Risk-driven long-time-sequence operation simulation scene decoupling acceleration solving method

Abstract

The invention belongs to the technical field of long-time-sequence operation simulation, and particularly relates to a long-time-sequence operation simulation scene decoupling acceleration solving method considering risk driving, which comprises the following steps: constructing a risk-driven long-time-sequence operation simulation model based on a typical scene; a risk-driven long-time-sequence operation simulation scene decoupling model is constructed, based on a Benders decomposition algorithm, risk coupling variables are optimized by the main problem, short-time operation simulation of each typical scene is carried out by the sub-problems, and the main and sub-problems are iterated to form a scene decoupling double-layer solving framework; a specific main and sub-problem model for scene decoupling is formed in combination with model characteristics, and all shaping variables in sub-problems are also regarded as coupling variables to be optimized by the main problem; and solving the simulation model by adopting a scene decoupling solving method. Time parallel solving and accurate linearization of long-time-sequence operation simulation are achieved, and the model solving time is greatly shortened while the model solving precision is guaranteed.

Description

Technical Field

[0001] The present invention belongs to the technical field of long-time series operation simulation, and particularly relates to a decoupling acceleration solution method for long-time series operation simulation scenarios considering risk drive. Background Art

[0002] In a new power system, the uncertainties on both the source and load sides are becoming increasingly significant. The system's supply-demand balance has changed from "using deterministic supply-side regulation resources to cope with uncertain loads" to "dynamically matching uncertain supply-side resources with uncertain demand-side resources", greatly intensifying the uncertainty of power system operation and posing severe challenges to system operation and risk control. Therefore, considering the flexibility and adequacy risks of the power system in long-time series operation simulation and then enhancing the flexibility and adequacy of the power system to reduce the operation risk of the system is the key to the sustainable development of current power system planning.

[0003] The computational complexity of long-time series operation simulation increases sharply with the expansion of the time range. The time series aggregation method can extract several typical cycles to reduce the complexity of the model, enabling long-time series operation simulation to be solved within a reasonable time. However, with the addition of risk costs, a large number of coupling variables between scenarios are introduced, causing the time series aggregation method to fail. Therefore, it is very important to study an effective solution technology for risk-driven long-time series operation simulation, which can reduce the complexity of model solution while ensuring the accuracy of calculation results as much as possible.

[0004] Existing accelerated solution methods for long-time series operation simulation lack effective means to handle the coupling variables between scenarios. In view of this situation, there is an urgent need for a decoupling acceleration solution method for long-time series operation simulation scenarios considering risk drive. Based on the risk-driven long-time series operation simulation model of typical scenarios, Benders decomposition is used to process the coupling variables, realizing the time parallel solution and accurate linearization of long-time series operation simulation, and greatly reducing the model solution time while ensuring the model solution accuracy. Summary of the Invention

[0005] The purpose of the present invention is to provide a decoupling acceleration solution method for long-time series operation simulation scenarios considering risk drive, which is characterized by including the following steps:

[0006] Step S1: Construct a risk-driven long-time series operation simulation model based on typical scenarios, and add the flexibility deficiency risk to the objective function based on the conditional value at risk theory; construct a risk-driven long-time series operation simulation based on typical scenarios, perform time series clustering on each annual scenario, and decompose the annual time series operation simulation into short-time series operation simulations of several typical scenarios;

[0007] Step S2: Construct a risk-driven long-term operation simulation scenario decoupling model. Based on the Benders decomposition algorithm, the master problem optimizes the risk coupling variables, and the sub-problem conducts short-term operation simulations for each typical scenario. The master and sub-problems iterate to form a scenario decoupling two-layer solution framework;

[0008] Step S3: Combine the characteristics of the risk-driven long-term operation simulation scenario decoupling model to form specific master and sub-problem models for scenario decoupling. Regard all integer variables in the sub-problem as coupling variables to be optimized by the master problem, and achieve parallel solution and accurate linearization of the risk-driven long-term operation simulation model based on typical scenarios;

[0009] Step S4: Based on the master and sub-problem models, use the scenario decoupling solution method to solve the risk-driven long-term operation simulation model based on typical scenarios, and achieve accelerated solution of the long-term operation simulation scenario decoupling considering risk drive.

[0010] The construction of the risk-driven long-term operation simulation model based on typical scenarios in Step S1 includes:

[0011] Step S11: Construct a risk index for insufficient flexibility; based on the conditional value-at-risk theory, construct the upward and downward insufficient flexibility risk costs of the system:

[0012] Step S12: Perform time-series clustering on all N annual time series to obtain N×Ns typical scenarios. Each annual scenario is replaced by Ns typical scenarios, and each typical scenario is solved in parallel without considering the risk cost.

[0013] The construction of the upward and downward insufficient flexibility risk costs of the system based on the conditional value-at-risk theory includes:

[0014]

[0015] In the formula, ρ s is the probability of the s-th annual scenario, χ curt is the penalty coefficient for abandoning new energy, χ shed is the penalty coefficient for load shedding; and are the upward and downward insufficient flexibility risk costs at time t of the ss-th typical scenario in the s-th annual time series scenario respectively, K c is the confidence level, and are auxiliary decision variables for calculating the CVaR value; the expression [x] + = max{x, 0}, is the upward flexibility demand at time t of the ss-th typical scenario in the s-th annual time series scenario, is the upward flexibility supply at time t of the ss-th typical scenario in the s-th annual time series scenario, is the downward flexibility demand at time t in the ss-th typical scenario of the s-th annual time series scenario, is the upward flexibility supply at time t in the ss-th typical scenario of the s-th annual time series scenario, and N is the number of annual time series.

