CN110137955A - A kind of decision-making technique counted and the robust Unit Combination of CVaR is dispatched - Google Patents

Abstract

The invention discloses a decision-making method for robust unit combination scheduling considering CVaR. Polyhedron uncertain set and robust optimization method to construct a robust unit combination optimization model that takes into account the acceptable wind power range; solve the model, and output the optimized system’s acceptable wind power range, risk loss, unit start/stop status, abandonment Wind and shedding capacity decisions. Compared with the method of setting the boundary of the uncertain set, the present invention has the boundary of the uncertain set constructed by optimization, and by adjusting the uncertainty parameters, the conservative control of the robust optimization model can be realized, and the model is based on the current power system The scheduling framework, following the adaptive mechanism between unit combination and robust optimization, can be directly used to solve the unit combination problem in current power systems.

Description

A Decision-Making Method for Robust Unit Combination Scheduling Considering CVaR

technical field

The invention relates to the technical field of power system analysis and scheduling, in particular to a decision-making method for robust unit combination scheduling that takes into account CVaR (Conditional Value at Risk, Conditional Value at Risk).

Background technique

Affected by the actual requirements of environmental protection and energy, wind power generation is showing a trend of rapid development in my country. The energy renewability and environmental non-pollution of wind power generation are the internal motivations for its development. In addition, the advantages of wind farms such as short construction time and flexible investment also promote the construction of this type of energy.

But now a high proportion of wind power with uncertain attributes is integrated into the power grid, which has brought many problems to the operation of the power system. The traditional concept and method of unit combination have been difficult to deal with effectively. Their decision-making is either aggressive or conservative, resulting in wind curtailment and load shedding. Or pay more. In order to effectively deal with the uncertainty of large-scale wind power, absorb more wind power, and achieve the best energy-saving and emission-reduction effects, the following key issues must be resolved:

① Boundary decision-making problem of power system to accommodate wind power uncertainty interval;

②The measurement of the system risk caused by the uncertainty of wind power beyond the safety boundary that the system can dispatch.

In view of this, it is urgent to provide a decision-making method that can be based on the current power system architecture and can be directly used to solve the unit combination scheduling problem of the current power system.

Contents of the invention

In order to solve the above-mentioned technical problems, the technical solution adopted in the present invention is to provide a decision-making method for robust unit combination scheduling that takes into account CVaR, including the following steps:

By weighing the unit combination cost and the risk loss cost, a polyhedron uncertain set is constructed from the interval, time and space dimensions;

Based on the idea of interval robust optimization and risk decision-making theory, the CVaR measurement method is used to measure the possible system risk of the system;

Based on the polyhedron uncertain set and the robust optimization method, a robust unit combination optimization model considering the acceptable wind power range is constructed;

Solve the model, and output the optimized system acceptable wind power range, risk loss, unit start/stop status, wind curtailment and load shedding decisions.

In the above method, the use of the CVaR measurement method to measure the possible system risk of the system includes:

Wind power curtailment risk due to underestimation and load loss risk due to overestimation.

In the above method, the establishment of a robust unit combination optimization model that takes into account the acceptable wind power range based on polyhedral uncertain sets and robust optimization methods includes the following steps:

Establish the first stage model for the day-ahead unit combination problem based on decision variables for unit start/stop status and operational risk;

According to the established first-stage model, based on the decision variables, the second-stage model is established for the system economic operation problem corresponding to the worst scenario in the wind power uncertainty set of unit output, load shedding and abandoned wind volume at each time;

Determine a two-stage robust optimization model that takes into account the acceptable wind power range.

In the above method, the solution model adopts an improved C&CG algorithm.

In the above method, the polyhedron uncertainty set is:

In the formula, W _wt , are the actual value, predicted value, lower boundary and upper boundary of the uncertainty interval of wind farm w in period t, respectively; and are the auxiliary {0,1} variables representing wind power output fluctuations; Γ ^T and Γ ^S are the uncertainty parameters of the uncertainty set in space and time, respectively.

In the above method, the expression of the CVaR value of the wind power curtailment risk caused by underestimation is:

The expression of the CVaR value of the loss of load risk due to overestimation is:

In the formula, is the wind power forecast error.

In the above method, the first stage model is specifically:

Objective function:

In the formula, is the quadratic cost function of the unit, which is linearized in three sections in the model, a _g , b _g , c _g are the coefficients of the quadratic cost function, is the output power of unit g in the first stage at period t;

is a typical exponential start-up cost function;

are the cost coefficients of charging and discharging of energy storage system e in time period t;

G and W are respectively the number of units and wind farms in the system; K is the penalty coefficient;

Constraints include:

The upper and lower limits of unit output power constraints, unit ramp rate constraints, unit minimum start-stop time constraints, active power balance constraints, node active power balance and transmission capacity constraints, risk constraints, energy storage system charging/discharging state constraints, energy storage system charging Discharge/discharge power constraints, energy storage system capacity constraints, and energy storage system charge/discharge control strategy constraints.

