A risk-based robust optimization method for distributed generation and energy storage virtual power plants
Through conditional risk value and segmented linearization processing, a two-stage robust optimization scheduling model is constructed, which solves the problem that traditional robust optimization model is difficult to determine the appropriate uncertain set. Using the probability distribution information of renewable energy output, the virtualization is achieved Accurate analysis of power plant operation risks and cost optimization ensure the economy and safety of virtual power plants.
Patent Information
- Application Number
- CN202210120824.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-02-09
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2042-02-09
AI Technical Summary
Traditional robust optimization models are difficult to determine the appropriate uncertain set, and cannot use the probability distribution information of renewable energy output for quantitative analysis, resulting in the scheduling strategy being too conservative or radical, and cannot effectively deal with the operating risks of distributed power supplies and energy storage virtual power plants. .
A robust optimization method for distributed power and energy storage virtual power plants that calculate risks is proposed. The operational risk is defined through the value of conditions and risk linearization is processed in segments, a two-stage robust optimization scheduling model is constructed, and the output of renewable energy is achieved. Probability distribution information, optimize the uncertain set of the robust model, and find the harshest operating scenarios within the optimized uncertain set, and formulate a scheduling plan.
Accurate quantitative analysis of scheduling plan risks, determine appropriate sets of uncertain models, formulate scheduling plans to deal with the harshest operating scenarios, ensure the economic and safety of virtual power plants, and coordinate the optimization of operating costs.
Smart Images

Figure FHA0000007003680000011 
Figure FHA0000007003680000012 
Figure FHA0000007003680000013
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of distributed power supply, and in particular to a risk-based robust optimization method for distributed power supply and energy storage virtual power plant. Background Art
[0002] Distributed power sources, especially distributed renewable energy, have been widely used due to their flexibility, environmental protection and economy. However, distributed power sources also have problems such as small capacity, intermittent and random nature, and difficulty in management, which bring challenges to their grid-connected operation. In order to solve these problems, the concept of virtual power plant (VPP) came into being. VPP aggregates distributed power sources, energy storage and other power resources through advanced control, metering, communication and other technologies, and realizes the coordinated and optimized operation of various resources through a higher-level software architecture. The establishment of VPP can effectively reduce the impact of distributed power grid connection on the distribution network and improve the stability of system operation.
[0003] At present, through the economic analysis of VPP, it is concluded that the coordination of multiple power resources can obtain greater benefits. By studying the intraday interaction mechanism between VPP and the power grid, and adjusting the adjustable resources inside VPP, VPP can be guaranteed to meet the adjustment plan of the power grid. There are also studies that consider the impact of the liberalization of the power sales side on the demand response of VPP and propose a bidding strategy for VPP. And electric vehicles are regarded as energy storage resources in VPP, and their response capabilities at multiple time scales are evaluated. All of the above studies have established deterministic models. However, there are often a large number of distributed renewable energy sources such as wind power and photovoltaics distributed in VPP, and their uncertainty makes it difficult to dispatch and operate VPP. In order to cope with the uncertainty of renewable energy, robust optimization methods have been widely used because they do not require a large number of samples and can meet the operating constraints of all random scenarios. At present, existing technologies use robust optimization methods to characterize the uncertainty of wind power in VPP and propose a robust bidding model for VPP. In addition, there are studies on the intraday scheduling problem of VPP based on robust optimization. The results of the example study show that the use of robust optimization methods can reduce the actual daily operating cost and improve the robustness of the optimization strategy. In addition, there are also literatures that propose VPP online equivalent technology that connects the robust aggregation of day-ahead distributed resources and intraday rolling corrections, and construct an interactive scheduling mode between VPP and power grid that considers differentiated peak-shaving needs.
[0004] However, the above traditional robust optimization model has two shortcomings. On the one hand, it is difficult to determine the appropriate uncertainty set. When the uncertainty set is too large, the scheduling strategy will be very conservative; on the contrary, it will be too radical. On the other hand, this method cannot utilize the probability distribution information of renewable energy output, nor can it quantitatively analyze the risks of the scheduling plan. Summary of the invention
[0005] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and provide a risk-based robust optimization method for distributed power sources and energy storage virtual power plants.
