Flexible resource planning method and terminal based on CG&S distributed computing method

Abstract

The present invention discloses a flexibility resource planning method and terminal based on the CG&S distributed computing method, including the following steps: step S1, establishing a multi-stage random flexibility resource planning model; step S2, using the Danz-Wolff formula to decompose and reformulate the objective function of the multi-stage random flexibility resource planning model into a main problem and a sub-problem; step S3, solving the decomposed main problem and sub-problem to obtain the optimal planning strategy of the objective function. The present invention solves complex resource planning problems by establishing a multi-stage random flexibility resource planning model and using the Danz-Wolff formula to decompose and reformulate the main problem and sub-problem to solve, thereby obtaining the optimal planning strategy for flexibility resources in the power system.

Description

Technical Field

[0001] The present invention relates to the technical field of power planning, and in particular to a flexibility resource planning method and terminal based on a CG&S distributed computing method. Background Art

[0002] For resource planning problems, the simplex method is usually used to solve them. The simplex method can ensure that the optimal solution is found after several iterations, but it is difficult to solve mixed integer programming problems in power systems. This type of optimization problem often has a limited number of constraints but a large number of variables, so all variables cannot be clearly expressed in the model. Therefore, the simplex method cannot solve such problems, which makes it difficult to solve the optimal planning strategy for flexible resources in power systems. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a flexibility resource planning method and terminal based on the CG&S distributed computing method, which can obtain the optimal planning strategy for flexibility resources in the power system.

[0004] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0005] The flexibility resource planning method based on the CG&S distributed computing method includes the following steps:

[0006] Step S1, establishing a multi-stage stochastic flexibility resource planning model;

[0007] Step S2, using the Danz-Wolf formula to decompose and reformulate the objective function of the multi-stage stochastic flexibility resource planning model into a main problem and sub-problems;

[0008] Step S3: Solve the decomposed main problem and sub-problems to obtain the optimal planning strategy of the objective function.

[0009] In order to solve the above technical problems, another technical solution adopted by the present invention is:

[0010] A flexible resource planning terminal based on a CG&S distributed computing method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are implemented:

[0011] Step S1, establishing a multi-stage stochastic flexibility resource planning model;

[0012] Step S2, using the Danz-Wolf formula to decompose and reformulate the objective function of the multi-stage stochastic flexibility resource planning model into a main problem and sub-problems;

[0013] Step S3: Solve the decomposed main problem and sub-problems to obtain the optimal planning strategy of the objective function.

[0014] The beneficial effects of the present invention are: a flexibility resource planning method and terminal based on the CG&S distributed computing method, which solves complex resource planning problems by establishing a multi-stage random flexibility resource planning model and using the Danz-Wolf formula to decompose and rephrase the main problem and sub-problems to obtain the optimal planning strategy for flexibility resources in the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic flow chart of a flexible resource planning method based on a CG&S distributed computing method according to an embodiment of the present invention;

[0016] Figure 2 A schematic diagram of a process for solving a decomposed multi-stage random flexibility resource planning model according to an embodiment of the present invention;

[0017] Figure 3 It is a structural schematic diagram of a flexible resource planning terminal based on a CG&S distributed computing method according to an embodiment of the present invention;

[0018] Figure 4 It is a schematic diagram of the structure of a processor involved in an embodiment of the present invention.

[0019] Description of labels:

[0020] 1. Flexible resource planning terminal based on CG&S distributed computing method; 2. Processor; 21. Main processor; 22. First auxiliary processor; 23. Second auxiliary processor; 3. Memory. DETAILED DESCRIPTION

[0021] In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in combination with the implementation modes and the accompanying drawings.

[0022] Please refer to Figure 1 , Figure 2 and Figure 4 , a flexibility resource planning method based on CG&S distributed computing method, comprising the steps of:

[0023] Step S1, establishing a multi-stage stochastic flexibility resource planning model;

[0024] Step S2, using the Danz-Wolf formula to decompose and reformulate the objective function of the multi-stage stochastic flexibility resource planning model into a main problem and sub-problems;

[0025] Step S3: Solve the decomposed main problem and sub-problems to obtain the optimal planning strategy of the objective function.