[0016] The risk-driven long-term operation simulation model based on typical scenarios is as follows:

[0017] The objective function of the risk-driven long-term operation simulation model based on typical scenarios is:

[0018]

[0019] In the formula, β is the risk aversion coefficient, and n ss is the number of the ss-th variable-scale typical period in the annual scenario, are respectively the unit operation cost and the curtailment and load shedding penalty cost at time t in the ss-th variable-scale typical scenario of the s-th annual scenario; and represent the start-up and shut-down costs of unit i, is the coal consumption cost of the unit, which is processed by piecewise linearization. N is the number of annual time series, Ns is the number of typical scenarios obtained by time series clustering in the annual time series, T is the duration of the typical scenario, generally 24 hours, and C Totle is the total cost of the risk-driven long-term operation simulation, and are both introduced intermediate variables, is the independent risk cost part of each annual scenario in the upward flexibility shortage cost, is the independent risk cost part of each annual scenario in the downward flexibility shortage cost, ng is the number of thermal power units, and n is the number of system nodes, and respectively represent the integer variables of unit on and off decisions; represents the output power of the new energy unit and the load demand power at node i, which are decision variables; represents the actual values of the output power of the new energy unit and the load demand power at node i in the ss-th variable-scale scenario of the s-th annual scenario, which are obtained by generating several deterministic scenarios of simulating new energy output from historical data and then through time series clustering;

[0020] The constraint conditions of the risk-driven long-term operation simulation model based on typical scenarios include:

[0021] Unit combination constraints, new energy curtailment and load shedding constraints, node power balance constraints, node phase angle constraints, line power flow constraints, and flexibility-related constraints.

[0022] The unit commitment constraints include: unit start-stop status constraints, minimum start-stop time constraints, unit power and ramping constraints;

[0023] The unit start-stop status constraints and minimum start-stop time constraints are:

[0024]

[0025] The unit power and ramping constraints are:

[0026]

[0027] In the formula, The integer variable representing the unit start-stop status, Indicates that unit i is in the on state at time t, otherwise it is in the off state; The integer variable representing the unit on-off decision, Indicates that unit i has an on decision at time t, Indicates that unit i has an off decision at time t; T i g ,on and T i g,off Respectively represent the minimum start-stop time of the unit;

[0028] The new energy curtailment and load shedding constraints are:

[0029]

[0030] In the formula, Represents the output power of the new energy unit and the load demand power at node i, which is a decision variable; Represents the actual value of the output power of the new energy unit and the load demand power at node i in the ss-th variable scale scenario of the s-th annual scenario, which is obtained by generating several deterministic scenarios simulating the new energy output from historical data and then through time series clustering;

[0031] The node phase angle constraints are:

[0032] -π ≤ θ i,t,ss,s <π (10)

[0033] In the formula, θ i,t,ss,s Represents the phase angle of node i at time t;

[0034] The line power flow constraints are:

[0035]

[0036] In the formula, B km Is the admittance of the transmission line km, The transmission capacity of line km is affected by thermal limits, stability constraints, etc.;

[0037] The node power balance constraint is:

[0038]

[0039] In the formula, Ω k represents the set of nodes adjacent to node k, represents the set of thermal power units connected to node k;

[0040] The flexibility-related constraints are:

[0041]

[0042] The flexibility-related constraints are to facilitate subsequent expressions by splitting the risk cost into two parts that are consistent for each annual scenario and independent of each annual scenario two parts.

[0043] The construction of the risk-driven long-term operation simulation scenario decoupling model includes:

[0044] In the objective function of the risk-driven long-term operation simulation model based on typical scenarios, and are relaxed to constants denoted as and

[0045] to form the objective function of the sub-problem model:

[0046]

[0047] The constraint conditions of the sub-problem model are the same as those of the risk-driven long-term operation simulation model based on typical scenarios;

[0048] The dual multipliers corresponding to the constraints involving the coupling variables are denoted as and The optimization result of the sub-problem model is denoted as Then the Benders cut constraint added to the master problem after solving the sub-problem is as follows:

[0049]

[0050] And update the lower bound LB:

[0051]

[0052] The objective function of the master problem model is as follows:

[0053]

[0054] Denote the optimization result after optimizing the main problem as the upper bound UB;

[0055] Transfer the optimized risk coupling variables and to the sub-problem. After the sub-problem receives the risk coupling variables, short-term operation simulations are carried out in parallel for each variable-scale typical scenario. Subsequently, according to the dual multipliers and return the Benders cut constraint to the main problem again. After the main problem adds the constraint, solve the risk coupling variables again. Iterate until the convergence condition is met, and the convergence condition is UB - LB ≤ σ.

[0056] The specific master-subproblem model of scenario decoupling formed by combining the characteristics of the risk-driven long-term operation simulation scenario decoupling model includes:

[0057] Linearize the sub-problem, and regard all integer variables in the sub-problem as coupling variables to be optimized by the main problem;

[0058] The modified main problem model is as follows:

[0059]

[0060] The constraint conditions of the modified main problem model include: the unit start-stop state constraint and the minimum start-stop time constraint in the unit commitment constraint, as well as the Benders cut constraint feedback from the sub-problem;

[0061] The decision variables of the modified main problem model are: After solving the main problem model, transfer the optimized to the sub-problem;

[0062] The modified sub-problem model is as shown in Equation (19):

[0063]

[0064] The constraint conditions of the modified sub-problem model include: the unit power and ramp constraint (Equation 8.2) in the unit commitment constraint, the new energy abandonment and load shedding constraint (Equation 9), the node phase angle constraint (Equation 10), the line power flow constraint (Equation 11), the node power balance constraint (Equation 12), and the flexibility-related constraint (Equation 13);

[0065] The decision variables of the modified sub-problem model are: unit output, new energy output, load shedding power, node phase angle; the sub-problem is a pure linear programming problem. After optimization and solution, obtain the dual multipliers corresponding to the constraints related to the coupling variables and return the constraints to the main problem.