[0006] The purpose of the present invention can be achieved by the following technical solutions:
[0007] A risk-based robust optimization method for distributed power generation and energy storage virtual power plant is applied to the optimal scheduling of a VPP composed of distributed power generation and energy storage. The method includes:
[0008] The operational risk is defined according to the conditional value at risk, and is divided into two categories: renewable energy abandonment risk and load loss risk. The classified risks are then piecewise linearized.
[0009] A two-stage robust optimization scheduling model taking into account risks is constructed for the VPP composed of distributed power sources and energy storage, and constraints are set for the two-stage model based on the results of piecewise linearization processing;
[0010] Solve the model, determine the uncertainty set of the robust model, and develop a scheduling plan.
[0011] The calculation formulas for the risk of abandoning renewable energy and the risk of load loss are as follows:
[0012]
[0013] Where: and are the risk of abandoning renewable energy and the risk of load loss, is the capacity of renewable energy units, ξ and Pr(ξ) are the probability density functions of renewable energy output and renewable energy, respectively. and They are respectively the upper and lower boundaries of the output of renewable energy unit r at time t, where the predicted output will not bring operating risks to the VPP.
[0014] The expression after piecewise linearization of the classified risk is:
[0015]
[0016] Where: s is the number of segments of piecewise linearization, and are the coefficients of the linearization constraints, respectively.
[0017] In the present invention, in the two-stage robust optimization scheduling model taking into account risk, the objective function of the first stage model is to minimize the operating risk brought to the VPP by the uncertainty of renewable energy, and the expression is:
[0018]
[0019] Where: T and R are the time set of the VPP optimization scheduling problem composed of distributed power sources and energy storage and the set of renewable energy units in the VPP, respectively. cur With C loss are the unit renewable energy abandonment penalty and the unit load loss penalty respectively. According to the above objective function, the uncertainty set of the first stage is obtained: and They are the risk of abandoning renewable energy and the risk of load loss.
[0020] The constraints of the first-stage model include the constraints for piecewise linearization of classified risks and the uncertainty set boundary constraints. The uncertainty set boundary constraints are the upper and lower boundary constraints for the output of the renewable energy unit r at time t, which will not bring operating risks to the VPP. The expression is:
[0021]
[0022] Where: and are the upper and lower bounds of the output of renewable energy unit r at time t, which will not bring operating risks to the VPP. It is the maximum value of the upper boundary of the output of renewable energy unit r at time t, which will not bring operating risks to the VPP.
[0023] In the present invention, in the two-stage robust optimization scheduling model taking into account risk, the objective function of the second stage model is to find the worst operating scenario in the uncertainty set determined in the first stage and minimize the operating cost under the worst operating scenario. The operating cost under the worst operating scenario includes the power purchase cost of the distribution network, the output cost of conventional generator sets within the VPP, and the energy storage operating cost.
[0024] The objective function of the second-stage model is in a two-layer form, expressed as:
[0025]
[0026] Where: G and E are the set of conventional generators and energy storage devices in the VPP, respectively. C b With P b,t are the unit power purchase cost and power purchase quantity of VPP from the upper distribution network, C g With P g,t are the unit power generation cost and power generation of conventional power generation units, and are the unit charging cost and charging capacity of the energy storage device, and They are the unit discharge cost and discharge capacity of the energy storage device respectively.
[0027] The constraints of the second-stage model include power balance constraints, power purchase constraints, conventional generator output and ramp constraints, energy storage device power and energy constraints, and renewable energy output constraints.
[0028] Furthermore, the present invention adopts a column constraint generation algorithm to solve the model in the form of main-subproblem iteration.