[0026] From the above description, it can be seen that the beneficial effects of the present invention are: by establishing a multi-stage random flexibility resource planning model and using the Danz-Wolf formula to decompose and rephrase the main problem and sub-problems for solving, it is possible to solve complex resource planning problems and obtain the optimal planning strategy for flexibility resources in the power system.

[0027] Furthermore, the step S1 specifically includes:

[0028] The scenario tree is used to divide the planning cycle into different decision-making stages. The root node represents the initial state of the system. This node only makes investment decisions, while the subsequent nodes are composed of both operation and investment stages.

[0029] From the above description, it can be seen that after one stage is determined, the next stage of operation is carried out. This multi-stage approach allows investment decisions to be made at several points in time and takes into account the information of the uncertain parameter set known in the current stage, reducing the impact of uncertain parameters on planning.

[0030] Furthermore, in step S1, a flexible resource comprehensive planning model considering short-term response and long-term uncertainty is specifically established, and its objective function is described as follows:

[0031]

[0032] In the formula, The unit investment cost of the power generation unit at node n, is the unit investment cost of ESS at node n, is the unit investment cost of the line at node n, is the capacity of the additional generating unit installed in node n, is the capacity of the ESS installed in node n, is the capacity of the line installed at node n, is the unit variable cost of the power generation unit in node n, is the startup cost of the power generation unit, P n，t，g is the hourly output power of the power generation unit in node n, S n，t，g is the number of startups per hour of the power generation unit in node n, is the unit variable cost of ESS in node n, is the power provided by ESS in node n per hour, Compensate for the unserved demand in node n, is the load reduction of node n on bus b at time t, which has the following constraints:

[0033]

[0034]

[0035]

[0036]

[0037]

[0038]

[0039]

[0040]

[0041]

[0042]

[0043]

[0044]

[0045]

[0046]

[0047]

[0048]

[0049]

[0050]

[0051]

[0052] In the formula, are the total number of installed power generation, energy storage and line units in node n, respectively. are the additional power generation units, ESS and lines installed in node n. is the maximum number of resources installed in node n, F n，t，l is the power flow on line l in node n at time t, They are the self-discharge, charging efficiency and discharging efficiency of ESS per hour respectively.

[0053] From the above description, it can be seen that it gives a flexible resource comprehensive planning model that considers short-term response and long-term uncertainty. The constraints associate the investment in different resources on the predecessor node of n with the total units installed in the node n, and also limit the total number of units installed in the node n, ensuring the power balance of each hour in each bus operation phase of the system, limiting the maximum flow through the transmission line to each installed capacity, and giving a simplified model of the power generation unit. Specifically, the maximum power generation limit of the thermal generator is given, and the power of the RES generator is limited by the availability of the installed units and primary energy. The energy balance, minimum and maximum energy storage capacity, and maximum charge and discharge power constraints are considered. The number of installed units of existing assets is set at each node n and the number of existing units at the root node, which limits the installation of additional power generation, energy storage or transmission units after the root node.

[0054] Furthermore, the step S3 specifically includes:

[0055] Step S31, solving the linear programming relaxation problem of the main problem, and obtaining the rate of change of the objective function;

[0056] Step S32, solving the sub-problem according to the rate of change of the objective function to generate a new CG column and obtain the corresponding minimum change of the objective function, each CG column corresponds to a pair of feasible total number of installation units and the optimization operation of the corresponding nodes, and the column with the change of the objective function less than zero is added to the main problem;

[0057] Step S33, calculate the change of the objective function in the linear programming, and check the conditions for terminating the algorithm calculation. If the change of the objective function is less than zero, the correlation coefficient of the column variable is added to the coefficient matrix of the main problem, and return to step 31 to perform a new iteration. If not, the termination condition is met, the calculation is terminated, and the optimal planning strategy of the objective function is obtained.