[0066] The described scenario decoupling and solving method includes:

[0067] 1) Initialize the coupling variables The upper bound UB = +∞, and the lower bound LB = -∞;

[0068] 2) Parallelly optimize and solve all typical scenario sub-problems, and obtain the dual multipliers of the corresponding constraints as well as the optimization results

[0069] 3) Add constraints to the master problem:

[0070]

[0071] and update the lower bound LB:

[0072]

[0073] 4) Solve the master problem after adding the Benders cut constraint, obtain the optimal solution and update it as the upper bound

[0074] 5) If UB - LB ≥ σ, return to step 2) for iteration; otherwise, output the optimization results, which include: the decision variables of all typical scenarios and various costs.

[0075] The described risk-driven long-time series operation simulation scenario decoupling and acceleration solving method further includes:

[0076] Step S5: Use the scenario decoupling and solving method and the overall model solving method to solve the proposed risk-driven long-time series operation simulation model, including:

[0077] After setting the parameters, use the scenario decoupling and solving method and the overall model solving method to solve the proposed risk-driven long-time series operation simulation model, and set the result comparison to verify the effectiveness of the planning scheme: compare the overall optimization costs and optimization times of the two calculation models; compare the specific optimization results of the decision variables at each moment in each typical scenario of the two calculation models

[0078] The beneficial effects of the present invention are as follows:

[0079] Applying a risk-driven long-time series operation simulation scenario decoupling and acceleration solving method provided by the present invention can improve the solving efficiency while ensuring the solving accuracy of the long-time series operation simulation.

[0080] By constructing a risk-driven long-time series operation simulation model, adding the risk of insufficient flexibility to the objective function based on the CVaR theory; constructing a risk-driven long-time series operation simulation based on typical scenarios, performing time series clustering on each annual scenario, and decomposing the annual time series operation simulation into short-time series operation simulations of several typical scenarios.

[0081] By constructing a risk-driven long-term operation simulation scenario decoupling model, based on the Benders decomposition algorithm, the master problem optimizes the risk coupling variables, and the sub-problems perform short-term operation simulations for all typical scenarios in parallel. The master and sub-problems iterate to form a scenario decoupling two-layer solution framework.

[0082] Combined with the characteristics of the model, specific master and sub-problem models for scenario decoupling are formed. Considering that the sub-problems must be linearized during the actual iteration process, all integer variables in the sub-problems are also regarded as coupling variables to be optimized by the master problem. In this way, not only can constraints be returned to the master problem based on the dual multipliers, but also the accurate linearization of the sub-problem model can be achieved.

[0083] The proposed risk-driven long-term operation simulation model is solved using the scenario decoupling solution method and the overall model solution method. The effectiveness of the proposed scenario decoupling model is verified by comparing the optimization results and calculation time.

[0084] Based on the risk-driven long-term operation simulation model based on typical scenarios, the Benders decomposition is used to process the coupling variables, realizing the time parallel solution and accurate linearization of the long-term operation simulation. While ensuring the model solution accuracy, the model solution time is significantly reduced. Brief Description of the Drawings

[0085] Figure 1 is a schematic flow chart of an accelerated solution method for long-term operation simulation scenario decoupling considering risk driving according to the present invention;

[0086] Figure 2 is a schematic structural diagram of the scenario decoupling two-layer solution framework provided by the present invention;

[0087] Figure 3 is a schematic flow chart of risk-driven long-term operation simulation scenario decoupling based on Benders decomposition provided by the present invention;

[0088] Figure 4 is a schematic structural diagram of the test system of the embodiment provided by the present invention;

[0089] Figure 5 is a comparison chart of the overall optimization results and solution times of two solution modes of the embodiment provided by the present invention from 2 annual scenario sets to 6 annual scenario sets;

[0090] Figure 6 is a comparison chart of the power balance of a single-day typical scenario of two solution modes of the embodiment provided by the present invention;

[0091] Figure 7 is a comparison chart of the power balance of a two-day typical scenario of two solution modes of the embodiment provided by the present invention;

[0092] Figure 8 It is a power balance comparison diagram of two solution modes of the embodiments provided by the present invention in a three-day typical scenario. Specific Embodiments

[0093] The present invention provides a decoupling and acceleration solution method for long-time sequential operation simulation scenarios considering risk drive. The following further elaborates on the present invention with reference to the accompanying drawings.

[0094] As Figure 1 shown, an embodiment of the present invention discloses a decoupling and acceleration solution method for long-time sequential operation simulation scenarios considering risk drive, including:

[0095] Step S1: Construct a risk-driven long-time sequential operation simulation model based on typical scenarios, and add the risk of insufficient flexibility to the objective function based on the conditional value at risk theory; construct a risk-driven long-time sequential operation simulation based on typical scenarios, perform time series clustering on each annual scenario, and decompose the annual time series operation simulation into short-time series operation simulations of several typical scenarios;

[0096] Step S2: Construct a decoupling model for risk-driven long-time sequential operation simulation scenarios. Based on the Benders decomposition algorithm, the master problem optimizes the risk coupling variables, and the sub-problem performs short-time operation simulations of each typical scenario. The master and sub-problems iterate to form a scenario decoupling two-layer solution framework;

[0097] Step S3: Combine the characteristics of the risk-driven long-time sequential operation simulation scenario decoupling model to form specific master and sub-problem models for scenario decoupling. Regard all integer variables in the sub-problem as coupling variables and optimize them by the master problem to achieve parallel solution and accurate linearization of the risk-driven long-time sequential operation simulation model based on typical scenarios;

[0098] Step S4: Based on the master and sub-problem models, use the scenario decoupling solution method to solve the risk-driven long-time sequential operation simulation model based on typical scenarios, and achieve decoupling and acceleration solution of long-time sequential operation simulation scenarios considering risk drive.

[0099] In this embodiment, a decoupling and acceleration solution method for risk-driven long-time sequential operation simulation scenarios is proposed. Based on the risk-driven long-time sequential operation simulation model based on typical scenarios, the Benders decomposition is used to process the coupling variables, realizing parallel solution of time for long-time sequential operation simulation and accurate linearization, and greatly reducing the model solution time while ensuring the model solution accuracy.