[0029] Compared with the prior art, the present invention has at least the following beneficial effects:
[0030] In order to deal with the impact of the uncertainty of distributed renewable energy on the operation of virtual power plants, the present invention establishes a two-stage robust optimization scheduling model that takes risks into account, and utilizes the probability distribution information of renewable energy output. The first stage of the model optimizes the uncertainty set of the robust model, and simultaneously evaluates the risk of abandoning renewable energy and the risk of load loss; the second stage finds the worst operating scenario in the optimized uncertainty set, and optimizes the operating cost of the virtual power plant under this scenario. The two-stage model is solved collaboratively to determine the appropriate uncertainty set of the robust model, formulate a scheduling plan to deal with the worst operating scenario, realize accurate quantitative analysis of scheduling plan risks, and ensure the economy and safety of virtual power plant operation. By linearizing the two types of risks, the proposed model can be solved using the most advanced column constraint generation algorithm, and the example analysis verifies the effectiveness of the proposed model. BRIEF DESCRIPTION OF THE DRAWINGS
[0031] Figure 1 Schematic diagram of the probability density curve of renewable energy output in the embodiment;
[0032] Figure 2 A schematic diagram of the electricity purchase cost and load curve in the embodiment;
[0033] Figure 3 Schematic diagram of energy and power of energy storage in the embodiment;
[0034] Figure 4 The energy and power of energy storage when the time-of-use electricity price and constraint (16) are ignored in the embodiment;
[0035] Figure 5 is the lower bound of the output of the four wind turbines under the three unit load loss penalties in the embodiment;
[0036] Figure 6 It is the electricity purchasing range of the virtual power plant in the embodiment. DETAILED DESCRIPTION
[0037] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work should belong to the scope of protection of the present invention.
[0038] Example
[0039] In order to address the shortcomings of traditional robust optimization models, existing literature has introduced probability distribution functions into two-stage robust optimization, and set the uncertainty set boundary as an optimization variable rather than a pre-set constant, thereby quantitatively analyzing the operational risk of the scheduling plan. There are also literatures that apply this idea in multi-zone AC / DC hybrid systems and transmission and distribution coupled power systems. The present invention provides a risk-based robust optimization method for distributed power sources and energy storage virtual power plants. The method combines robust optimization with risk and is applied to the VPP optimization scheduling problem composed of distributed power sources and energy storage, thereby constructing a risk-based two-stage robust optimization scheduling model. In the first stage of the model, the uncertainty set of the robust model is optimized and the corresponding operational risks are quantified; in the second stage, the operating cost under the worst scenario is optimized. The collaborative solution of the two-stage model can achieve a balance between the economic efficiency and safety of VPP operation.
[0040] The robust optimization method of distributed power sources and energy storage virtual power plants taking into account risks proposed in the present invention first defines operating risks based on the concept of conditional value-at-risk (CVaR), and divides them into two categories: renewable energy abandonment risk and load loss risk.
[0041] Figure 1 A general probability density curve of renewable energy output is given. r,t represents the predicted output of renewable energy unit r, t is the time, that is, A r,t represents the predicted output of renewable energy unit r at time t; and They represent the upper and lower limits of the output that will not bring operational risks to the VPP. When the output is less than When VPP power generation resources are insufficient, it will lead to the risk of load loss.
[0042] Formula (1) and Formula (2) respectively give the calculation formulas for the risk of abandoning renewable energy and the risk of load loss based on CVaR:
[0043]
[0044] Where: and are the risk of abandoning renewable energy and the risk of load loss, is the capacity of renewable energy units, ξ and Pr(ξ) represent the renewable energy output and its probability density function respectively.
[0045] Since equations (1)-(2) are strong nonlinear constraints in integral form, in order to improve the efficiency of solving the optimization model, the present invention performs piecewise linearization on the above two constraints. The method for piecewise linearization of the above two constraints can adopt the existing technology, such as the method in the literature ZHANG Ning, KANG Chongqing, XIA Qing. et al. A convex model of risk-based unit commitment for day-ahead market clearing considering wind power uncertainty. The piecewise linearization results obtained according to this method are shown in equations (3)-(4):
[0046]
[0047] Where: s represents the number of segments of piecewise linearization, and are the coefficients of the linearization constraints.
[0048] The method of the present invention is aimed at VPPs composed of distributed power sources and energy storage, and also proposes a two-stage robust optimization scheduling model that takes risks into account. In order to solve the problem that it is difficult for traditional robust optimization models to determine a suitable uncertainty set, the first stage of the model optimizes the boundary of the uncertainty set of the robust optimization model and minimizes the corresponding operating risks. The second stage searches for the worst operating scenario in the optimized uncertainty set and optimizes the VPP operating cost under the scenario. The two-stage model is solved collaboratively to determine a suitable robust model uncertainty set and formulate a scheduling plan to deal with the worst operating scenario, thereby ensuring the economy and safety of VPP operation. It should be noted that the method of the present invention is aimed at intraday scheduling problems, and the scheduling time interval can be set to 1h with reference to the prior art.