[0058] From the above description, it can be seen that a method for solving the main problem of the decomposed objective function is given, and the optimal planning strategy of the objective function is calculated.

[0059] Further, the step S31 specifically includes: using the main processor to solve the linear programming relaxation problem of the main problem, obtaining the rate of change of the objective function, and transmitting the rate of change to the auxiliary processor;

[0060] The step S32 specifically includes: in the auxiliary processor, solving the sub-problem to generate a new CG column, sharing the column between different auxiliary processors and generating a new CG column, sending the CG column with a change in the objective function less than zero to the main processor, and adding it to the main problem.

[0061] From the above description, it can be seen that the main problem is solved by the main processor, and the sub-problems are solved by the auxiliary processors. Different auxiliary processors share columns and generate new columns to improve the computational efficiency of the algorithm.

[0062] Please refer to Figure 3 and Figure 4 , a flexible resource planning terminal based on the CG&S distributed computing method, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program:

[0063] Step S1, establishing a multi-stage stochastic flexibility resource planning model;

[0064] Step S2, using the Danz-Wolf formula to decompose and reformulate the objective function of the multi-stage stochastic flexibility resource planning model into a main problem and sub-problems;

[0065] Step S3: Solve the decomposed main problem and sub-problems to obtain the optimal planning strategy of the objective function.

[0066] From the above description, it can be seen that the beneficial effects of the present invention are: by establishing a multi-stage random flexibility resource planning model and using the Danz-Wolf formula to decompose and rephrase the main problem and sub-problems for solving, it is possible to solve complex resource planning problems and obtain the optimal planning strategy for flexibility resources in the power system.

[0067] Furthermore, the step S1 specifically includes:

[0068] The scenario tree is used to divide the planning cycle into different decision-making stages. The root node represents the initial state of the system. This node only makes investment decisions, while the subsequent nodes are composed of both operation and investment stages.

[0069] From the above description, it can be seen that after one stage is determined, the next stage of operation is carried out. This multi-stage approach allows investment decisions to be made at several points in time and takes into account the information of the uncertain parameter set known in the current stage, reducing the impact of uncertain parameters on planning.

[0070] Furthermore, in step S1, a flexible resource comprehensive planning model considering short-term response and long-term uncertainty is specifically established, and its objective function is described as follows:

[0071]

[0072] In the formula, The unit investment cost of the power generation unit at node n, is the unit investment cost of ESS at node n, is the unit investment cost of the line at node n, is the capacity of the additional generating unit installed in node n, is the capacity of the ESS installed in node n, is the capacity of the line installed in node n, is the unit variable cost of the power generation unit in node n, is the startup cost of the power generation unit, P n，t，g is the hourly output power of the power generation unit in node n, S n，t，g is the number of startups per hour of the power generation unit in node n, is the unit variable cost of ESS in node n, is the power provided by ESS in node n per hour, Compensate for the unserved demand in node n, is the load reduction of node n on bus b at time t, which has the following constraints:

[0073]

[0074]

[0075]

[0076]

[0077]

[0078]

[0079]

[0080]

[0081]

[0082]

[0083]

[0084]

[0085]

[0086]

[0087]

[0088]

[0089]

[0090]

[0091]

[0092] In the formula, are the total number of installed power generation, energy storage and line units in node n, respectively. are the additional power generation units, ESS and lines installed in node n. is the maximum number of resources installed in node n, F n，t，l is the power flow on line l in node n at time t, They are the self-discharge, charging efficiency and discharging efficiency of ESS per hour respectively.