[0100] The following separately explains each step in detail.

[0101] Step S1: Construct a risk-driven long-term operation simulation model based on typical scenarios, and add the risk of insufficient flexibility to the objective function based on the conditional value-at-risk theory; construct a risk-driven long-term operation simulation based on typical scenarios, perform time series clustering on each annual scenario, and decompose the annual time series operation simulation into short-term operation simulations of several typical scenarios.

[0102] The construction of the risk-driven long-term operation simulation model in step S1 includes:

[0103] Step S11: Construct an insufficient flexibility risk index; based on the conditional value-at-risk theory, construct the upward and downward insufficient flexibility risk costs of the system:

[0104] Step S12: Perform time series clustering on all N annual time series to obtain N×Ns typical scenarios. Each annual scenario is replaced by Ns typical scenarios for the annual time series, and each typical scenario is solved in parallel without considering the risk cost.

[0105] In this embodiment, in step S1, a risk-driven long-term operation simulation model is constructed, and the risk of insufficient flexibility is added to the objective function based on the CVaR theory; a risk-driven long-term operation simulation based on typical scenarios is constructed, time series clustering is performed on each annual scenario, and the annual time series operation simulation is decomposed into short-term operation simulations of several typical scenarios.

[0106] In step S11, an insufficient flexibility risk index is constructed, and the system flexibility deficit is the flexibility demand minus the flexibility supply The flexibility demand of the system is the fluctuation of the system net load as shown in Equation (2); the flexibility resources of the system include thermal power units and energy storage, etc., and the flexibility supply provided by them is as shown in Equations (3)-(5); therefore, the system flexibility deficit is as shown in Equation (3):

[0107]

[0108] In the formula, represents the upward and downward flexibility deficit of the system, represents the upward and downward flexibility demand of the system, represents the upward and downward flexibility supply of the system; and respectively represent the unit output, new energy output, load demand, charge and discharge power of the energy storage, and state of charge of the energy storage at time t in the ss-th typical scenario of the s-th annual scenario; respectively represent the maximum and minimum outputs of unit i and its upward and downward ramp rates, and respectively represent the maximum charge and discharge power of energy storage i, and respectively represent the maximum and minimum state of charge of energy storage i.

[0109] The lack of upward flexibility in the system includes two situations: insufficient available power generation adequacy and insufficient upward ramp - up capacity, which will lead to load shedding; the lack of downward flexibility includes two situations: insufficient valley - peaking capacity and insufficient downward ramp - up capacity, which will lead to wind curtailment.

[0110] The risk costs of insufficient upward and downward flexibility of the system constructed based on the conditional value - at - risk theory (CVaR) include:

[0111]

[0112] In the formula, ρ s is the probability of the s - th annual scenario, χ curt is the penalty coefficient for abandoning new energy, χ shed is the penalty coefficient for load shedding; and are the risk costs of insufficient upward and downward flexibility at time t of the ss - th typical scenario in the s - th annual time - series scenario respectively, K c is the confidence level, and are auxiliary decision variables for calculating the CVaR value; the expression [x] + = max{x, 0}, is the upward flexibility demand at time t of the ss - th typical scenario in the s - th annual time - series scenario, is the upward flexibility supply at time t of the ss - th typical scenario in the s - th annual time - series scenario, is the downward flexibility demand at time t of the ss - th typical scenario in the s - th annual time - series scenario, is the upward flexibility supply at time t of the ss - th typical scenario in the s - th annual time - series scenario, and N is the number of annual time - series.

[0113] In step S12, all N annual time - series are clustered in time series to obtain N×Ns typical scenarios. Each annual scenario is replaced by Ns typical scenarios, and each typical scenario can be solved in parallel without considering the risk cost. However, since the VaR of the variables corresponding to the same time in all annual scenarios in the risk cost should be the same, there are coupled variables in the typical scenarios in the risk - driven long - time - series operation simulation.

[0114] The risk - driven long - time - series operation simulation model based on typical scenarios is:

[0115] The objective function of the risk - driven long - time - series operation simulation model based on typical scenarios is:

[0116]

[0117] In the formula, β is the risk aversion coefficient, and n ss is the number of the s-th variable-scale typical time periods in the annual scenario, are respectively the unit operation cost and the curtailment and load-shedding penalty cost at time t in the s-th variable-scale typical scenario of the s-th annual scenario; and represent the start-stop cost of unit i, is the coal consumption cost of the unit, which is processed by piecewise linearization. N is the number of annual time series, Ns is the number of typical scenarios obtained by time series clustering in the annual time series, T is the duration of the typical scenario, generally 24 hours, and C Totle is the total cost of risk-driven long-term operation simulation; and are both introduced intermediate variables, is the risk cost part independent of each annual scenario in the cost of insufficient upward flexibility, is the risk cost part independent of each annual scenario in the cost of insufficient downward flexibility. ng is the number of thermal power units, and n is the number of system nodes. and represent the integer variables of unit start-up and shutdown decisions respectively; represents the output power of the new energy unit and the load demand power at node i, which are decision variables; represents the actual values of the output power of the new energy unit and the load demand power at node i in the s-th variable-scale scenario of the s-th annual scenario, which are obtained by generating several deterministic scenarios of simulated new energy output through historical data and then through time series clustering;

[0118] The constraint conditions of the risk-driven long-term operation simulation model based on typical scenarios include:

[0119] Unit combination constraints, new energy curtailment and load-shedding constraints, node power balance constraints, node phase angle constraints, line power flow constraints, and flexibility-related constraints.