In the above method, the second stage model is specifically:

Objective function:

In the formula, c _wt and c _dt are the costs of wind curtailment and load shedding respectively;

ΔD _dt is the load shedding amount of the dth load in the time period t;

ΔW _wt is the abandoned wind volume of the wth wind farm in time period t;

Constraint conditions include: unit output power upper and lower limit constraints, unit ramp rate constraints, active power balance constraints, load shedding constraints, abandoned air volume constraints and node active power balance constraints.

In the above method, the two-stage robust optimization model considering the acceptable wind power range is specifically:

In the formula, F ₁ is the objective function of the first stage; F ₂ is the objective function of the second stage;

C ₁ represents the constraints satisfied by the variables in the first stage; C2 represents the constraints satisfied by the variables in the second stage;

u _gt is the 0/1 decision variable in the first stage, indicating the operating status of unit g in period t, 1 means running, 0 means outage; and are the auxiliary {0,1} variables representing wind power output fluctuations;

P _gt , ΔD _dt and ΔW _wt are the continuous variables in the second stage, among which, P _gt is the output power of unit g in period t; ΔD _dt is the load shedding amount of d load in period t; ΔW _wt is the The curtailed wind volume of a wind farm in time period t;

W _wt , are the actual output value of wind farm w in period t, the lower boundary and upper boundary of the uncertainty interval, respectively.

Compared with the method of setting the boundary of the uncertain set, the present invention has the boundary of the uncertain set constructed by optimization, and by adjusting the uncertainty parameters, the conservative control of the robust optimization model can be realized, and the model is based on the current power system The scheduling framework, following the adaptive mechanism between unit combination and robust optimization, can be directly used to solve the unit combination problem in current power systems.

Description of drawings

Fig. 1 is the flowchart that the present invention provides;

Fig. 2 is the normal distribution figure of the probability density function of wind farm prediction error among the present invention;

Fig. 3 solves two-stage robust optimization unit combination model C&CG algorithm solution flowchart in the present invention;

Fig. 4 is a 6-node system wiring diagram in the case analysis of the present invention;

Fig. 5 is a curve diagram of the consumption range of wind power prediction value, wind power prediction error band and uncertainty set in the case of the present invention;

Fig. 6 is a graph of wind power acceptance range under different node systems in the case of the present invention;

(a), Curve diagram of wind power acceptance range under 6-node system, (b), Curve diagram of wind power acceptance range under 6-node system;

Fig. 7 is a curve diagram of total scheduling cost and risk cost under different node systems and different risk thresholds in the case of the present invention;

(a), 6-node system, operating cost and risk cost curves of the system under different risk thresholds, (b), 118-node system, operating cost and risk cost curves of the system under different risk thresholds;

Fig. 8 is a dispatch result diagram of whether to consider the energy storage control strategy in the case of the present invention;

(a) Result graph of energy storage scheduling in the first stage, (b) and (c) are graphs of energy storage scheduling results with or without constraints (34) and (35), respectively.

Detailed ways

Based on the idea that the energy storage system can deal with the uncertainty of wind power to reduce the system risk, the present invention proposes a two-stage robust unit combination decision-making model considering CVaR (conditional value at risk, conditional value at risk). The specific implementation methods are combined below The present invention is described in detail with accompanying drawing of description.

As shown in Figure 1, the present invention provides a decision-making method for robust unit combination scheduling that takes into account CVaR, including the following steps:

S1. By weighing the unit combination cost and the risk loss cost, construct a polyhedral uncertainty set from the interval, time and space to realize the flexible adjustment of the uncertainty set, thereby avoiding the conservatism of the robust scheduling results.

The polyhedral uncertainty set is as follows:

In the formula, W _wt , are the actual value, predicted value, lower boundary and upper boundary of the uncertainty interval of wind farm w in period t, respectively; and are the auxiliary {0,1} variables representing wind power output fluctuations; Γ ^T and Γ ^S are the uncertainty parameters of the uncertainty set in space and time, respectively.

Equation (1) represents the actual wind power output, which is an uncertain set represented by a polyhedron structure

Equations (2) and (3) are used to control the scenarios covered by the uncertain set, and the parameters Γ ^T and Γ ^S are reasonably selected to reduce the conservatism of robust scheduling.

Equation (4) represents the pole constraint of the polyhedron, and the same wind power can only reach an uncertain upper limit or lower limit at the same time, but not at the same time.

S2. Based on the idea of interval robust optimization and risk decision-making theory, the CVaR measurement method is used to measure the possible systemic risk of the system; among them, the systemic risk includes the risk of abandoning wind and the risk of load shedding, as follows:

In this embodiment, the wind power prediction error is:

Assuming that the wind power prediction error obeys the normal distribution with mean value 0 and variance σ _w ² , its probability density function is shown in Figure 2.