[0093] From the above description, it can be seen that it gives a flexible resource comprehensive planning model that considers short-term response and long-term uncertainty. The constraints associate the investment in different resources on the predecessor node of n with the total units installed in the node n, and also limit the total number of units installed in the node n, ensuring the power balance of each hour in each bus operation phase of the system, limiting the maximum flow through the transmission line to each installed capacity, and giving a simplified model of the power generation unit. Specifically, the maximum power generation limit of the thermal generator is given, and the power of the RES generator is limited by the availability of the installed units and primary energy. The energy balance, minimum and maximum energy storage capacity, and maximum charge and discharge power constraints are considered. The number of installed units of existing assets is set at each node n and the number of existing units at the root node, which limits the installation of additional power generation, energy storage or transmission units after the root node.

[0094] Furthermore, the step S3 specifically includes:

[0095] Step S31, solving the linear programming relaxation problem of the main problem, and obtaining the rate of change of the objective function;

[0096] Step S32, solving the sub-problem according to the rate of change of the objective function to generate a new CG column and obtain the corresponding minimum change of the objective function, each CG column corresponds to a pair of feasible total number of installation units and the optimization operation of the corresponding nodes, and the column with the change of the objective function less than zero is added to the main problem;

[0097] Step S33, calculate the change of the objective function in the linear programming, and check the conditions for terminating the algorithm calculation. If the change of the objective function is less than zero, the correlation coefficient of the column variable is added to the coefficient matrix of the main problem, and return to step 31 to perform a new iteration. If not, the termination condition is met, the calculation is terminated, and the optimal planning strategy of the objective function is obtained.

[0098] From the above description, it can be seen that a method for solving the main problem of the decomposed objective function is given, and the optimal planning strategy of the objective function is calculated.

[0099] Further, the processor includes a main processor and a plurality of auxiliary processors, and the step S31 specifically includes: using the main processor to solve the linear programming relaxation problem of the main problem, obtaining the rate of change of the objective function, and transmitting the rate of change to the auxiliary processor;

[0100] The step S32 specifically includes: in the auxiliary processor, solving the sub-problem to generate a new CG column, sharing the column between different auxiliary processors and generating a new CG column, sending the CG column with a change in the objective function less than zero to the main processor, and adding it to the main problem.

[0101] From the above description, it can be seen that the main problem is solved by the main processor, and the sub-problems are solved by the auxiliary processors. Different auxiliary processors share columns and generate new columns to improve the computational efficiency of the algorithm.

[0102] The flexibility resource planning method and terminal based on the CG&S distributed computing method of the present invention are applied to the flexibility resource planning of the power system to solve the optimal planning strategy for the flexibility resources in the system.

[0103] Please refer to Figure 1 , Embodiment 1 of the present invention is: a flexibility resource planning method based on a CG&S distributed computing method, which comprises the following steps:

[0104] Step S1, establishing a multi-stage stochastic flexibility resource planning model.

[0105] Specifically, through the scenario tree-based method, the basic model is described with the objective function and constraints, and a large-scale mixed integer programming problem is obtained. In the subsequent steps, the optimal planning strategy for the flexibility resources in the system can be obtained by solving the model.

[0106] In planning problems, long-term uncertainty is mainly related to uncertain parameters such as fuel prices, operating and investment costs, and changes in load demand. The scenario tree is used to divide the planning cycle into different decision stages. The root node represents the initial state of the system, and this node only makes investment decisions. The subsequent nodes are composed of both the operation and investment stages. After one stage is determined, the next stage is operated. This multi-stage approach allows investment decisions to be made at several time points and takes into account the information of the known uncertain parameter set in the current stage, reducing the impact of uncertain parameters on planning.

[0107] Based on the above method, a flexible resource comprehensive planning model considering short-term response and long-term uncertainty is established. Its objective function is to minimize the expected sum of investment and operating costs. The description formula (1) is as follows:

[0108]

[0109] In the formula, They are the unit investment costs of the power generation unit, ESS (Energy Storage System) and line of node n respectively. is the capacity of the additional generation units, ESS and lines installed in node n. is the unit variable cost of the power generation unit in node n, is the startup cost of the power generation unit. n，t，g is the hourly output power of the power generation unit in node n, S n，t，g is the number of starts per hour of the power generation unit in node n. is the unit variable cost of ESS in node n, is the power provided by ESS in node n per hour. Compensate for the unserved demand in node n. is the load reduction of node n on bus b at time t.

