Distribution network probabilistic planning method and device considering multi-time scale uncertainty

Abstract

The present invention discloses a distribution network probabilistic planning method and device considering multi-time scale uncertainty. The method comprises: establishing a multi-stage hybrid scenario tree about multi-time scale uncertainty factors based on the uncertainty factors of the distribution network at different times; establishing an objective function of a multi-stage probabilistic planning model with the minimum comprehensive cost under multiple scenarios within the planning period as the goal according to the multi-stage hybrid scenario tree; establishing constraints on distribution network planning investment and operation, and obtaining a distribution network probabilistic planning model according to the objective function and the constraints; solving the distribution network probabilistic planning model based on an improved Benders decomposition algorithm to obtain a probabilistic planning scheme.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network planning, and in particular to a distribution network probabilistic planning method and device considering multi-time scale uncertainties. Background Art

[0002] The future development trend of the power grid will fully integrate the large-scale development of clean energy power generation such as solar energy and wind turbines, deeply promote the integration of new energy vehicles into urban transportation, power grids and other fields, and improve the carrying capacity and flexibility of distribution networks for new energy and flexible loads. In this context, the access of large-scale new energy and flexible loads is an inevitable trend in the development of high-voltage distribution networks in the future. The disorderly access of large-scale new energy power generation and flexible loads has changed the conventional operating characteristics of high-voltage distribution networks, making it difficult to predict power demand, increasing the operating risks of high-voltage distribution network planning schemes, and greatly increasing the difficulty of accurate and efficient system planning. The construction period of high-voltage distribution networks is long, and during the planning period, they will face multi-time scale uncertainties such as dynamic growth of sources and loads and fluctuations in electricity prices, which have an important impact on the final plan. How to properly handle the multivariate uncertainties at multiple time scales is a key issue that needs to be solved in the planning of new high-voltage distribution networks in the future.

[0003] Patent document CN110414836A discloses a probabilistic substation planning method that takes into account the uncertainty of electricity demand of integrated energy stations: statistics on the load value of conventional electricity load in the target year in the planned area, determine the total electricity load in the target year in the planned area; determine the number of new substations and several schemes of capacity combination of all new substations; determine the initial site of each new substation; obtain the new power supply range of each substation; obtain the final result of substation planning under the first scheme in the new substation capacity combination scheme, and calculate the minimum cost required; determine whether all new substation capacity combination schemes have been traversed, if all have been traversed, take the substation planning result under the scheme with the lowest cost in the new substation capacity combination scheme as the final substation planning result of the planning area. However, this method does not take into account the uncertainty of load and has low accuracy. Summary of the invention

[0004] The present invention provides a distribution network probabilistic planning method and device taking into account multi-time scale uncertainties, which can improve the accuracy of distribution network probabilistic planning.

[0005] A distribution network probabilistic planning method considering multi-time scale uncertainties includes:

[0006] Based on the uncertainties of the distribution network at different times, a multi-stage hybrid scenario tree of multi-time scale uncertainties is established.

[0007] According to the multi-stage hybrid scenario tree, an objective function is established with the goal of minimizing the comprehensive cost under multiple scenarios within the planning period;

[0008] Establishing constraints on distribution network planning, investment and operation, and obtaining a distribution network probabilistic planning model based on the objective function and the constraints;

[0009] The distribution network probabilistic planning model is solved based on the improved Benders decomposition algorithm to obtain a probabilistic planning solution.

[0010] Furthermore, the uncertainties at different times include uncertainties in the planning stage and uncertainties within a preset time; the uncertainties in the planning stage include load growth and electricity price status; and the uncertainties within the preset time include wind power load and photovoltaic load.

[0011] Furthermore, based on the long-term uncertainties of the distribution network, a multi-stage hybrid scenario tree of multi-time scale uncertainties is established, including:

[0012] Determine the planning stage of the distribution network;

[0013] The current state of the distribution network is taken as the root node, the load growth and electricity price state in each planning stage are taken as nodes, the planning path between the planning stages are taken as line segments, and the root node and each node are connected through the line segments;

[0014] The Markov chain method is used to obtain the probability transfer characteristics of the connected nodes between different stages and the occurrence probability of the planned path;

[0015] At each node, scenarios are clustered based on wind power load and photovoltaic load to form a typical scenario daily set.