In this embodiment, the CVaR calculation formula is used to represent the maximum range of wind power uncertainties that the system can accommodate The size of the average potential loss that may be caused by the situation, as shown in the shaded part in Figure 2, that is, in the case of fully exploiting the maximum adjustment capacity of the system, the average loss due to wind power fluctuations exceeding the system adjustment range is defined as the wind power acceptance of the system CVaR.

If the actual output power of wind power exceeds the upper boundary of the system regulation range, measures to abandon wind can be taken to ensure the safety of system operation. Therefore, the expression of the CVaR value of wind power curtailment risk due to underestimation is:

In the formula, Corresponding to the average value of expected wind curtailment risk in the shaded part on the right side of Figure 2.

Similarly, if the actual output power of wind power is lower than the lower boundary of the system control range, load shedding measures can be taken to ensure the safety of system operation. At this time, the expression of the CVaR value of the loss of load risk due to overestimation is:

In formula (8), Corresponds to the average value of the expected loss of load risk in the shaded area on the left in Figure 2.

However, both equations (7) and (8) are nonlinear integral expressions, and it is difficult to solve them directly with commercial solvers, so they need to be linearized; The expression of the CVaR value of the loss of load risk is linearized. Expression after linear transformation:

In the formula, Represents the CVaR value of the wind power abandonment risk caused by underestimation of the wind farm w in the period t; Respectively represent the CVaR value of the wind farm w in the time period t due to the overestimation of the risk of load loss.

To sum up, equations (9) to (12) establish the functional correspondence between the acceptable wind power uncertainty interval of the system and the CVaR value faced by the system.

S3. Based on the polyhedron uncertain set and the robust optimization method, construct a robust unit combination optimization model that takes into account the acceptable wind power range; specifically includes the following steps:

This embodiment is based on a two-stage robust optimization problem established by polyhedron uncertain sets:

S31. Establish the first stage model for the day-ahead unit combination problem based on the decision variable for unit start/stop status and operation risk;

S32. According to the established first-stage model, the second-stage model is established for the system economic operation problem corresponding to the worst scenario in the wind power uncertainty set of unit output, load shedding and curtailed wind volume at each time based on the decision variables.

S33. According to steps S31 and S32, determine a two-stage robust optimization model taking into account the acceptable wind power interval. in,

(1) Phase 1 model

The first stage is to make decisions on the start-stop status of the unit, the range of wind power output that the system can accommodate, and the decision-making volume related to energy storage.

The objective function of the first stage is to minimize the start-stop cost, operating cost, energy storage charge and discharge cost of the unit, and the operating risk cost of the system; that is

Objective function:

In the formula, is the quadratic cost function of the unit, which is linearized in three sections in the model, and a _g , b _g , and c _g are the coefficients of the quadratic cost function; is the output power of unit g in the first stage at period t;

Is a typical exponential start-up cost function, in order to facilitate the analysis, it can be linearized by using three-section piecewise linearization method;

are the cost coefficients of charging and discharging of energy storage system e in time period t;

G and W are the number of units and wind farms in the system respectively; K is the penalty coefficient.

In this embodiment, the decision-maker can adjust the penalty coefficient K to weigh the reliability of the system and the economy of operation, so as to determine the range that the system can accommodate wind power;

Constraints include:

① The upper and lower limits of the output power of the unit

In the formula, are the maximum and minimum output power allowed by the unit g, respectively.

② Unit climbing rate constraints

In the formula, are the upward and downward climbing rates of unit g, respectively.

③ The minimum start and stop time constraints of the unit

In the formula, Respectively, the starting and stopping time of unit g at the initial moment; are the minimum start-up and stop-time of unit g respectively.

④Active power balance constraints

In the formula, D _dt represents the predicted value of active power of load d in time period t; D is the total number of loads.

⑤ Node active power balance and transmission capacity constraints

where B _ij is the line admittance between node i and node j; θ _it is the phase angle of node i at period t; f _ij,t is the active transmission power of the transmission line between node i and node j; is the limit value of the active transmission power of the transmission line; B and L are the number of nodes and transmission lines in the system respectively.

⑥ Risk constraints

In the formula, Risk ^da is the threshold value of the day-ahead risk target; Indicates the installed capacity of the wind farm w.

In this embodiment, the charging and discharging behavior of the energy storage system is actually that the energy storage can not only provide power to the system as a power source, but also absorb power from the system like a load. It is precisely because of the operational flexibility of the energy storage system that it can cooperate with conventional thermal power units to effectively deal with the uncertainty of wind power and reduce the operational risk of the system. In addition, from the perspective of spatial location, the decentralized access of renewable energy will inevitably lead to the spatially dispersed configuration of energy storage to deal with its uncertainty, and the energy storage is also an energy source that can effectively cope with network constraints. From the perspective of the time course, energy storage is also a bridge for energy transfer, and the essence of energy storage is backup transmission. Therefore, the scheduling strategy of energy storage will not only affect the system's ability to absorb wind power, but also affect the security of the system.