[0110] Constraints are established for the system, generator sets, and energy storage. Specifically, they include system and bus power balance and transmission capacity constraints, investment constraints, power generation unit output constraints, and energy storage level and charging and discharging power constraints, as follows:

[0111]

[0112]

[0113]

[0114] The above constraints (2)-(4) are unexpected constraints. are the total number of installed power generation, energy storage and line units in node n, respectively. are the additional power generation units, ESS and lines installed in node n. The maximum number of units installed for each resource in node n. They relate the investment of different resources on n's predecessor nodes to the total units installed in node n, and also limit the total number of units installed in node n.

[0115]

[0116]

[0117] The above constraints (5)-(6) are related to the transmission network. Constraint (5) ensures the power balance of each busbar operation phase in the system every hour, and constraint (6) limits the maximum flow through the transmission line to the respective installed capacity. In the formula, F n，t，l is the power flow on line l in node n at time t.

[0118]

[0119]

[0120]

[0121]

[0122]

[0123] Among the above constraints, constraint (7) gives a simplified model of the power generation unit, specifically giving the maximum power generation limit of the thermal generator, constraint (8) gives the power of the RES generator limited by the installed units and the availability of primary energy, and the simplified model of ESSs takes into account the energy balance (9), the minimum and maximum energy storage capacity (10) and the maximum charge and discharge power constraints (11), where constraint (10) allows limiting the energy level of the energy storage to prevent accelerated degradation. They are the self-discharge, charging efficiency and discharging efficiency of ESS per hour respectively.

[0124]

[0125]

[0126]

[0127]

[0128]

[0129]

[0130]

[0131]

[0132]

[0133] Among the above constraints, constraints (12)–(14) set the number of installed units of existing assets at each node n and the number of existing units at the root node. To maintain the consistency of the model, constraints (15)–(17) set the number of additional units installed at the root node to the number of existing units. Constraints (18)–(20) restrict that no additional generation, storage or transmission units can be installed after the root node.

[0134] Since it is difficult to solve mixed integer programming problems in power systems, such optimization problems often have a limited number of constraints but a large number of variables. Therefore, all variables cannot be explicitly expressed in the model, so the simplex method cannot solve such problems.

[0135] However, in the solution process, the basis variables are only related to the number of constraints, and only one new non-basic variable will be generated in each iteration, so only a small number of variables will be involved in the entire solution process. Therefore, the column generation algorithm was generated based on the simplex method. The column generation and sharing algorithm is a very efficient algorithm for solving large-scale linear optimization problems. Its theoretical basis was proposed by Danz et al. in 1960. In essence, the column generation algorithm is a form of the simplex method. At present, there are some studies and improvements on distributed solution algorithms in China, but there are few discussions on the column generation and sharing algorithm. This method no longer needs to traverse all variables, and can even work in the case of an unknown total number of variables. As long as one or more variables can be generated in each iteration, the purpose of improving optimization can be achieved. By continuously adding new variables until there is no test number of non-basic variables less than 0, the optimal solution to the original problem is obtained.

[0136] Step S2: Use the Dantzig-Wolfe formula to decompose the objective function of the multi-stage stochastic flexibility resource planning model and reformulate it into a main problem and sub-problems.

[0137] First, the objective function of the multi-stage stochastic problem is described by the matrix (A, 1a):

[0138]

[0139] In the formula, represents the probability of node n, I n represents the additional unit vector installed at node n, X n A vector representing the variables running in node n. are the unit costs of investment and operation respectively.

[0140] Combining the investment decision made at the predecessor node of n with the total installed units in node n gives the constraint:

[0141]

[0142] In the formula, Z n is the vector of total installed units in node n.

[0143] Relating the operation decision in node n to the total installed units in node n gives the constraint:

[0144]

[0145] In the formula, A n The matrix represents the coupled operation and investment decisions in node n. Summarizing the maximum investment limit gives the constraint:

[0146]

[0147] In the formula, is the maximum number of units that can be installed in node n.