[0016] Furthermore, the typical scenario set includes wind power load and photovoltaic load under typical scenario days;

[0017] At each node, scenarios are clustered based on wind power load and photovoltaic load to form a typical scenario set, including:

[0018] At each node, the Weibull model parameters are obtained based on the historical wind speed data to generate simulated wind speed, and the photovoltaic Beta model parameters are obtained based on the historical photovoltaic data to generate simulated photovoltaic power;

[0019] The simulated wind speed and the simulated photovoltaic power within a preset time period are taken as a sample set, and a clustering algorithm is used to perform a clustering operation on the sample set to obtain a typical scene set.

[0020] Furthermore, the objective function is as follows:

[0021]

[0022] Wherein, the subscript l represents the planning path between planning stages, and Ω l represents the set of planned paths, Ω t represents the set of planning stages, ρ l represents the occurrence probability of the planned path l, f l represents the planning cost under planning path l, γt represents the initial year of planning stage t, ψ represents the annualized interest rate, They represent the equipment investment cost, equipment maintenance cost, daily operation cost and power failure cost of the high-voltage distribution network in the planning stage t respectively, and f represents the comprehensive cost.

[0023] Furthermore, the constraints on distribution network planning, investment and operation include equipment construction constraints, power flow constraints, grid radial constraints and upper and lower limit constraints of operating variables.

[0024] Furthermore, the distribution network probability planning model is solved based on the improved Benders decomposition algorithm to obtain a probability planning scheme, including:

[0025] converting the distribution network probabilistic programming model into a compact form;

[0026] Based on the compact form, the distribution network probabilistic programming model is decoupled into an investment main problem and an operation sub-problem;

[0027] Introduce slack variables into the subproblem as penalty terms, obtain the relaxed running subproblem, and form a unified dual cut;

[0028] Solve the main problem and record the planning variables;

[0029] Feeding back the planning variables to the relaxed running subproblem and recording the unified dual cut;

[0030] It is determined whether the upper bound and the lower bound formed by the main problem and the relaxed running subproblem converge. If so, the iteration is stopped; otherwise, the main problem is solved again.

[0031] Further, the compact form is as follows:

[0032]

[0033] The main problem is as follows:

[0034]

[0035] The relaxed running subproblem is as follows:

[0036]

[0037] The unified dual cut is as follows:

[0038]

[0039] Where x and y represent investment 0-1 variables and operation continuous variables, respectively; b, c, d, e, A, B, and D represent coefficient matrices; the superscript T represents the transposed matrix; and the subscript s represents the number of the typical scenario day. is the intermediate variable, μ s and λ s represents the dual variable in the iteration process, μ s Used to generate the optimal dual cut, λ s Used to generate feasible dual cuts. The superscript * indicates the optimal value during the iteration. s Indicates the relaxation of constraints when the subproblem is infeasible, d s represents the coefficient matrix of a typical scenario day s, x s represents the investment 0-1 variable on the typical scenario day s, y s represents the running continuous variable of the typical scenario day s, M is a positive integer, v s represents the slack variable, A s represents the coefficient matrix of a typical scenario day s, B s represents the coefficient matrix of the typical scenario day s, b s Represents the coefficient matrix of the typical scenario day s, π s Represents the dual variable of the unified dual cut under the typical scenario day s.

[0040] A distribution network probabilistic planning device considering multi-time scale uncertainties, comprising:

[0041] The scenario building module builds a multi-stage hybrid scenario tree for multi-time scale uncertainty factors based on the uncertainty factors of the distribution network at different times;

[0042] A target establishment module, used to establish an objective function with the goal of minimizing the comprehensive cost under multiple scenarios within the planning period according to the multi-stage hybrid scenario tree;

[0043] A planning module, used to establish the constraints of distribution network planning investment and operation, and obtain a distribution network probabilistic planning model according to the objective function and the constraints;

[0044] The analytical module solves the distribution network probability planning model based on the improved Benders decomposition algorithm to obtain a probability planning solution.

[0045] Furthermore, the uncertainties at different times include uncertainties in the planning stage and uncertainties within a preset time; the uncertainties in the planning stage include load growth and electricity price status; and the uncertainties within the preset time include wind power load and photovoltaic load.

