The architecture of the VPP grid interaction architecture is: The VPP grid interaction architecture includes two layers: independent system operator ISO and virtual power plant. The virtual power plant aggregates distributed energy resources DERs such as distributed power sources, energy storage systems ESS, wind power WPP, photovoltaic generators (PV), electric vehicles EV, etc. in different regions into a whole, and directly participates in the coordinated management of the power grid through transactions with ISO.

3. The virtual power plant optimization scheduling method considering the controllable charging and discharging of electric vehicles according to claim 1 is characterized by: The specific contents of the VPP operation model are as follows: Assuming that VPP participates in the day-ahead energy market and the real-time power balancing market as a price receiver, the operation mode of VPP accepting ISO dispatch and participating in the power market is as follows: before the end of the energy market on the same day, the operator needs to coordinate internal resources according to the controllable capacity curve of EVG, the predicted output of wind power and photovoltaic units, and the load demand curve issued by ISO, and submit the power purchase and sale plan and dispatch plan for each period of the next day to ISO; in the day-ahead energy market, ISO will not settle and impose a fine on the surplus power of VPP that exceeds the load demand, and the deficit power that is lower than the load demand will be sold at the power purchase price in the energy market; in the subsequent real-time market, in order to cope with the output fluctuation of renewable energy units, VPP will give priority to adjusting internal controllable resources to reduce power deviation; for the excess power generation that VPP still cannot suppress, ISO will not clear it and impose a heavier fine than that in the energy market, and sell the missing power at the real-time market power purchase price; in addition, after each dispatch cycle, ISO should conduct a deviation assessment on the actual power generation and power consumption of VPP to improve the VPP's level of absorption of wind and solar energy resources.

4. The virtual power plant optimization scheduling method considering the controllable charging and discharging of electric vehicles according to claim 1 is characterized by: The state vector of electric vehicles after entering the network can be divided into the following three categories: 1) Class I electric vehicles: Electric vehicles in Class I status have enough time to connect to the grid and discharge the remaining power on the grid to the minimum power, and then charge to the off-grid demand before leaving the grid; VPP has sufficient time to flexibly arrange the charging and discharging behavior of these electric vehicles; 2) Electric vehicles in Class II state: Electric vehicles in Class II state do not have enough time to fully discharge to the minimum power due to the restrictions of grid connection time and off-grid demand. Therefore, VPP can only give priority to allowing them to discharge as much as possible and switch such electric vehicles from the discharge state to the charging state within a certain period of time until the off-grid time; 3) Electric vehicles in Class III state: Electric vehicles in Class III state are connected to the grid for too short a time and cannot even meet their own basic charging needs. Therefore, VPP cannot schedule the charging and discharging behavior of such electric vehicles. In this state, after the electric vehicle is connected to the grid, VPP must charge it immediately to meet its charging needs to the maximum extent.

5. The virtual power plant optimization scheduling method considering the controllable charging and discharging of electric vehicles according to claim 1 is characterized by: The process of processing the variables of electric vehicles by summing the Minkowski method is as follows: Minkowski summation is a kind of addition applicable to Euclidean space, and its expression is shown in formula (10). Its physical essence is the expansion set of multiple spaces. After calculating the Minkowski sum of two variable spaces, the envelope of the variable sum can be obtained. Based on this idea, the electric vehicle group can be aggregated into a generalized energy storage device. Where: A and B are two variable spaces; a and b are elements of A and B respectively; represents the Minkowski sum of A and B; the premise of Minkowski summation is that the two variable spaces have the same domain. However, due to the differences in the grid-connected time of individual electric vehicles, their domain T EVn is heterogeneous, so it is necessary to extend it to the same scheduling time set T, thereby obtaining the generalized energy storage model shown in formula (11): Based on the electric vehicle classification results, Minkowski summation is performed on all EV variables to obtain a decision envelope space consisting of all possible charging and discharging behaviors of EVG; this envelope space is the controllability of the electric generator.

6. The virtual power plant optimization scheduling method considering the controllable charging and discharging of electric vehicles according to claim 1 is characterized by: The VPP two-stage robust optimization scheduling model is: 1) Objective function Based on the electricity price information released by ISO, load demand curve, and the predicted output of WPP and PV, VPP takes maximizing operating income as the optimization goal, integrates internal resources, and provides a day-ahead dispatch plan. The income of VPP can be expressed as formula (21): In the formula, R CGT,t Refers to the operating income of CGT at time t, R W,t refers to WPP’s operating income at time t, R PV,t Refers to the operating income of PV at time t, R EV,t Refers to the operating income of EV at time t, R ESS,t Refers to the operating income of ESS at time t, ΔR t It refers to the cost of purchasing and disposing electricity at the VPP in the DAM at time t; 2) Constraints ①CGT constraints In the formula, The minimum output of the i-th CGT at time t is: The maximum output of the i-th CGT at time t, U CGT,i,t is a binary variable, indicating whether CGTi is running at time t. If yes, it takes 1, otherwise it takes 0; ②EVG constraints Formulas (21)-(22) are the upper and lower limits of the electric vehicle charging and discharging power; Formulas (23)-(24) represent the constraint relationship between the electric vehicle power and the charging and discharging power; ③ Energy storage system constraints ④Power balance constraints 7. The virtual power plant optimization scheduling method considering the controllable charging and discharging of electric vehicles according to claim 6 is characterized in that: the two-stage robust optimization method is used to solve the optimization scheduling of the VPP. The specific contents are as follows: The maximum form of the objective function in equation (21) is converted into the minimum form of the objective function, as shown in equation (27), and the constraints are equations (22)-(25). The simplified format of formula (27) can be expressed as: in, The min part is the first stage problem in the two-stage robust optimization model, the decision vector is x, and the optimization goal is to minimize the startup and shutdown costs of CGT. The max-min part is the second stage problem, u is the uncertainty variable introduced after considering the uncertainty of WPP and PV output, and the decision vector is y. The max part aims to find the "worst case" that leads to the maximum operating cost of VPP, while the min part aims to find the minimum operating cost of the remaining "worst cases" except the startup and shutdown costs of CGT. a, b, c are the column vectors of objective function coefficients; d, h, m are the constant column vectors; A, B, C, D, E, F, G are the coefficient matrices.

8. A non-volatile storage medium, characterized in that: The non-volatile storage medium includes a stored program, wherein when the program is executed, the device where the non-volatile storage medium is located is controlled to execute the method according to any one of claims 1 to 7.

9. An electronic device, characterized in that: It comprises a processor and a memory; the memory stores computer-readable instructions, and the processor is used to execute the computer-readable instructions, wherein the computer-readable instructions execute the method described in any one of claims 1 to 7 when executed.

Images