[0148] Summarize the operations that are not explicitly dependent on the total installed base to give the constraints:

[0149]

[0150] Where, X n A vector representing the running variables of node n, χ n is a set of feasible action decisions in node n.

[0151] Give the relevant constraints that describe the completeness of the investment:

[0152]

[0153] In the formula, I n represents the vector of additional installed units in node n, G, ε, and L are a set of generators, ESS, and transmission lines, respectively.

[0154]

[0155]

[0156]

[0157] The above constraints (A.1g)-(A.1i) summarize the constraints on the investment decision of existing resources. 0 A 0-1 variable indicating whether the resource already exists or is a candidate asset.

[0158] The problem is then reformulated using a discretization approach. To this end, the feasible area of ​​the total installed units (including generators, ESSs and lines) in node n is defined as:

[0159]

[0160] For every feasible vector of total units installed in node n There is at least one associated optimal operation plan Therefore X n It can be expressed as:

[0161]

[0162] The main problem (A.4a) of the Danz-Wolff decomposition is obtained by substituting Zn and Xn in (A.1a) into (A.2) and (A.3) respectively. A discrete variable representing installation or non-installation. Constraint (A.4b) is equivalent to constraint (A.1b). Constraints (A.4c) and (A.4d) are to ensure that each node in the scene tree selects only one vector of the total installation unit, that is, only one operation plan. The objective function change rates of the relevant constraints (A.4b) and (A.4c) are πn and μn, respectively.

[0163]

[0164]

[0165]

[0166]

[0167]

[0168] diag(U 0 )I1=Z 0 (A.4f);

[0169]

[0170] After the solution is decomposed, the decomposed problem can be solved and calculated to obtain the expected investment, operation and total costs, load shortfall and renewable energy reduction, and the expected installed capacity of each planning model.

[0171] Step S3: Solve the decomposed main problem and sub-problems to obtain the optimal planning strategy of the objective function.

[0172] Please refer to Figure 2 , specifically including:

[0173] Step S31, using the main processor to solve the linear programming (LP) relaxation problem of the main problem. At this time, each node will obtain the rate of change of the objective function, and this rate of change will be transmitted to the auxiliary processor at the same time for solving the sub-problem.

[0174] Step S32: In the auxiliary processor, solve the sub-problem to generate a new CG column. The generation of this column will simultaneously produce the minimum objective function change (test amount). In this embodiment, one column corresponds to a pair of optimization operations of the total number of feasible installation units and their corresponding nodes.

[0175] Different auxiliary processors share columns and generate new columns to improve the computational efficiency of the algorithm. Columns with objective function changes less than zero are sent to the main processor and added to the main problem.

[0176] Step S33, calculate the change of the objective function in the linear programming, and check the conditions for terminating the algorithm calculation. If the change of the objective function is less than zero, the correlation coefficient of the column variable is added to the coefficient matrix of the main problem, and return to step 31 to perform a new iteration. If not, the termination condition is met, the calculation is terminated, and the optimal planning strategy of the objective function is obtained.

[0177] The column generation and sharing algorithm proposed in this paper is also superior to the existing solution algorithms in terms of computational performance. Usually, the problem to be solved is a mixed integer programming problem, and the size of the problem will increase with the increase of the solution time scale, making the problem more complicated. In the process of solving the existing algorithm, the size of the sub-problem will increase linearly or even exponentially, which increases the difficulty of solving. However, the method proposed in this invention uses the scene tree to consider long-term uncertainty, so the size of the sub-problem can be guaranteed to remain unchanged. Both the solution time and the number of iterations can illustrate the advantages of using the novel column generation and sharing method. As the number of stages increases, the size of the problem also increases, and the proposed method will be significantly faster than other decomposition methods. Due to the large number of nodes in the scene tree and investment decision, general methods usually require a large number of iterations to converge. In contrast, this phenomenon is improved when the column sharing program is included, because the column sharing program is effective in generating new feasible columns for each node of the scene tree in each iteration. These additional columns add more information to the main problem and increase the rate of change of the objective function. In addition, adding these columns can quickly reduce the upper limit of the objective function while increasing convergence.