[0046] Furthermore, the scenario building module builds a multi-stage hybrid scenario tree of multi-time scale uncertainty factors based on the uncertainty factors of the distribution network at different times, including:

[0047] Determine the planning stage of the distribution network;

[0048] The current state of the distribution network is taken as the root node, the load growth and electricity price state in each planning stage are taken as nodes, the planning path between the planning stages are taken as line segments, and the root node and each node are connected through the line segments;

[0049] The Markov chain method is used to obtain the probability transfer characteristics of the connected nodes between different stages and the occurrence probability of the planned path;

[0050] At each node, scenarios are clustered based on wind power load and photovoltaic load to form a typical scenario set.

[0051] Furthermore, the typical scenario set includes wind power load and photovoltaic load under typical scenario days;

[0052] The scenario building module clusters scenarios based on wind power load and photovoltaic load at each node to form a typical scenario set, including:

[0053] At each node, the Weibull model parameters are obtained based on the historical wind speed data to generate simulated wind speed, and the photovoltaic Beta model parameters are obtained based on the historical photovoltaic data to generate simulated photovoltaic power;

[0054] The simulated wind speed and the simulated photovoltaic power within a preset time period are taken as a sample set, and a clustering algorithm is used to perform a clustering operation on the sample set to obtain a typical scene set.

[0055] Furthermore, the objective function is as follows:

[0056]

[0057] Wherein, the subscript l represents the planning path between planning stages, and Ω l represents the set of planned paths, Ω t represents the set of planning stages, ρ l represents the occurrence probability of the planned path l, f l represents the planning cost under planning path l, γt represents the initial year of planning stage t, ψ represents the annualized interest rate, They represent the equipment investment cost, equipment maintenance cost, daily operation cost and power failure cost of the high-voltage distribution network in the planning stage t respectively, and f represents the comprehensive cost.

[0058] Furthermore, the constraints on distribution network planning, investment and operation include equipment construction constraints, power flow constraints, grid radial constraints and upper and lower limit constraints of operating variables.

[0059] Furthermore, the analytical module solves the distribution network probability planning model based on the improved Benders decomposition algorithm to obtain a probability planning scheme, including:

[0060] converting the distribution network probabilistic programming model into a compact form;

[0061] Based on the compact form, the distribution network probabilistic programming model is decoupled into an investment main problem and an operation sub-problem;

[0062] Introduce slack variables into the subproblem as penalty terms, obtain the relaxed running subproblem, and form a unified dual cut;

[0063] Solve the main problem and record the planning variables;

[0064] Feeding back the planning variables to the relaxed running subproblem and recording the unified dual cut;

[0065] It is determined whether the upper bound and the lower bound formed by the main problem and the relaxed running subproblem converge. If so, the iteration is stopped; otherwise, the main problem is solved again.

[0066] Further, the compact form is as follows:

[0067] min x∈{0,1},y {c T x+d T y:Dx≤e,Ax+By≤b};

[0068] The main problem is as follows:

[0069]

[0070] The relaxed running subproblem is as follows:

[0071]

[0072] The unified dual cut is as follows:

[0073]

[0074] Where x and y represent investment 0-1 variables and operation continuous variables, respectively; b, c, d, e, A, B, and D represent coefficient matrices; the superscript T represents the transposed matrix; and the subscript s represents the number of the typical scenario day. is the intermediate variable, μ s and λ s represents the dual variable in the iteration process, μ sUsed to generate the optimal dual cut, λ s Used to generate feasible dual cuts. The superscript * indicates the optimal value during the iteration. s Indicates the relaxation of constraints when the subproblem is infeasible, d s represents the coefficient matrix of a typical scenario day s, x s represents the investment 0-1 variable on the typical scenario day s, y s represents the running continuous variable of the typical scenario day s, M is a positive integer, v s represents the slack variable, A s represents the coefficient matrix of a typical scenario day s, B s represents the coefficient matrix of the typical scenario day s, b s Represents the coefficient matrix of the typical scenario day s, π s Represents the dual variable of the unified dual cut under the typical scenario day s.

[0075] A computer storage medium stores computer instructions, which are used to execute the above method when called.

[0076] The distribution network probabilistic planning method and device considering multi-time scale uncertainty provided by the present invention have at least the following beneficial effects:

[0077] By considering uncertainties at multiple time scales, long-term planning uncertainties and short-term operating uncertainties can be coordinated and combined; multi-scenario hybrid trees can cover a wider range of future scenarios, thereby improving the comprehensiveness and reliability of planning results; the planning scheme obtained by the multi-time scale probabilistic planning model can better cope with the multiple uncertainties faced by the high-voltage distribution network during the planning period, and improve the economy, flexibility and adaptability of the planning scheme; the improved solution algorithm can improve the solution efficiency, shorten the calculation time, and make it possible to solve large-scale problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 A flow chart of an embodiment of a distribution network probabilistic planning method considering multi-time scale uncertainties provided by the present invention.