The energy storage scheduling model is shown in formulas (29)-(35):

⑦ Energy storage system charging/discharging state constraints

In the formula, are the status flags of the energy storage system in discharge or charge, respectively. The imposition of this constraint can ensure that the energy storage system is only in the charging or discharging state at a certain moment.

⑧ Energy storage system charging/discharging power constraints

In the formula, are the discharge and charge power of the energy storage system in the first stage model and the maximum allowable discharge and charge power, respectively.

⑨ Energy storage system capacity constraints

In the formula, _Eet is the stored energy value of the energy storage system in time period ^t ; η ^ch and ηd are the charging and discharging efficiencies of the energy storage system, respectively; The maximum and minimum values allowed for the storage energy of the energy storage system; E _e,int , E _e,T are the power values of the energy storage system in the initial period and the final period, respectively. In order to ensure that the energy storage system can function normally in the next dispatch cycle, it is required that the power value of the energy storage system at the end of each cycle is equal to the power value of the initial period.

⑩Constraints on charging/discharging regulation strategy of energy storage system

Considering that the wind power output during the peak load period is higher than the predicted value or the wind power output is lower than the predicted value during the low load period, the impact on the grid peak regulation is not significant. In this embodiment, the charging and discharging modes of the energy storage system are limited in these two situations, so as to avoid the situation of discharging during the valley period or charging during the peak period, so as to reduce the loss caused by disorderly charging and discharging of the energy storage system.

In the formula, are the discharge and charge power of the energy storage system in the second-stage model, respectively.

(1) Stage 2 model

The second stage is to minimize the cost of wind curtailment and load shedding corresponding to the worst case described by the uncertainty set when the start-stop state of the unit obtained by the decision-making in the first stage and the boundary of the acceptable interval of wind power are given. Based on adaptive robust optimization, a two-stage robust optimization model considering the acceptable wind power range is constructed as follows:

Objective function:

In the formula, c _wt and c _dt are the costs of wind curtailment and load shedding respectively, and the values in this embodiment are 10$/MW and 1000$/MWh respectively.

Restrictions:

① The upper and lower limits of the output power of the unit

② Unit climbing rate constraints

③Active power balance constraints

④ Load shedding constraints

⑤ Abandoned air volume constraints

⑥Nodal Active Power Balance Constraints

From formula (36) to formula (43), energy storage system control strategy constraint formula (30) to formula (35), uncertainty set constraint formula (1) to formula (5), and transmission capacity constraint formula (21) to formula (24) constitute the second-stage model.

Based on the above ideas, a two-stage robust optimization model considering the acceptable wind power range is given:

In the formula, F ₁ is the objective function of the first stage; F ₂ is the objective function of the second stage;

C ₁ represents the constraints satisfied by the variables in the first stage; C2 represents the constraints satisfied by the variables in the second stage;

u _gt is the 0/1 decision variable in the first stage, indicating the operating status of unit g in period t, 1 means running, 0 means outage;

P _gt , ΔD _dt and ΔW _wt are continuous variables in the second stage, among which, P _gt is the output power of unit g in period t; ΔD _dt is the load shedding amount of d load in period t; ΔW _wt is the The curtailed wind volume of a wind farm in time period t.

S4. Solve the model, and output the optimized system acceptable wind power range, risk loss, start/stop status of the unit, wind curtailment and load shedding decision.

The constructed two-stage problem constitutes a two-stage robust optimal unit combination model considering CVaR. However, due to the multi-layer structure of the model, it cannot be solved directly. For the convenience of expression, the model expressed in matrix form is given, and the improved C&CG (Column and Constraint Generation) algorithm is used to solve the model.

Phase 1: Master Problem (MP)

The decision variables in the first stage include u _gt , and Since wind curtailment and load shedding will not occur in the first stage, u _gt , solution feasibility and the existence of a solution.

In the formula, x represents the binary state vector of the generating set;

vector and vector y respectively represent the continuous vector of generator output and the phase angle vector of each node;

u represents the binary state vector of the charging and discharging state of the energy storage system;

and z represent the charge and discharge power and capacity vectors of the energy storage system;

w represents the output boundary vector of the wind farm;

Q represents the operational risk vector;

a, b, c, d, e, A, B, C, D, E, and F are all corresponding constant coefficient matrices; for example, matrix A can be derived from constraints (14)-(20). Compared with the traditional robust unit combination, the proposed model introduces more decision variables and constraints, for example, w,u, and z, which will greatly increase the computational complexity. The main problem is a MILP problem, so it can be solved by the existing commercial solver CPLEX.