[0049] In the present invention, the objective function of the first-stage model is to minimize the operational risk of the VPP caused by the uncertainty of renewable energy, as shown in formula (5).
[0050]
[0051] Where: T and R represent the time set of the studied problem and the set of renewable energy units in the VPP, respectively. cur With Closs They are the unit renewable energy abandonment penalty and the unit load loss penalty respectively.
[0052] The constraints of the first stage model include constraints (3)-(4) and (6), where equation (6) constrains the uncertainty set boundary and The value range of .
[0053]
[0054] In the formula, It is the maximum value of the upper boundary of the output of renewable energy unit r at time t, which will not bring operating risks to the VPP.
[0055] In the present invention, the second-stage model finds the worst operating scenario in the uncertainty set determined in the first stage, and minimizes the operating cost under this scenario, including the cost of purchasing electricity from the distribution network, the output cost of conventional generator sets within the VPP, and the energy storage operating cost. The objective function of the second stage is a two-layer form, as shown in formula (7).
[0056]
[0057] Where: G and E represent the set of conventional generators and energy storage devices in the VPP, respectively, and C b With P b,t are the unit power purchase cost and power purchase quantity of VPP from the upper distribution network, C g With P g,t are the unit power generation cost and power generation of conventional power generation units, and are the unit charging cost and charging capacity of the energy storage device, and They are the unit discharge cost and discharge capacity of the energy storage device respectively.
[0058] The constraints of the second-stage model are as follows.
[0059] 1) Power balance constraints:
[0060]
[0061] Where: D represents the load set in VPP, P r,t With P d,t They are renewable energy output and load respectively.
[0062] 2) Power purchase constraints:
[0063]
[0064] Where: and They are the upper and lower limits of the amount of electricity that VPP can purchase from the upper distribution network. When the value is negative, it means that the VPP can send power to the distribution network, thus playing the role of a power plant.
[0065] 3) Output and climbing constraints of conventional generator sets:
[0066]
[0067] Where: is the capacity of the generator set, and ΔP is the ramp rate of the generator set between adjacent moments.
[0068] 4) Power and energy constraints of energy storage devices:
[0069]
[0070] Where: and are the maximum power of charging and discharging of the energy storage device, and are the charging and discharging efficiency of the energy storage device, E e,t The amount of electricity stored in the energy storage device, and are the minimum and maximum values of the amount of electricity stored in the energy storage device, E e,|T| Indicates the power of the energy storage device at the last moment of the study. To facilitate the scheduling of energy storage in the next study time interval, constraint (16) indicates that in the study time interval, the energy of the energy storage device at the last moment must be equal to the energy of the energy storage device at the initial moment E e,0 same.
[0071] 5) Renewable energy output constraints:
[0072]
[0073] Where: represents the predicted output of renewable energy, and It is an auxiliary variable describing the uncertainty of renewable energy output, and its value is between 0 and 1. Equations (17)-(186) constitute the uncertainty set of the robust model. It should be noted that since equation (17) requires that all possible renewable energy outputs in the uncertainty set must be fully absorbed, and the second-stage objective function (7) minimizes the charging and discharging cost of the energy storage device, according to the prior art, the binary variable characterizing the charging and discharging state of the energy storage device can be omitted.
[0074] Since the second-stage model established by the present invention is a max-min double-layer form, commercial solvers cannot solve it directly. In order to solve this problem, the most advanced column-and-constraint generation (C&CG) algorithm can be used to solve the model in the form of main-subproblem iteration. In order to better illustrate the solution process of the C&CG algorithm, the proposed two-stage model is written in a compact form as shown in equations (19)-(22). Among them, equation (19) is the total objective function of the two-stage problem, equation (20) is the first-stage constraint, and equations (21)-(22) are the second-stage constraints.
[0075]
[0076] stFx≤f (20)
[0077]
[0078] Kz≤k (22)
[0079] Where x and y represent the vector form of the optimization variables of the first and second stage problems respectively. The variables of the first stage problem include and The second stage question variables include P b,t , P g,t , P r,t With E e,t ;c x With c y They represent the corresponding coefficient vectors in the objective function; z represents the auxiliary vector describing the uncertainty of renewable energy output; F, H, I, J, K are the coefficient matrices related to the optimization variables in the constraints; f, h, k are the coefficient vectors in the constraints; symbols Represents the Hadamard product.