[0079] Figure 2 A flowchart of an embodiment of establishing a multi-stage hybrid scenario tree in a distribution network probabilistic planning method considering multi-time scale uncertainties provided by the present invention.

[0080] Figure 3 A schematic diagram of an embodiment of a multi-stage hybrid scenario tree in a distribution network probabilistic planning method considering multi-time scale uncertainties provided by the present invention.

[0081] Figure 4A flow chart of an embodiment of solving a distribution network probabilistic planning model in a distribution network probabilistic planning method considering multi-time scale uncertainties provided by the present invention.

[0082] Figure 5 A schematic structural diagram of an embodiment of a distribution network probabilistic planning device considering multi-time scale uncertainties provided by the present invention. DETAILED DESCRIPTION

[0083] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0084] refer to Figure 1 In some embodiments, a distribution network probabilistic planning method considering multi-time scale uncertainties is provided, comprising:

[0085] S1. Based on the uncertainties of the distribution network at different times, a multi-stage hybrid scenario tree of multi-time scale uncertainty factors is established;

[0086] S2. Establishing an objective function with the goal of minimizing the comprehensive cost under multiple scenarios within the planning period according to the multi-stage hybrid scenario tree;

[0087] S3, establishing constraints on distribution network planning, investment and operation, and obtaining a distribution network probabilistic planning model based on the objective function and the constraints;

[0088] S4. Solve the distribution network probability planning model based on the improved Benders decomposition algorithm to obtain a probability planning solution.

[0089] Specifically, the probabilistic planning in this embodiment targets multi-time scale uncertainties, including a long planning stage and a short time scale within a preset time. Both uncertainties have probabilistic characteristics, the probability of occurrence of the planned path and the probability of occurrence of the typical scenario on a daily basis. The obtained probabilistic planning scheme includes the probability of occurrence of the planned path and the probability of occurrence of the typical scenario on a daily basis.

[0090] Specifically, in step S1, the uncertainties at different times include uncertainties in the planning stage and uncertainties within a preset time; the uncertainties in the planning stage include load growth and electricity price status; the uncertainties within the preset time include wind power load and photovoltaic load.

[0091] refer to Figure 2 In step S1, based on the uncertainties of the distribution network at different times, a multi-stage hybrid scenario tree of multi-time scale uncertainties is established, including:

[0092] S11, determine the planning stage of the distribution network;

[0093] S12, taking the current state of the distribution network as the root node, taking the load growth and electricity price state in each planning stage as nodes, taking the planning path between the planning stages as line segments, and connecting the root node and each node through the line segments;

[0094] S13, using the Markov chain method to obtain the probability transfer characteristics of the connected nodes between different stages, and obtain the occurrence probability of the planned path;

[0095] S14. At each node, scenarios are clustered based on wind power load and photovoltaic load to form a typical scenario set.

[0096] Specifically, refer to Figure 3 , taking the planning period as the long time scale, the uncertainty of load growth and electricity price state under the long time scale within the planning period is modeled by a multi-stage scenario tree. The scenario tree consists of multiple nodes P and line segments l; the planning path label of the planning stage is represented by l, and the planning stage is represented by t. The root node of the scenario tree contains a node, which represents the current state. Each subsequent stage corresponds to one or more nodes. Each node in each stage t represents a load growth / electricity price state and is connected to the next node through a line segment. The Markov chain method is used to obtain the probability transfer characteristics of the connected nodes between different stages. The sum of the probabilities of the subsequent connection branches of each node is equal to 1, and the number of planning paths is equal to the number of nodes in the termination stage.

[0097] In step S13, the Markov chain method is used to obtain the probability transfer characteristics of the connected nodes between different stages, and the occurrence probability of the planned path is obtained, including:

[0098] S13a, define the state space and transition probability matrix of the Markov chain, where the state space is all possible nodes and the transition probability matrix is ​​the probability of transitioning from one node to another;

[0099] S13b. Calculate the probability of occurrence of the planned path between connected nodes. For each possible planned path, the probability of occurrence of the planned path is the product of the transition probability of each step on the planned path.