Phase 2: Subproblem (SP)

The decision variables in the second stage include P _gt , ΔD _dt and ΔW _wt and wind power uncertainty set

In the formula, s represents the wind curtailment and load shedding vector, and v describes the binary vector of wind power uncertainty set;

f, g, h, G, H, I, J, K, L, M and N are all corresponding constant coefficient matrices;

Represents the Hadamard product.

In this embodiment, since the subproblem is a max-min structure, the model of this structure cannot be solved directly, and it is necessary to convert the max-min structure model into a single-layer MILP problem to solve it conveniently.

First, the way of finding the dual of the internal minimization problem is transformed into a max problem. According to Wang Dan et al. in 2014 in the Chinese Society of Electrical Engineering, "Household Temperature Control Load Demand Response and Energy Efficiency Power Plant Modeling Considering User Comfort Constraints" , the mentioned strong duality theory solves the dual of the inner layer min problem, then combines with the outer layer problem, and finally converts it into a single layer max problem. The matrix form of the transformed sub-problem is expressed as follows:

λ ^T ≥ 0 (49)

Nv≤h (50)

In the formula, λ is the dual variable of constraints (37)~(43). Observing that the objective function (47) contains a bilinear term λ ^T v, we can use the outer approximation method proposed by Zhang Li in his doctoral dissertation "Research on Unit Combination Theory in Electric Power Market" in 2006 or by Sheng Wanxing et al. In 2013, he proposed "Characteristics and Research Framework of Automatic Demand Response in Intelligent Power Consumption" in the Journal of Power System Automation, and mentioned the large M linearization method to solve it.

In some cases, the outer approximation may fail to find the global optimal solution. Therefore, the bilinear term λ ^T v is solved using the large-M linearization method:

Through the above processing, the sub-problem (46) is transformed into a standard single-layer MILP problem, such as formula (55):

In the formula, is the auxiliary vector. q is a continuous vector;

Equations (52) to (54) are auxiliary constraints generated by the large M method, so that the bilinear term can be equivalently transformed into the following equation:

After the above-mentioned linear transformation, the main problem (45) and the sub-problem (55) constitute a standard two-stage MILP problem. This embodiment adopts the " Green power voluntary purchases: price elasticity and policy analysis (green energy voluntary purchase: price elasticity and policy analysis)", mentioned the improved C&CG algorithm to solve this two-stage unit combination model. Compared with the Benders decomposition algorithm, the algorithm greatly reduces the number of iterations and greatly improves the convergence speed. The overall flow of the algorithm solution is shown in Figure 3, and the steps are as follows:

A1. Initialization

Set lower bound LB=0, upper bound UB=+∞, convergence error ε≥0; number of iterations k=0, solution space is O=φ; go to step A2;

A2, solve the following main problem formula (45);

ψ≥f ^T s (57)

get the optimal solution renew Go to step A3;

A3. Solving subproblems

Based on the given unit combination state x _k+1 , u _k+1 , w _k+1 , solve the sub-problem (55) to obtain the worst fluctuation scenario of wind power output and abandoned air volume and load shedding s _k+1 ;

update upper bound

If UB-LB≤ε, stop the iteration and output the optimal solution;

Otherwise, set l=l+1, and the worst output scenario of wind power corresponding to the optimal solution of the subproblem Pass to the main question and go to step A4;

A4. Add variables and constraints

Add variables y _k+1 and s _k+1 , add constraints (59) and (60) and return to the main problem; update l=l+1 and O=O∪{k+1}. Solve the outer layer problem of the updated sub-problem, return to step A2.

ψ≥f ^T s _k+1 (59)

Beneficial effects of this embodiment:

This embodiment proposes a two-stage robust unit combination optimization model considering CVaR, and the beneficial effects are as follows:

1) Compared with the method of setting the boundary of the uncertain set, the boundary of the uncertain set constructed in this embodiment is obtained by optimization, and by adjusting the uncertainty parameter, the conservative control of the robust optimization model can be realized; at the same time, it can be Select an appropriate risk threshold, or make a compromise between system operating risk and operating cost, in order to obtain the optimal scheduling solution.

2) The configuration of energy storage increases the flexibility resources of thermal power units, improves the system's ability to deal with wind power uncertainties, and eases the system's ramp rate limit and network constraints.

3) The model is based on the current power system dispatching architecture, complies with the adaptive mechanism between unit combination and robust optimization, and can be directly used to solve the problem of unit combination in the current power system, which verifies the correctness of the model. At the same time, according to the characteristics of the model in this example, the C&CG algorithm is used to solve it, and the advantages of the proposed method in terms of calculation efficiency and calculation speed are verified.

The present embodiment is described below through a specific case.

In this case, a 6-node system and a modified IEEE118-node system are taken as examples to analyze the validity of the model proposed in the above embodiments. As shown in Figure 4, it is a 6-node system, which includes a wind farm with a capacity of 250MW, an energy storage system, 3 thermal power units, and 7 lines. Technical data such as thermal power unit capacity, ramp rate, and lines refer to "Distributionally robust solution to the reserve scheduling problem with partial" proposed by Qiaoyan Bian et al. information of wind power" (distribution robust solution method considering part of wind power information in standby dispatching).