[0080] The main problem in the C&CG algorithm includes the first-stage model and the worst operating scenario constraints found according to the second-stage model. The main problem in the i-th iteration process can be expressed by equations (23)-(26).
[0081] min(c x ) T x+η (23)
[0082] stFx≤f (24)
[0083]
[0084] Where: η is the second stage objective function value to be optimized; y (j)The decision variables added to the main problem are used to construct the objective function and operation constraints of the second-stage problem under the worst operation scenario found in the jth iteration; Characterizes the worst operating scenario found by the subproblem in the jth iteration.
[0085] The subproblem in the C&CG algorithm is a two-level max-min problem in the second stage, which can be transformed into a single-level maximization problem through strong duality theory. The subproblem in the i-th iteration process can be expressed by equations (27)-(30).
[0086]
[0087] st(H) T ξ≤c y (28)
[0088] ξ≤0 (29)
[0089] Kz≤k (30)
[0090] Where: is the optimal value of the decision variable of the first stage problem obtained by optimizing the main problem in the i-th iteration; ξ is the dual multiplier of the second stage constraint condition. The transformed subproblem contains the bilinear term z T ξ, so it needs to be processed. According to the prior art (such as the literature SHAO Chengcheng, WANG Xifan, SHAHIDEHPOUR Mohammad, et al. Security-Constrained Unit Commitment with Flexible Uncertainty Set for Variable Wind Power), the worst scenario found by the subproblem will exist on a vertex in the uncertainty set, so the optimized variable z will take the value of 0 or 1, that is, converted into a binary variable. Therefore, the bilinear term can be strictly converted into the form shown in equations (31)-(32) by the big M method.
[0091]
[0092] Where: is a new variable introduced in the linearization process; M is a sufficiently large constant.
[0093] According to the above main and sub-problems, the solution steps of the C&CG algorithm are as follows:
[0094] Step 1: Initialize, set the number of iterations i = 1, the upper bound UB = ∞, and the lower bound LB = -∞ of the objective function; set the convergence criterion e.
[0095] Step 2: Solve the main problem and obtain its objective function value V i , control variables Update the lower bound of the objective function to LB = V i .
[0096] Step 3: Solve the equation and get the objective function value J i And the worst operating scenario value Return constraints (25)-(26) to the main problem and update the upper bound of the objective function to
[0097] Step 4: Convergence judgment, if |(UB-LB) / LB|≤e, the problem converges, stop iteration, and the objective function value is UB; otherwise, continue iteration, i=i+1, and return to step 2.
[0098] This embodiment uses a certain VPP as an example for example analysis, in which the model used in the example analysis is the two-stage robust optimization scheduling model taking into account the risk proposed by the method of the present invention, as shown in formulas (3)-(18). The VPP consists of 2 conventional generator sets, 2 energy storage devices and 4 wind turbine sets. Among them, the capacity of the 2 conventional generator sets is 10MW and 20MW respectively, the climbing rate is 5MW and 10MW respectively, and the unit output cost is $7.3 / MW and $11.4 / MW respectively; the capacity of the 2 energy storage devices is 10MWh, the initial capacity is 5MWh, the maximum charge and discharge power is 5MW, the charge and discharge efficiency is 90%, and the unit charge and discharge cost is $1 / MW; the capacity of the 4 wind turbine sets is 10MW, and their output prediction errors all obey the normal distribution, with a standard deviation of 10% of the predicted value. Without special instructions, the unit abandonment penalty for renewable energy and the unit loss of load penalty are $100 / MW and $1000 / MW respectively. At the same time, the unit cost of the VPP purchasing electricity from the distribution network is the time-of-use electricity price. The specific price within 24 hours and the load of the VPP can be found in Figure 2 .
[0099] To illustrate the role of energy storage, Table 1 compares the operating costs of VPPs under different energy storage capacities. Among them, when the energy storage capacity changes, its initial capacity and maximum charge and discharge power change proportionally. As can be seen from the table, as the energy storage capacity increases, the power purchase cost of VPP in the worst scenario decreases linearly, while the energy storage operating cost increases linearly, and conventional generators remain in full power, indicating that VPPs prefer to use energy storage and conventional generators with lower costs in the region.