[0120] (1) The unit combination constraints include: unit start-stop state constraints and minimum start-stop time constraints, unit power and ramp constraints;

[0121] The unit start-stop state constraints and minimum start-stop time constraints are:

[0122]

[0123] The unit power and ramp constraints are:

[0124]

[0125] In the formula, An integer variable representing the start-up and shut-down state of the unit Indicates that unit i is in the on state at time t, otherwise it is in the off state; An integer variable representing the start-up and shut-down decision of the unit Indicates that unit i has a start-up decision at time t Indicates that unit i has a shut-down decision at time t; T i g ,on and T i g,off respectively represent the minimum start-up and shut-down time of the unit;

[0126] (2) The new energy curtailment and load shedding constraint is as follows:

[0127]

[0128] In the formula, Represents the output power of the new energy unit and the load demand power at node i, which is a decision variable; Represents the actual value of the output power of the new energy unit and the load demand power at node i in the ss-th variable-scale scenario of the s-th annual scenario, which is obtained by generating several deterministic scenarios simulating new energy output from historical data and then through time series clustering;

[0129] (3) The node phase angle constraint is as follows:

[0130] -π ≤ θ i,t,ss,s <π (10)

[0131] In the formula, θ i,t,ss,s Represents the phase angle of node i at time t;

[0132] (4) The line power flow constraint is as follows:

[0133]

[0134] In the formula, B km Is the admittance of the transmission line km, Is the transmission capacity of the line km, which is affected by thermal limit, stability constraint, etc.;

[0135] (5) The node power balance constraint is as follows:

[0136]

[0137] In the formula, Ω k Represents the set of nodes adjacent to node k, Represents the set of thermal power units connected to node k;

[0138] (6) The flexibility-related constraint is as follows:

[0139]

[0140] The flexibility-related constraints are to facilitate subsequent expressions. The risk cost is split into two parts that are consistent for each annual scenario and independent of each annual scenario .

[0141] In this embodiment, the above-mentioned risk-driven long-term operation simulation model based on typical scenarios is a mixed integer linear programming model, which can directly call the Gurobi solver for solution.

[0142] Step S2: Construct a risk-driven long-term operation simulation scenario decoupling model. Based on the Benders decomposition algorithm, the master problem optimizes the risk coupling variables, and the sub-problems perform short-term operation simulations for each typical scenario. The master and sub-problems are iteratively composed into a scenario decoupling two-layer solution framework;

[0143] The construction of the risk-driven long-term operation simulation scenario decoupling model includes:

[0144] In the objective function of the risk-driven long-term operation simulation model based on typical scenarios, and are relaxed to constants denoted as and

[0145] In this embodiment, in the objective function of the risk-driven long-term operation simulation model based on typical scenarios for overall solution, if and are relaxed to constants, there is no coupling between each typical period in the first term of the objective function in Equation (7), and it can be directly decomposed into several small optimization problems.

[0146] In this embodiment, the variables passed by the master problem to the sub-problem of the ss-th variable scale scenario in the s-th annual scenario are denoted as and Then the objective function of the sub-problem model is denoted as Equation (14), forming the objective function of the sub-problem model:

[0147]

[0148] The constraint conditions of the sub-problem model are the same as those of the risk-driven long-term operation simulation model based on typical scenarios;

[0149] Taking and as flexibility-related constraints and adding them to the sub-problem, as shown in Equation (15),

[0150]

[0151] Denote the dual multipliers corresponding to the two constraints in Equation (15) as and The optimization result of the sub-problem model is denoted as

[0152] In this embodiment, denote the dual multipliers corresponding to the constraints involving the coupling variables as and The optimization result of the sub-problem model is denoted as Then, the Benders cut constraint added to the master problem after solving the sub-problem is as follows:

[0153]

[0154] And update the lower bound LB:

[0155]

[0156] The objective function of the master problem model is expressed as Equation (16):

[0157]

[0158] Wherein, and are both decision variables of the master problem model; and Their own definitions are common knowledge and are not specifically limited in this embodiment.

[0159] Denote the optimization result after optimizing the master problem as the upper bound UB;

[0160] After adding the Benders cut constraint to the master problem, solve to obtain and Then transmit them to the sub-problem for running simulations of each typical scenario.

[0161] Transmit the optimized risk coupling variables and to the sub-problem. After the sub-problem receives the risk coupling variables, perform parallel short-term running simulations for each variable-scale typical scenario. Subsequently, based on the dual multipliers and return the Benders cut constraint to the master problem again. After the master problem adds the constraint, solve the risk coupling variables again. Iterate like this until the convergence condition is satisfied, and the convergence condition is UB - LB ≤ σ. Those skilled in the art should know that the value of σ is set according to the specific environment, and the value of σ is not specifically limited in this embodiment.

[0162] In this embodiment, the scenario decoupling double-layer solution framework structure is asFigure 2 As shown, the risk coupling variables optimized by the main problem are passed to the sub-problems. After the sub-problems receive the risk coupling variables, each variable-scale typical scenario conducts parallel short-term operation simulations. Subsequently, according to the dual multipliers, Benders cut constraints are returned to the main problem. After the main problem adds the constraints, the risk coupling variables are solved again. Such iteration continues until the convergence condition is met.

[0163] Step S3: Combine the characteristics of the risk-driven long-term operation simulation scenario decoupling model to form the specific master-subproblem model of scenario decoupling. All integer variables in the sub-problems are also regarded as coupling variables to be optimized by the main problem, realizing the parallel solution and accurate linearization of the risk-driven long-term operation simulation model based on typical scenarios;

[0164] In this embodiment, considering that constraints need to be returned according to the dual multipliers, the sub-problems of Benders decomposition must be linear optimization problems; to ensure the feasibility of Benders decomposition, the sub-problems must be linearized.

[0165] In this embodiment, combine the model characteristics to form the specific master-subproblem model of scenario decoupling. Considering that the sub-problems must be linearized during the actual iteration process, all integer variables in the sub-problems are also regarded as coupling variables to be optimized by the main problem. In this way, not only can constraints be returned to the main problem based on the dual multipliers, but also the accurate linearization of the sub-problem model can be achieved.