[0178] Please refer to Figure 3 , Embodiment 2 of the present invention is:

[0179] A flexible resource planning terminal 1 based on a CG&S distributed computing method includes a memory 3, a processor 2, and a computer program stored in the memory 3 and executable on the processor 2. Figure 4 The processor specifically includes a main processor 21, a first auxiliary processor 22 and a second auxiliary processor 23. The main processor 21, the first auxiliary processor 22 and the second auxiliary processor communicate with each other.

[0180] The main processor 21, the first auxiliary processor 22 and the second auxiliary processor 23 implement the steps of the above-mentioned embodiment one when executing the computer program, wherein the main processor 21 implements steps S1, S2, S31 and S33 of the above-mentioned embodiment one when executing the computer program, and the first auxiliary processor 22 and the second auxiliary processor 23 implement step S32 of the above-mentioned embodiment one when executing the computer program.

[0181] In summary, the flexibility resource planning method and terminal based on the CG&S distributed computing method provided by the present invention achieve the solution of complex resource planning problems by establishing a multi-stage random flexibility resource planning model and using the Danz-Wolf formula to decompose and rephrase the main problem and sub-problems for solution. The main problem is solved on the main processor, and the sub-problems are solved on the auxiliary processor. Different auxiliary processors share columns and generate new columns to improve the computational efficiency of the algorithm. The scene tree method is adopted to describe the basic model with objective functions and constraints, reduce the influence of uncertainty parameters on planning, and provide an example resource planning objective function that takes short-term response and long-term uncertainty into consideration.

[0182] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A flexibility resource planning method based on CG&S distributed computing method, characterized in that: Includes steps: Step S1, establishing a multi-stage stochastic flexibility resource planning model; Step S2, using the Danz-Wolf formula to decompose and reformulate the objective function of the multi-stage stochastic flexibility resource planning model into a main problem and sub-problems; Step S3, solving the decomposed main problem and sub-problems to obtain the optimal planning strategy of the objective function; The step S1 specifically includes: The scenario tree is used to divide the planning cycle into different decision stages. The root node represents the initial state of the system. This node only makes investment decisions, while the subsequent nodes are composed of both operation and investment stages. The step S1 specifically establishes a flexible resource comprehensive planning model that considers short-term response and long-term uncertainty, and its objective function is described as follows: In the formula, is the unit investment cost of the power generation unit at node n, is the unit investment cost of ESS at node n, is the unit investment cost of the line at node n, is the capacity of the additional generating unit installed in node n, is the capacity of the ESS installed in node n, is the capacity of the line installed in node n, is the unit variable cost of the power generation unit in node n, is the startup cost of the power generation unit, P n,t,g is the hourly output power of the power generation unit in node n, S n,t,g is the number of startups per hour of the power generation unit in node n, is the unit variable cost of ESS in node n, is the power provided by ESS in node n per hour, Compensate for the unserved demand in node n, is the load reduction of node n on bus b at time t, which has the following constraints: In the formula, are the total number of installed power generation, energy storage and line units in node n, respectively. are the additional power generation units, ESS and lines installed in node n. is the maximum number of resources installed in node n, F n，t，l is the power flow on line l at time t, They are the self-discharge, charging efficiency and discharging efficiency of ESS per hour respectively; The step S3 specifically includes: Step S31, solving the linear programming relaxation problem of the main problem, and obtaining the rate of change of the objective function; Step S32, solving the sub-problem according to the rate of change of the objective function to generate a new CG column and obtain the corresponding minimum change of the objective function, each CG column corresponds to a pair of feasible total number of installation units and the optimization operation of the corresponding nodes, and the column with the change of the objective function less than zero is added to the main problem; Step S33, calculate the change of the objective function in the linear programming, and check the conditions for terminating the algorithm calculation; if the change of the objective function is less than zero, add the correlation coefficients of the column variables to the coefficient matrix of the main problem, and return to step 31 to perform a new iteration; if the opposite is true, the termination condition is met, the calculation is terminated, and the optimal planning strategy for the objective function is obtained.