[0100] Table 1 Operating costs of virtual power plants under different energy storage capacities
[0101]
[0102] At the same time, different from the results of the literature (CHEN Zhe, DONG Shufeng, GUO Chuangxin, et al. Fully distributed risk-based robust reserve scheduling for bulk hybrid AC-DC systems) and (CHEN Zhe, GUO Chuangxin, DONG Shufeng, et al. Distributed robust dynamic economic dispatch of integrated transmission and distribution systems), since the conventional generators have been kept in full power, when the output of renewable energy suddenly increases at adjacent moments, the system is no longer constrained by the ramp rate of the conventional generators, but the upper distribution network quickly adjusts the power to respond to the fluctuation of renewable energy, thereby achieving full absorption of renewable energy, and thus does not bring the risk of abandoning renewable energy to the VPP. This also reflects the positive role of the construction of VPP in absorbing distributed renewable energy. Overall, the total operating cost decreases linearly with the increase of energy storage capacity, which illustrates the role of energy storage in improving the flexibility of VPP operation.
[0103] Figure 3 A curve showing the change of power and stored energy of one of the energy storages is given to illustrate the mechanism of energy storage. A positive value of the energy storage power indicates discharge, and a negative value indicates charging. Figure 2 It can be seen from the change curve of the unit power purchase cost that due to the time-of-use electricity price, the energy storage is charged before the high electricity price moment (moment 8). Due to the limits of charging power and efficiency, it is fully charged after two charges at moments 4 and 6; it is discharged at moments 8 and 9 when the electricity price is high, and the stored energy at the end of the discharge is its lower limit; in order to meet the power requirement at the end of constraint (16), it is charged at the low electricity price moment 21, so that the stored energy at the end is the same as the initial moment. From the above analysis, it can be seen that the energy storage is arbitraged under the guidance of the time-of-use electricity price, thereby achieving the effect of peak shaving and valley filling, and improving the operational flexibility of the VPP.
[0104] In order to further verify that energy storage works under the guidance of time-of-use electricity prices, this embodiment further simulates the dispatch results when the unit electricity purchase cost is constant. The results show that in this case, the energy storage device is not used within 24 hours. Although the operating cost of energy storage is lower than the electricity purchase cost, due to the requirement of storing energy at the end of the moment and the power loss of charging and discharging, its overall cost of use is higher. Figure 4The energy and power of energy storage without considering the time-of-use electricity price and constraint (16) are given. It can be seen from the figure that due to ignoring the constraint of energy storage at the end moment, VPP can use energy storage with lower price at a certain moment.
[0105] In order to solve the problem that it is difficult for traditional robust optimization models to determine a suitable uncertainty set, the present invention optimizes the boundary of the uncertainty set of the robust optimization model. Figure 5 The lower bounds of the output of four wind turbines under three unit load loss penalties, that is, the lower bound of the uncertainty set, are shown. As can be seen from the figure, as the unit load loss penalty increases, the lower bound of the uncertainty set gradually moves downward, indicating that VPP has the motivation to ensure the safety of system operation at a higher operating cost. From this perspective, the model proposed in this paper can directly reflect the dispatcher's preference for VPP to improve operating economy and reduce operating risks by optimizing the boundary of the uncertainty set.
[0106] Combined with the optimized upper and lower boundaries of the uncertainty set, Figure 6 The power purchase range of the VPP from the upper distribution network is given when the unit load loss penalty is $100 / MW, so that the upper distribution network can further make reasonable decisions and dispatch. Among them, the negative value means that the VPP sells electricity to the distribution network. As can be seen from the figure, the VPP can both purchase and sell electricity from the distribution network, reflecting the two-way interactive characteristics of the VPP.
[0107] The present invention proposes a two-stage robust optimization scheduling model that takes risks into account for a VPP composed of distributed power sources and energy storage, and solves it using the most mainstream C&CG algorithm. The following conclusions can be drawn through examples. 1) Energy storage arbitrages the time-of-use electricity price of VPP purchasing electricity from the superior distribution network, thereby achieving the effect of peak shaving and valley filling, and improving the economy and flexibility of VPP operation. 2) By optimizing the uncertainty set boundary of the robust optimization model, a reasonable uncertainty set can be obtained, and the scheduling operation risk is evaluated at the same time. The construction of VPP is conducive to the absorption of distributed renewable energy, and the preference of dispatchers for improving operational economy and reducing operational risks has a significant impact on the lower bound of the uncertainty set considered. 3) The model proposed in the present invention can also give a reasonable range of power purchases from the superior distribution network of VPP, which reflects the two-way interactive characteristics of VPP. The advantages of VPP in managing distributed power sources are fully utilized.