[0166] Therefore, after adding the integer variables to the main problem together, the modified main problem model of the scenario decoupling model is as follows:

[0167]

[0168] For the modified main problem, the start-up and shut-down costs related to the unit start-up and shut-down variables are additionally added to the objective function; the constraints include unit start-up and shut-down state constraints, minimum start-up and shut-down time constraints, and Benders cut constraints fed back by the sub-problems. In the main problem, is the decision variable. After the main problem is solved, the optimized is passed to the sub-problems.

[0169] The objective function of the modified sub-problem model is as shown in Equation (19):

[0170]

[0171] The constraint conditions of the modified sub-problem model are given by equations (8)-(13). Equation (8) includes equations (8.1) and (8.2). The constraint conditions of the modified sub-problem model include: the unit power and ramping constraint (Equation 8.2) in the unit commitment constraint, the new energy curtailment and load shedding constraint (Equation 9), the node phase angle constraint (Equation 10), the line power flow constraint (Equation 11), the node power balance constraint (Equation 12), and the flexibility-related constraint (Equation 13);

[0172] That is: in the sub-problem, the constraints include the unit power and ramping constraint, the new energy curtailment and load shedding constraint, the node phase angle constraint, the line power flow constraint, the node power balance constraint, and the flexibility-related constraint, etc.; among which the unit output, new energy output, load shedding power, and node phase angle are decision variables. The sub-problem is a pure linear programming problem. After optimization and solution, the dual multipliers corresponding to the constraints related to the coupling variables are obtained, and the constraints are returned to the master problem.

[0173] The flow chart of the scenario decomposition algorithm is as Figure 3 shown, and the specific solution process is as follows:

[0174] 1) Initialize the coupling variables The upper bound UB = +∞, and the lower bound LB = -∞.

[0175] 2) Parallelly optimize and solve all typical scenario sub-problems, and obtain the dual multipliers of the corresponding constraints and the optimization results

[0176] 3) The master problem adds constraints:

[0177]

[0178] and updates the lower bound

[0179]

[0180] 4) Solve the master problem after adding the Benders cut constraint, obtain the optimal solution and update it as the upper bound

[0181] 5) If UB - LB ≥ σ, return to step 2 for iteration; otherwise, output the optimization results, including the decision variables of all typical scenarios and various costs.

[0182] Step S4: Based on the master-subproblem model, use the scenario decoupling solution method to solve the risk-driven long-term operation simulation model based on typical scenarios, and realize the decoupled and accelerated solution of the long-term operation simulation scenario considering risk drive.

[0183] In this embodiment, a scenario decoupling solution method and a model overall solution method are adopted to solve the proposed risk-driven long-time series operation simulation model, and the effectiveness of the proposed scenario decoupling model is verified by comparing the optimization results and the calculation time.

[0184] In an alternative embodiment, the risk-driven long-time series operation simulation scenario decoupling acceleration solution method further includes:

[0185] Step S5: Solve the proposed risk-driven long-time series operation simulation model by using the scenario decoupling solution method and the model overall solution method, including:

[0186] After setting the parameters, use the scenario decoupling solution method and the model overall solution method to solve the proposed risk-driven long-time series operation simulation model, and set the result comparison to verify the effectiveness of the planning scheme: compare the overall optimization cost and optimization time of the two calculation models; compare the specific optimization results of the decision variables at each moment in each typical scenario of the two calculation models.

[0187] In summary, the present invention discloses a risk-driven long-time series operation simulation scenario decoupling acceleration solution method. Based on the risk-driven long-time series operation simulation of typical scenarios, analyze the risk coupling variables between each annual scenario introduced, and use Benders decomposition to decouple between each typical scenario, so that all typical scenarios can be solved in parallel. Compare the proposed acceleration solution method with the solution time and solution results of the original model, and verify that the proposed acceleration solution method can improve the solution efficiency while ensuring the calculation accuracy.

[0188] To verify the effectiveness of the risk-driven long-time series operation simulation scenario decoupling acceleration solution method disclosed in the present invention, the following comparative verification experiments are carried out.

[0189] The test case uses Figure 4 the test system shown. Based on the actual historical data of new energy output, 1000 new energy output scenarios are randomly generated by ARMA, and the scenarios are reduced by K-means. Finally, 2-6 new energy annual output scenario sets are retained to simulate the real-time output power of new energy.

[0190] Cluster the new energy output data and load data of each annual scenario from 2-year scenarios to 6-year scenarios. A total of 8 types of typical scenarios are generated for each annual scenario for one day, two days, and three days. Relevant calculations are all completed on a computer with an Intel Core i7-12700H processor and 16GB of memory. Python is used to program the test case and Gurobi is called to solve it. Two calculation modes are set to compare the calculation results and calculation time: Mode 1 directly solves all annual scenarios; Mode 2 uses the proposed scenario decoupling decomposition solution.

[0191] To compare and analyze the effectiveness of the method proposed in this embodiment, two types of results are compared:

[0192] Result comparison 1: Compare the overall optimization cost and optimization time of the two calculation models. Table 1 shows the comparison of the three types of costs and calculation time in the optimization results from the 2-year scenario to the 6-year scenario.

[0193] Table 1 Comparison table of optimization results and calculation time of two calculation modes

[0194]

[0195] Through comparison, the maximum error of the total operating cost is 0.28%, and the errors of the risks of insufficient upward and downward flexibility are both within 0.1%. However, the calculation time of the scenario decoupling mode is significantly shortened. The calculation time of the 6-year scenario is shortened by 91%. Moreover, with the increase of the number of annual scenarios and the number of clustering typical periods, as well as the expansion of the system scale, the improvement effect of the calculation efficiency of scenario decoupling will be more obvious. The overall optimization results and solution times of the two solution modes from 2 annual scenario sets to 6 annual scenario sets are as Figure 5 shown, and it can be more intuitively reflected from Figure 5 the effectiveness of the proposed embodiment in calculation accuracy and calculation efficiency.