2. The flexibility resource planning method based on the CG&S distributed computing method according to claim 1 is characterized in that: The step S31 specifically includes: using the main processor to solve the linear programming relaxation problem of the main problem, obtaining the change rate of the objective function, and transmitting the change rate to the auxiliary processor; The step S32 specifically includes: in the auxiliary processor, solving the sub-problem to generate a new CG column, sharing the column between different auxiliary processors and generating a new CG column, sending the CG column with a change in the objective function less than zero to the main processor, and adding it to the main problem.

3. A flexible resource planning terminal based on a CG&S distributed computing method, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the following steps are implemented: Step S1, establishing a multi-stage stochastic flexibility resource planning model; Step S2, using the Danz-Wolf formula to decompose and reformulate the objective function of the multi-stage stochastic flexibility resource planning model into a main problem and sub-problems; Step S3, solving the decomposed main problem and sub-problems to obtain the optimal planning strategy of the objective function; The step S1 specifically includes: The scenario tree is used to divide the planning cycle into different decision stages. The root node represents the initial state of the system. This node only makes investment decisions, while the subsequent nodes are composed of both operation and investment stages. The step S1 specifically establishes a flexible resource comprehensive planning model that considers short-term response and long-term uncertainty, and its objective function is described as follows: In the formula, is the unit investment cost of the power generation unit at node n, is the unit investment cost of ESS at node n, is the unit investment cost of the line at node n, is the capacity of the additional generating unit installed in node n, is the capacity of the ESS installed in node n, is the capacity of the line installed in node n, is the unit variable cost of the power generation unit in node n, is the startup cost of the power generation unit, P n，t，g is the hourly output power of the power generation unit in node n, S n，t，g is the number of startups per hour of the power generation unit in node n, is the unit variable cost of ESS in node n, is the power provided by ESS in node n per hour, Compensate for the unserved demand in node n, is the load reduction of node n on bus b at time t, which has the following constraints: In the formula, are the total number of installed power generation, energy storage and line units in node n, respectively. are the additional power generation units, ESS and lines installed in node n. is the maximum number of resources installed in node n, F n，t，l is the power flow on line l at time t, They are the self-discharge, charging efficiency and discharging efficiency of ESS per hour respectively; The step S3 specifically includes: Step S31, solving the linear programming relaxation problem of the main problem, and obtaining the rate of change of the objective function; Step S32, solving the sub-problem according to the rate of change of the objective function to generate a new CG column and obtain the corresponding minimum change of the objective function, each CG column corresponds to a pair of feasible total number of installation units and the optimization operation of the corresponding nodes, and the column with the change of the objective function less than zero is added to the main problem; Step S33, calculate the change of the objective function in the linear programming, and check the conditions for terminating the algorithm calculation; if the change of the objective function is less than zero, add the correlation coefficients of the column variables to the coefficient matrix of the main problem, and return to step 31 to perform a new iteration; if the opposite is true, the termination condition is met, the calculation is terminated, and the optimal planning strategy for the objective function is obtained.

4. The flexible resource planning terminal based on the CG&S distributed computing method according to claim 3, characterized in that: The processor includes a main processor and a plurality of auxiliary processors, and the step S31 specifically includes: using the main processor to solve the linear programming relaxation problem of the main problem, obtaining the change rate of the objective function, and transmitting the change rate to the auxiliary processor; The step S32 specifically includes: in the auxiliary processor, solving the sub-problem to generate a new CG column, sharing the column between different auxiliary processors and generating a new CG column, sending the CG column with a change in the objective function less than zero to the main processor, and adding it to the main problem.

Images