The modified IEEE118 node system has 53 generators, 91 load nodes, 186 lines, 14 capacitors, and 9 tap changers; 3 wind farms with a capacity of 250MWh are respectively connected to nodes 59, 66 and 94 node. For detailed IEEE118 node system data, please refer to motor.ec.iit.edu/data/scuc_118. The parameters of the energy storage system are shown in Table 1. The charging/discharging prices of the energy storage system are 0.4/0.6 ($/kWh) respectively. The test calculation uses Visual Studio 2016C++ software to call the CPLEX12.8 solver for solving. The computer configuration is Win10 system, Intel Core i7-8700k series, main frequency 3.0GHz, memory 16G. The simulation time scale is 1 day, divided into 24 periods.

Table 1 Energy storage system parameters

The wind power prediction value and wind power prediction error band used in the example are shown in Figure 5. In the calculation example, the selected confidence levels are β ^T = 95% and β ^S = 95%, that is, in the 6-node system and

In a 118-node system The mean value of the wind power prediction error Δw _wt is 0, and the variance is calculated according to the following formula.

(1) The scheduling results of the CVaR robust unit combination

Taking the 6-node test system as an example, the model of this embodiment and the document "Optimizing demand response price and quantity in Wholesale markets" (wholesale market demand response price and quantity optimization) for a comparative analysis of the scheduling results of the risk-taking robust unit combination model, see Table 2.

Table 2 Comparison of scheduling results

It can be seen from Table 2 that compared with the scheduling results of the literature, the most economical unit G1 in the model of this example always runs during the research period, the most expensive unit G2 quits operation during the period from 11:00 to 12:00, and unit G3 only operates under load During the peak hours 9:00, 11:00～12:00 are in the running state, the corresponding total system cost is 94236.95$, saving 94236.95$-94632.321=395.37$; it illustrates the peak-shaving and valley-filling effect of the energy storage system, on the one hand The thermal power unit can provide enough flexible resources to cope with the uncertainty of wind power, thereby improving the system's ability to accommodate wind power. On the other hand, it can reduce the pressure of thermal power unit configuration backup during peak hours, reduce the number of unit start/stop, and as an additional source of flexibility, reduce the cost of UC dispatch.

(2) Comparison with other unit combination models

Discussed now that the model of this embodiment is compared with the deterministic unit combination (10% uncertainty interval), the given symmetric uncertain set robust unit combination, and the robust unit combination considering the risk of the asymmetric uncertain set The scheduling results obtained by these four models are shown in Table 3 below. Among them, the given symmetric uncertain set robust unit combination comes from FahriogluM. et al. in 2001 in IEEE Transactions on Power Systems (International Institute of Electrical and Electronics Engineers Power System Transactions) "Using utility information to calibrate customer demand management behavior models" (models that use utility information to calibrate customer demand management behavior).

Table 3. Scheduling results of different unit combination models

It can be seen from Table 3 that the operating cost and risk cost of the model in this embodiment are lower than those of the other three models, which shows that the model in this embodiment has better operational flexibility and the ability to reduce system operational risks. In addition, even though the configuration of the energy storage system increases the flexibility and adjustment capability of the system, there is still some wind curtailment or load shedding. This is due to the limitation of the ramp rate of the thermal power unit, which makes the system unable to cope with the intermittent problem of a high proportion of wind power at a certain moment. .

By comparing the number of iterations and running time, taking 118 nodes as an example, the total calculation time of the model in this embodiment is 8.677 seconds, and the number of iterations is 3 times, which shows that the method in this embodiment maintains the linear nature of the model and has high calculation efficiency. To meet the computational efficiency requirements of solving the actual system.

(3) Flexible uncertainty set and scope of accommodation

The wind power uncertainty set of the model in this embodiment is flexible, adjustable and asymmetrical, mainly due to two reasons:

(1) In order to limit the conservatism of the robust method, the robustness of the system is restricted from the time dimension formula (2) and space dimension formula (3).

(2) By setting different penalty costs for wind curtailment and load shedding, it leads to asymmetry in the scope of wind power consumption, that is, asymmetry in the uncertain set of wind power. It can be seen from Table 3 that even though the amount of load shedding is much smaller than the amount of abandoned air, the cost of load shedding is greater than the cost of abandoned air. This is because we set a higher penalty cost for load shedding, which makes the risk of load shedding lower.

The accommodation range obtained from model optimization in this embodiment is compared with the dispatching result of the risk-considered robust unit combination model, as shown in FIG. 6 .