[0196] Result comparison 2: Compare the specific optimization results of the decision variables at each moment in each typical scenario of the two calculation models. For a more intuitive comparison, take the first annual scenario in the 3-year scenario as an example. Figure 6 、 Figure 7 and Figure 8 show the power balance diagrams of three types of typical scenarios with different durations. Taking Mode 1 as the comparison benchmark, it can be found that Mode 2 achieves an approximate optimal decision of the decision variables, including the output of thermal power units, the power of cutting off new energy and load, etc. The system power balance result of Mode 2 is basically the same as that of Mode 1, and the decision variable results of more than 98% of the periods are the same as those of Mode 1.

[0197] In an embodiment of the present invention, according to the risk of insufficient flexibility existing in system operation, a long-term operation simulation model based on conditional value-at-risk is constructed to measure the extreme risk losses brought by low-probability scenarios under a specific probability distribution, and to quantify the economic loss risk of tail abandoning new energy and load shedding due to insufficient flexibility. In order to effectively solve the risk-driven long-term operation simulation model, on the basis of the typical scenario method, an accelerated solution model with scenario decoupling is proposed. The upper-level master problem optimizes the risk coupling variables and integer variables and transfers the coupling parameters to the lower level. The lower level performs short-term operation simulation considering the risk of insufficient flexibility. Since there are no risk coupling variables, each typical scenario is solved in parallel, and then the constraints are returned to the master problem. The two are alternately iteratively solved to finally obtain the long-term operation simulation results. By comparing the optimization results and calculation times of direct solution and solution using the proposed model, it is verified that the proposed accelerated solution method can effectively improve the calculation efficiency of risk-driven operation simulation while ensuring the solution accuracy.

Claims

1. A decoupling and accelerated solution method for long-term simulation scenarios considering risk-driven, characterized in that: The steps include: Step S1: construct a risk-driven long-term operation simulation model based on typical scenarios, and add the risk of insufficient flexibility into the objective function based on the conditional risk value theory; Construct risk-driven long-term operation simulation based on typical scenarios, perform time series clustering on each annual scenario, and decompose the annual time series operation simulation into several short-term operation simulations of typical scenarios; Step S2: Construct a risk-driven long-term operation simulation scenario decoupling model. Based on the Benders decomposition algorithm, the main problem optimizes the risk coupling variables, and the sub-problems perform short-term operation simulations of various typical scenarios. The main and sub-problems are iterated to form a scenario decoupling double-layer solution framework; Step S3: Combine the characteristics of the risk-driven long-term operation simulation scenario decoupling model to form a specific main-subproblem model for scenario decoupling, and treat all the shaping variables in the subproblem as coupling variables to be optimized by the main problem, so as to achieve parallel solution and accurate linearization of the risk-driven long-term operation simulation model based on typical scenarios; Step S4: Based on the main-subproblem model, a scenario decoupling solution method is used to solve the risk-driven long-term operation simulation model based on a typical scenario, so as to achieve decoupling and accelerated solution of the risk-driven long-term operation simulation scenario.

2. The method for decoupling and accelerating the solution of a long-term simulation scenario considering risk-driven operation according to claim 1 is characterized in that: The step S1 of constructing a risk-driven long-term operation simulation model based on typical scenarios includes: Step S11: Construct the risk index of insufficient flexibility; based on the conditional value at risk theory, construct the upward and downward insufficient flexibility risk cost of the system: Step S12: All N annual time series are clustered to obtain N×Ns typical scenarios. Each annual scenario is replaced by Ns typical scenarios, and each typical scenario is solved in parallel without considering the risk cost.

3. The method for decoupling and accelerating the solution of a long-term simulation scenario considering risk-driven operation according to claim 2 is characterized in that: Based on the conditional risk value theory, the risk costs of insufficient upward and downward flexibility in building a system include: In the formula, ρ s is the probability of the s-th year scenario, χ curt is the penalty coefficient for abandoning new energy, χ shed is the load shedding penalty coefficient; and are the upward and downward inflexibility risk costs of the ssth typical scenario at time t, K c is the confidence level, and It is an auxiliary decision variable for calculating CVaR value; the expression [x] + =max{x,0}, is the upward flexibility demand at time t for the ssth typical scenario in the sth year time series scenario, is the upward flexibility demand supply at time t in the ssth typical scenario in the sth year time series scenario, is the downward flexibility demand at time t for the ssth typical scenario in the sth year time series scenario, is the upward flexibility supply at time t in the ssth typical scenario in the sth annual time series scenario, and N is the number of annual time series.

4. The method for decoupling and accelerating the solution of a long-term simulation scenario considering risk-driven operation according to claim 1 is characterized in that: The risk-driven long-term operation simulation model based on typical scenarios is: The objective function of the risk-driven long-term operation simulation model based on typical scenarios is: In the formula, β is the risk aversion coefficient, n ss is the number of the ssth variable-scale typical period in the annual scenario, are the unit operation cost and wind curtailment load shedding penalty cost at time t in the ss-th variable-scale typical scenario in the s-th year scenario; and represents the start-up and shutdown cost of unit i, is the coal consumption cost of the unit, piecewise linear processing, N is the number of annual time series, Ns is the number of typical scenarios obtained by time series clustering in the annual time series, T is the duration of the typical scenario, generally 24 hours, C Totle The total cost of running a simulation for a long time series driven by risk, and are all introduced intermediate variables. It is the independent risk cost part of each annual scenario in the upward flexibility deficiency cost. is the independent risk cost part of each annual scenario in the downward flexibility deficiency cost, ng is the number of thermal power units, n is the number of system nodes, and Integer variables representing the startup and shutdown decisions of the units respectively; Represents the output power of the new energy unit at node i and the load demand power, which are decision variables; It represents the actual value of the output power of the new energy unit at node i and the load demand power in the ss-th variable-scale scenario in the s-th year scenario. It is obtained by using historical data through scenario generation to obtain several deterministic scenarios simulating the output of new energy and then through time series clustering; The constraints of the risk-driven long-term simulation model based on typical scenarios include: Unit combination constraints, renewable energy load shedding constraints, node power balance constraints, node phase angle constraints, line flow constraints and flexibility related constraints.