[0100] In step S14, the typical scenario set includes wind power load and photovoltaic load on a typical scenario day; at each node, scenario clustering is performed based on wind power load and photovoltaic load to form a typical scenario set, including:

[0101] At each node, the Weibull model parameters are obtained based on the historical wind speed data to generate simulated wind speed, and the photovoltaic Beta model parameters are obtained based on the historical photovoltaic data to generate simulated photovoltaic power;

[0102] The simulated wind speed and the simulated photovoltaic power within a preset time period are taken as a sample set, and a clustering algorithm is used to perform a clustering operation on the sample set to obtain a typical scene set.

[0103] Among them, at each node, the uncertainty of source and load under short time scale is modeled, the Weibull model parameters are obtained according to the historical wind speed data, and the simulated wind speed is generated. The photovoltaic Beta model parameters are obtained according to the historical photovoltaic data, and the simulated photovoltaic power is generated.

[0104] The uncertainty of source load in short time scale is modeled, and the Weibull model parameters are obtained according to the historical wind speed data to generate simulated wind speed. The Weibull distribution probability density of wind turbine output is shown as follows:

[0105]

[0106] Where P is the output power of the wind turbine, k is the shape factor, c is the scale factor, and F(P,c,k) represents the Weibull distribution probability density of the wind turbine output.

[0107] Based on the historical data of wind turbines, the shape parameter k and scale parameter c are estimated by the maximum likelihood estimation method to obtain the probability distribution of wind speed and then the simulated wind speed.

[0108] The photovoltaic Beta model parameters are obtained based on the photovoltaic historical data to generate simulated photovoltaic power. The beta distribution probability density of photovoltaic output is shown as follows:

[0109]

[0110] Where P is the photovoltaic output power, P max is the maximum output power of photovoltaic power; α and β are shape parameters, Γ is the gamma function, and f(E) represents the distribution probability of photovoltaic power output.

[0111] Based on the historical photovoltaic data, the shape parameters α and β are estimated by the maximum likelihood estimation method to obtain the distribution probability of photovoltaic output and then the simulated photovoltaic power.

[0112] In some embodiments, scene clustering can be performed on the wind and solar load data of 8760 hours, and the clustering data set S contains n samples, that is, S = {x1, x2, ..., x n}, using the k-means algorithm to form a typical scene set {x1,x2,…,x s}, the typical scene day number is represented by s.

[0113] A clustering algorithm is used to perform a clustering operation on the sample set to obtain a typical scene set, including:

[0114] 1) Initialize the center point:

[0115] Randomly select N s points as the initial cluster centers, and for each data point in the data set, calculate the Euclidean distance between it and each cluster center.

[0116] 2) Allocate clusters:

[0117] Assign each sample point to the cluster to which the cluster center closest to it belongs.

[0118] 3) Update the center point:

[0119] Recalculate the center point of each cluster, and the new cluster center is the mean of all sample points in the cluster.

[0120] 4) Repeat:

[0121] Repeat steps 2) and 3) until the cluster centers no longer change significantly (ie, the cluster allocation no longer changes) or a predetermined number of iterations is reached.

[0122] The final distribution of cluster center sample points is the typical daily scene of K-means clustering.

[0123] Furthermore, in step S2, the objective function of the multi-stage probabilistic programming model is as follows:

[0124]

[0125]

[0126] Wherein, the subscript l represents the planning path between planning stages, and Ω l represents the set of planned paths, Ω t represents the set of planning stages, ρ l represents the occurrence probability of the planned path l, f l represents the planning cost under planning path l, γt represents the initial year of planning stage t, ψ represents the annualized interest rate, They represent the equipment investment cost, equipment maintenance cost, daily operation cost and power failure cost of the high-voltage distribution network in stage t respectively, and f represents the comprehensive cost.

[0127] Specifically, the objective function of the multi-stage probabilistic planning model of the high-voltage distribution network includes equipment investment cost, equipment maintenance cost, daily operation cost and power outage cost. The specific expression is as follows:

[0128]