It can be seen from Fig. 6 that, compared with the literature, the upper and lower boundaries of wind power consumption determined by the model of this embodiment are smaller in most periods of time. For example, in a 6-node system, time periods 1:00～3:00, 5:00, 7:00, 9:00, 12:00, 14:00, 16:00, 18:00, 21:00～22 :00;

In the 118-node system, the time slots are 2:00, 8:00, and 11:00-16:00. It proves that configuring the energy storage system enables the system to have more schedulable flexible resources to cope with the uncertainty of wind power output. On the one hand, it reduces the conservatism of traditional robust methods; System operation risk.

(4) Conservative analysis of uncertain sets

Adjusting the uncertainty parameters Γ ^T and Γ ^S can control the robustness of the model, thereby reducing the conservatism of the scheduling results. In order to analyze the influence of the change of uncertainty parameters on the degree of conservatism of the results obtained by the model of this embodiment, the calculation results of different combinations of Γ ^T and Γ ^S are listed in Table 4.

Table 4 Scheduling results under different combinations of Γ ^T and Γ ^S

It can be seen from Table 4 that when the fixed Γ ^S is constant, with the gradual increase of Γ ^T , more flexible resources are needed to deal with the uncertainty of wind power, so that the total operating cost gradually increases, but the risk cost gradually decreases, while the operating risk The size is always less than or equal to the given risk threshold. It can also be seen that when Γ ^T is fixed, the wider the distribution of wind farms (the larger Γ ^S ), the smaller the total operating cost. Increasing the strategy of geographical distribution of wind farms and relaxing network constraints can improve the system's response to wind power. Uncertainty robustness; but the risk cost increases gradually, because: in order to improve the robustness of the system to cope with wind power uncertainty, an over-conservative strategy is adopted, resulting in a suboptimal solution. Obviously, the values of Γ ^T and Γ ^S have a decisive effect on the robustness and conservatism of the robust optimization results. If the values of Γ ^T and Γ ^S are too large, the robustness of the optimization result will be better, but it is extremely conservative, and the economy will be poor; and if the values of Γ ^T and Γ ^S are too small, the allowable output will reach the forecast limit of wind power. If there are too few farms, this will not fully reflect the uncertainty of wind power.

Therefore, by setting the uncertainty parameters reasonably, the adjustable robust optimization method can eliminate the situation where the probability of the uncertainty parameters is extremely small, making the model more realistic and more applicable to engineering practice. Table 4 also lists the calculation time under different wind farm uncertainty sets ^ΓT and ^ΓS , which proves the effectiveness of the method again.

(5) The impact of the risk threshold Risk ^da

In the embodiment, the risk threshold Risk ^da is also an important parameter, which not only affects the scheduling strategy but also affects the feasibility of the scheduling solution. In practice, it can be selected based on historical data, risk appetite of decision makers, power contracts, etc. Figure 7 shows the changing trend of the system's operating cost and risk cost curve under different risk thresholds Risk ^da .

As shown in Figure 7, taking the 118-node system as an example, as the risk threshold Risk ^da increases gradually, the operating cost of the system decreases gradually, while the risk cost gradually increases. When the risk threshold of the system is Risk ^da ≈ 1000, the operating cost and risk cost of the system tend to be stable. When the risk threshold of the system Risk ^da ≥ 1400, the risk cost of the system decreases slightly. On the contrary, when the risk threshold of the system is reduced to Risk ^da ≤ 10, the system has no feasible solution. This means that the minimum feasible risk threshold for the system is 10. At the same time, it can be observed that the relationship between the operating cost and risk cost of the system and the risk threshold Risk ^da is not strictly linear. Similarly, the upper and lower limits of the risk threshold for the existence of a feasible solution to the system can also be obtained.

(6) The role of energy storage charge/discharge control strategy

Taking the 6-node system as an example, Figure 8 shows the effects of the energy storage charge/discharge control strategy constraints (34) and (35).

Among them, Figure 8(a) is the result of energy storage scheduling in the first stage. It can be seen that during the peak period of electricity consumption, the energy storage responds to the increased load demand by discharging, and during the low period of electricity consumption, it responds to the anti-peaking characteristics of wind power by charging the energy storage. Use the peak-shaving and valley-filling effect of energy storage to smooth the load curve. Figure 8(b) and Figure 8(c) compare the energy storage scheduling results with and without constraints (34) and (35). It can be seen that the energy storage charging/discharging control strategy constraints restrict the charging/discharging state of the energy storage system during the peak and low periods of the net load: the unit combination plan in the first stage requires that the energy storage system can only charge during the low period, and only charge during the peak period. Can discharge. In the second stage of real-time dispatching, if the actual output of wind power is higher than the predicted output, the excess wind power can be charged to the energy storage system; however, if the actual output of wind power is lower than the predicted value, the charging power can be reduced but cannot be discharged. Such constraints not only ensure that the scheduling strategy of the energy storage system is consistent with the peak shaving of the power grid, but also avoid the loss caused by disorderly charging and discharging of the energy storage.