5. The method for decoupling and accelerating the solution of a long-term simulation scenario considering risk-driven operation according to claim 4 is characterized in that: The unit combination constraints include: unit start and stop state constraints and minimum start and stop time constraints, unit power and climbing constraints; The unit start-stop state constraint and minimum start-stop time constraint are: The unit power and ramp constraints are: In the formula, Integer variable indicating the start and stop status of the unit, Indicates that unit i is in the on state at time t, otherwise it is in the off state; Integer variable representing the unit start-up and shutdown decision, Indicates that unit i has a startup decision at time t, represents the shutdown decision of unit i at time t; and Respectively represent the minimum start and stop time of the unit; The load shedding constraint of abandoning new energy is: In the formula, Represents the output power of the new energy unit at node i and the load demand power, which are decision variables; It represents the actual value of the output power of the new energy unit at node i and the load demand power in the ss-th variable-scale scenario in the s-th year scenario. It is obtained by using historical data through scenario generation to obtain several deterministic scenarios simulating the output of new energy and then through time series clustering; The node phase angle constraint is: -π≤θ i,t,ss,s <π(10) In the formula, θ i,t,ss,s represents the phase angle of node i at time t; The line power flow constraint is: In the formula, B km is the admittance of the transmission line km, is the transmission capacity of the line km, which is affected by thermal limits, stability constraints, etc.; The node power balance constraint is: In the formula, Ω k represents the set of nodes adjacent to node k, represents the set of thermal power units connected to node k; The flexibility-related constraints are: The flexibility-related constraints are to facilitate subsequent expression, splitting the risk cost into consistent annual scenarios. Independent of each year scene Two parts.

6. The method for decoupling and accelerating the solution of a long-term simulation scenario considering risk-driven operation according to claim 1 is characterized in that: The construction of the risk-driven long-term operation simulation scenario decoupling model includes: In the objective function of the risk-driven long-term operation simulation model based on typical scenarios, and Relaxation to a constant is denoted by and The objective function of the subproblem model is formed as: The constraints of the sub-problem model are the same as those of the risk-driven long-term operation simulation model based on typical scenarios; The dual multiplier corresponding to the constraint involving the coupled variable is denoted as and The optimization result of the subproblem model is recorded as Then the Benders cut constraints added to the main problem after solving the subproblem are as follows: And update the lower bound LB: The objective function of the main problem model is as follows: The optimization result after optimizing the main problem is recorded as the upper bound UB; Coupling the optimized risk variables and After the subproblem receives the risk coupling variable, each variable scale typical scenario is simulated in parallel for a short time, and then according to the dual multiplier and Then return the Benders cut constraint to the main problem, add the constraint to the main problem and solve the risk coupling variable, and iterate until the convergence condition is met, where UB-LB≤σ.

7. The method for decoupling and accelerating the solution of a long-term simulation scenario considering risk-driven operation according to claim 1 is characterized in that: The above-mentioned scenario decoupling model characteristics of the risk-driven long-term operation simulation scenario decoupling model are combined to form a specific main and sub-problem model of scenario decoupling, including: Linearize the subproblem and treat all the integer variables in the subproblem as coupled variables to be optimized by the main problem; The modified main problem model is as follows: The constraints of the modified main problem model include: unit start / stop state constraints and minimum start / stop time constraints in the unit combination constraints, and Benders cut constraints fed back by the sub-problems; The decision variables of the modified main problem model are: After solving the main problem model, the optimized Pass to sub-problem; The modified sub-problem model is as follows: The constraints of the modified sub-problem model include: unit power and ramp constraints in unit combination constraints (Equation 8.2), load shedding constraints for abandoning new energy (Equation 9), node phase angle constraints (Equation 10), line flow constraints (Equation 11), node power balance constraints (Equation 12) and flexibility-related constraints (Equation 13); The decision variables of the modified sub-problem model are: unit output, new energy output, load shedding power, and node phase angle. The sub-problem is a pure linear programming problem. After optimization and solution, the dual multipliers corresponding to the constraints related to the coupling variables are obtained, and the constraints are returned to the main problem.

8. The risk-driven decoupling and accelerated solution method for long-term simulation scenarios according to claim 1 is characterized in that: The scene decoupling solution method includes: 1) Initialize coupling variables The upper bound UB = +∞, the lower bound LB = -∞; 2) Parallel optimization solves all typical scenario sub-problems and obtains the dual multipliers of the corresponding constraints And optimization results 3) Add constraints to the main problem: And update the lower bound LB: 4) Solve the main problem after adding the Benders cut constraint, obtain the optimal solution and update it as the upper bound 5) If UB-LB ≥ σ, return to step 2) and iterate; otherwise, output the optimization result, which includes: decision variables and various costs of all typical scenarios.

9. The risk-driven long-term operation simulation scenario decoupling and accelerated solution method according to claim 1 is characterized in that: Also includes: Step S5: Solve the proposed risk-driven long-term operation simulation model using scenario decoupling solution method and model overall solution method, including: After setting the parameters, the scenario decoupling solution method and the model overall solution method are used to solve the proposed risk-driven long-term operation simulation model, and the result comparison is set to verify the effectiveness of the planning scheme: compare the overall optimization cost and optimization time of the two calculation models; compare the specific optimization results of the decision variables at each moment in each typical scenario of the two calculation models.

Images