[0129] Among them, the subscripts t, i, k, a, s, and τ represent the planning stage, high-voltage distribution network node, type of equipment to be constructed, high-voltage distribution line, typical daily operation scenario, and intraday scheduling period, respectively; SS, L, PV, WT, and ES represent substations, high-voltage distribution lines, photovoltaics, wind turbines, and energy storage devices, respectively; represents the equipment investment cost of the high-voltage distribution network at the planning stage t, represents the equipment maintenance cost of the high-voltage distribution network at planning stage t, represents the daily operating cost of the high-voltage distribution network at planning stage t, represents the power outage cost of the high-voltage distribution network at planning stage t, Ω SS represents the set of substation nodes, Ω L represents the set of high-voltage distribution lines, Ω PV represents the set of photovoltaic nodes, Ω WT represents the set of wind turbine nodes, Ω ES represents the set of energy storage nodes, Ω PL represents the set of load nodes, K L Represents the type set of high-voltage distribution lines to be constructed, represents the 0-1 investment variable of the ith substation in planning stage t, represents the unit investment cost of the ith substation, represents the 0-1 investment variable of the a-th high-voltage distribution line, represents the unit investment cost of the transformer, l a Indicates the length of the high-voltage distribution line, represents the investment capacity of the ith PV node in planning stage t, represents the investment capacity of the ith PV node in the t-1th planning stage, Represents the unit investment cost of photovoltaic equipment, represents the investment capacity of the i-th wind turbine in planning stage t, represents the investment capacity of the i-th wind turbine in the t-1-th planning stage, represents the unit investment cost of the fan equipment, represents the investment capacity of the i-th energy storage device in the t-th planning stage, represents the investment capacity of the i-th energy storage device in the t-1-th planning stage, represents the unit investment cost of the energy storage device, represents the maintenance cost of the ith substation, represents the maintenance cost of the PV node in the tth planning stage, represents the maintenance cost of the wind turbine in the tth planning stage, represents the maintenance cost of the energy storage device in the tth planning stage, D srepresents the number of typical operating days s, Δ τ represents the length of the time period τ, represents the unit operating cost of the substation, Represents the active power of the substation, C O,PV represents the unit operating cost of the photovoltaic node, Represents the active power of photovoltaic, C O,WT represents the unit operating cost of the fan, Indicates the active power of the fan, C O,ES represents the unit operating cost of the energy storage device, Indicates the active power of the energy storage device charging, Represents the active power of the energy storage device discharged, C U,D represents the unit reduction cost of the load, Indicates load shedding power, C U,PV represents the unit reduction cost of photovoltaics, Indicates the abandoned optical power, C U,WT represents the unit reduction cost of the fan, Indicates the wind power abandoned.

[0130] Furthermore, in step S3, the constraints for distribution network planning investment and operation include equipment construction constraints, power flow constraints, grid radial constraints, and upper and lower limit constraints of operating variables.

[0131] Among them, the equipment construction constraints of the high-voltage distribution network probabilistic planning model must meet:

[0132]

[0133] In the formula, represents the 0-1 investment variable of the ith substation in planning stage t, represents the 0-1 investment variable of the a-th high-voltage distribution line, {PV, WT, ES} represents the set of photovoltaic, wind turbine, and energy storage equipment, σ represents the equipment number, represents the equipment investment capacity in the tth planning stage, Indicates the upper limit of equipment construction capacity, represents the equipment investment capacity in the t-1th planning stage. Equations (7) and (8) indicate that substations and lines can only be built once during the planning period, Equation (9) represents the inter-stage connection constraints of photovoltaic, wind turbines, and energy storage, and Equation (10) represents the upper limit constraints on the investment capacity of photovoltaic, wind turbines, and energy storage.

[0134] The power flow constraints of the probabilistic planning model of high-voltage distribution network must meet the following requirements:

[0135]

[0136]

[0137] In the formula, R a , X a Respectively represent the resistance and reactance of high-voltage distribution line a, They represent the set of high-voltage lines with i as the starting node and the ending node respectively. Indicates the load power factor, 0-1 variables representing the forward and reverse operation of the high voltage line, Indicates the line active power, Indicates the active power of the electrical load, Respectively represent the reactive power of substation and line, U itsτ , U jtsτ Represent the square values ​​of the voltages at nodes i and j, respectively. Represents the square value of the line current, M represents a large positive integer, Indicates the active power of the substation, Represents the active power of photovoltaic, Indicates the active power of the fan, Indicates the active power of the energy storage device charging, Indicates the active power discharged by the energy storage device, Represents the load shedding power. The power flow constraints of the high-voltage distribution network branch include node active power balance constraints (11), reactive power balance constraints (12), line voltage drop constraints (13) and line power balance constraints (14).