The present invention is not limited to the above-mentioned best implementation mode, and anyone should know that any structural changes made under the inspiration of the present invention, and any technical solutions that are identical or similar to the present invention, all fall within the protection scope of the present invention.

Claims (9)

1. A decision method for considering CVaR robust unit combination scheduling is characterized by comprising the following steps:

constructing a polyhedral uncertain set from interval, time and space dimensions by balancing unit combination cost and risk loss cost;

based on an interval robust optimization thought and a risk decision theory, a CVaR measurement method is adopted to measure system risks possibly existing in the system;

constructing a robust generator set combination optimization model considering an acceptable wind power interval based on a polyhedron uncertain set and a robust optimization method;

and solving the model, and outputting the wind power range, risk loss, start/stop state of the unit, wind curtailment and load shedding amount decision which can be accepted by the optimized system.

2. The decision method as claimed in claim 1, wherein the measuring the system risk that the system may have using the CVaR measurement method comprises:

the risk of wind curtailment due to underestimation and the risk of load loss due to overestimation.

3. The decision method according to claim 1, wherein the construction of the robust wind turbine combination optimization model taking into account the acceptable wind power interval based on the polyhedral uncertain set and the robust optimization method comprises the following steps:

establishing a stage 1 model for the unit combination problem of the day before the unit start/stop state and the operation risk based on the decision variables;

according to the established stage 1 model, establishing a stage 2 model for the system economic operation problem corresponding to the worst scenario in the wind power uncertain set of the unit output, the load shedding amount and the wind abandoning amount at each moment based on a decision variable;

and determining a two-stage robust optimization model considering the acceptable wind power interval.

4. The decision method of claim 1, wherein the solution model employs a modified C & CG algorithm.

5. The decision method of claim 2, wherein the set of polyhedral uncertainties is:

in the formula, W_wt、Respectively obtaining an actual value, a predicted value and a lower boundary and an upper boundary of an uncertain interval of the wind power plant w at a time t;andrespectively representing auxiliary {0,1} variables of the fluctuation condition of the wind power output; gamma-shaped^TAnd Γ^SUncertainty parameters of the uncertainty set in space and time are respectively.

6. Decision method according to claim 5 characterized in that the CVaR value expression of the wind curtailment risk due to underestimation is:

the CVaR value expression for the risk of loss of load due to overestimation is:

in the formula,and the wind power prediction error is obtained.

7. The decision method according to claim 3, wherein the stage 1 model is specifically:

an objective function:

in the formula,is a quadratic cost function of the unit and is linearized in three sections in a model, a_g、b_g、c_gRespectively the coefficients of the quadratic cost function,the output power of the unit g in the 1 st stage in the time period t;

is a typical exponential start-up cost function;

respectively representing the cost coefficients of charging and discharging of the energy storage system e in the time period t;

g and W are the number of the units and the wind power plant in the system respectively; k is a penalty coefficient;

the constraint conditions include:

the method comprises the following steps of unit output power upper and lower limit constraint, unit climbing rate constraint, unit minimum on-off time constraint, active power balance constraint, node active power balance and transmission capacity constraint, risk constraint, energy storage system charging/discharging state constraint, energy storage system charging/discharging power constraint, energy storage system capacity constraint and energy storage system charging/discharging regulation strategy constraint.

8. The decision method according to claim 7, wherein the stage 2 model is specifically:

an objective function:

in the formula, c_wt、c_dtRespectively the cost of wind abandoning and load shedding;

ΔD_dtthe load shedding amount of the d load in the time period t;

ΔW_wtthe wind curtailment quantity of the w wind power plant in the time period t is obtained;

the constraint conditions include: the method comprises the following steps of unit output power upper and lower limit constraint, unit climbing rate constraint, active power balance constraint, load shedding amount constraint, air curtailment amount constraint and node active power balance constraint.

9. The decision method according to claim 3, wherein the two-stage robust optimization model taking into account the acceptable wind power interval is specifically:

in the formula, F₁Is the objective function of stage 1; f₂Is the objective function of the 2 nd stage;

C₁represents the constraints satisfied by the stage 1 variables; c2 represents the constraint satisfied by the stage 2 variable;

u_gta decision variable 0/1 in the 1 st stage represents the running state of the unit g in the time period t, 1 represents running, and 0 represents shutdown;andrespectively representing auxiliary {0,1} variables of the fluctuation condition of the wind power output;

P_gt、ΔD_dtand Δ W_wtIs a continuous variable of stage 2, wherein P_gtThe output power of the unit g in the time period t is obtained; delta D_dtThe load shedding amount of the d load in the time period t; Δ W_wtThe wind curtailment quantity of the w wind power plant in the time period t is obtained;

W_wt、respectively the actual value of the wind farm w output in the time period t, the lower boundary and the upper boundary of the uncertainty interval.