[0108] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person familiar with the technical field can easily think of various equivalent modifications or substitutions within the technical scope disclosed by the present invention, and these modifications or substitutions should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.
Claims
1. A risk-based robust optimization method for distributed power and energy storage virtual power plants, applied to the optimal scheduling of VPPs composed of distributed power and energy storage, characterized in that: include: The operational risk is defined according to the conditional value at risk, and is divided into two categories: renewable energy abandonment risk and load loss risk. The classified risks are then piecewise linearized. A two-stage robust optimization scheduling model taking into account risks is constructed for the VPP composed of distributed power sources and energy storage, and constraints are set for the two-stage model based on the results of piecewise linearization processing; Solve the model, determine the uncertainty set of the robust model, and develop a scheduling plan; Among them, in the two-stage robust optimization scheduling model taking into account risks, the objective function of the first stage model is to minimize the operational risk brought to the VPP by the uncertainty of renewable energy, and the expression is: Where: T and R are the time set of the VPP optimization scheduling problem composed of distributed power sources and energy storage and the set of renewable energy units in the VPP, respectively. cur With C loss are the unit renewable energy abandonment penalty and the unit load loss penalty respectively. According to the above objective function, the uncertainty set of the first stage is obtained: and They are the risk of abandoning renewable energy and the risk of load loss; The objective function of the second-stage model is to find the worst operating scenario in the uncertainty set determined in the first stage and minimize the operating cost under the worst operating scenario; the operating cost under the worst operating scenario includes the power purchase cost of the distribution network, the output cost of conventional generators within the VPP, and the energy storage operating cost; The objective function of the second-stage model is in a two-layer form, expressed as: Where: G and E are the set of conventional generators and energy storage devices in the VPP, respectively. C b With P b,t are the unit power purchase cost and power purchase quantity of VPP from the upper distribution network, C g With P g,t are the unit power generation cost and power generation of conventional power generation units, and are the unit charging cost and charging capacity of the energy storage device, and are the unit discharge cost and discharge capacity of the energy storage device respectively; The constraints of the second-stage model include power balance constraints, power purchase constraints, conventional generator output and ramp constraints, energy storage device power and energy constraints, and renewable energy output constraints; among them, the power purchase constraints are expressed as: Where: and are the upper and lower limits of the amount of electricity that VPP purchases from the upstream distribution network. When it is negative, it means that the VPP sends power to the distribution network, thus having the effect of a power plant.
2. The risk-based robust optimization method for distributed power sources and energy storage virtual power plants according to claim 1, characterized in that: The calculation formula for the risk of abandoning renewable energy and the risk of load loss is: Where: and are the risk of abandoning renewable energy and the risk of load loss, is the capacity of renewable energy units, ξ and Pr(ξ) are the probability density functions of renewable energy output and renewable energy, respectively. and They are respectively the upper and lower boundaries of the output of renewable energy unit r at time t, where the predicted output will not bring operating risks to the VPP.
3. The risk-based robust optimization method for distributed power sources and energy storage virtual power plants according to claim 2, characterized in that: The expression after piecewise linearization of the classified risk is: Where: s is the number of segments of piecewise linearization, and are the coefficients of the linearization constraints, respectively.
4. The risk-based robust optimization method for distributed power sources and energy storage virtual power plants according to claim 1, characterized in that: In the two-stage robust optimization scheduling model taking into account risk, the constraints of the first stage model include the constraints of piecewise linearization of classified risks and the uncertainty set boundary constraints. The uncertainty set boundary constraints are the upper and lower boundary constraints of the output of the renewable energy unit r at time t that will not bring operating risks to the VPP. The expression is: Where: and are the upper and lower bounds of the output of renewable energy unit r at time t, which will not bring operating risks to the VPP. It is the maximum value of the upper boundary of the output of renewable energy unit r at time t, which will not bring operating risks to the VPP.
5. The risk-based robust optimization method for distributed power sources and energy storage virtual power plants according to claim 1, characterized in that: The column constraint generation algorithm is used to solve the model through the iteration of main and sub-problems.