[0138] The grid radial constraints of the probabilistic planning model of high-voltage distribution network must meet the following requirements:

[0139]

[0140] In the formula, υ represents the planning stage, 0-1 variables representing the forward and reverse operation of the high voltage line, Represents the 0-1 investment variable of the a-th high-voltage distribution line.

[0141] High-voltage distribution networks usually use the design concept of closed-loop design and open-loop operation. In daily operation, the grid needs to satisfy the radial state. The specific constraints are as follows: Equation (15) indicates that unbuilt lines cannot be put into operation. Equations (16)-(17) use a tree graph structure to constrain the radial state of the grid. Equation (16) indicates that nodes outside the substation are leaf nodes with power inflow. Equation (17) indicates that the substation is the root node with no power inflow.

[0142] The upper and lower limit constraints of the variables in the high-voltage distribution network probabilistic planning model must meet the following requirements:

[0143]

[0144] In the formula, Indicates the active power of the substation, represents the 0-1 investment variable of the ith substation in planning stage t, Indicates the transformer capacity of type k, U, Respectively represent the lower and upper limits of the node voltage, U itsτ represents the square value of the voltage at node i, represents the square value of the line current, Indicates the upper limit of high voltage line current. represents the 0-1 investment variable of the a-th high-voltage distribution line, They represent the output ratio of photovoltaic and wind turbines in the typical day period s, represents the power ratio of the load in the typical day s period τ, ξ PV , WT Represent the power factors of photovoltaic and wind turbine respectively, Indicates the maximum load reduction rate, Represent the reduction of photovoltaic and wind turbines, P ES Indicates the upper limit of energy storage charging and discharging power. Indicates the basic power value of the load. Represents the active power of photovoltaic, Indicates the active power of the fan, represents the investment capacity of the ith PV node in planning stage t, represents the investment capacity of the i-th wind turbine in planning stage t, represents the investment capacity of the i-th energy storage device in the t-th planning stage, Indicates the active power of the energy storage device charging, Indicates the active power discharged by the energy storage device, Indicates the active power load of the node.

[0145] Further, refer to Figure 4 In step S4, the distribution network probability planning model is solved based on the improved Benders decomposition algorithm to obtain a probability planning scheme, including:

[0146] S41, converting the distribution network probabilistic planning model into a compact form;

[0147] S42. Based on the compact form, decoupling the distribution network probabilistic planning model into an investment main problem and an operation sub-problem;

[0148] S43, introducing the slack variable into the subproblem as a penalty term, obtaining the relaxed running subproblem, and forming a unified dual cut;

[0149] S44, solving the main problem and recording planning variables;

[0150] S45, feeding back the planning variables to the relaxed operation subproblem, and recording the unified dual cut;

[0151] S46, judging whether the upper bound and the lower bound formed by the main problem and the relaxed running subproblem converge, if so, stopping the iteration, otherwise solving the main problem again.

[0152] Wherein, in step S41, the compact form is as follows:

[0153] min x∈{0,1},y {c T x+d T y:Dx≤e,Ax+By≤b}; (25)

[0154] In step S42, the main question is as follows:

[0155]

[0156] The sub-problems are as follows:

[0157]

[0158] Where x and y represent investment 0-1 variables and operation continuous variables, respectively; b, c, d, e, A, B, and D represent coefficient matrices; the superscript T represents the transposed matrix; and the subscript s represents the number of the typical scenario day. is the intermediate variable, μ s and λ s represents the dual variable in the iteration process, μ s Used to generate the optimal dual cut, λ s Used to generate feasible dual cuts. The superscript * indicates the optimal value during the iteration. s Indicates the relaxation of constraints when the subproblem is infeasible, d s represents the coefficient matrix of a typical scenario day s, x s represents the investment 0-1 variable on the typical scenario day s, y s represents the running continuous variable of the typical scenario day s, M is a positive integer, v s represents the slack variable, A s represents the coefficient matrix of a typical scenario day s, B s represents the coefficient matrix of the typical scenario day s, b s Represents the coefficient matrix of a typical scenario day s.

[0159] In step S43, the slack variable v sIntroduced into the subproblem, and the slack variable is added as a penalty term to the objective function, the modified subproblem is shown in (28), where M represents a large positive integer. After the improvement, the optimal dual cut and the relaxed dual cut do not need to be solved separately, forming a unified dual cut as shown in (29), where π s Denotes the dual variable of the uniform dual cut.

[0160]

[0161] Where x and y represent investment 0-1 variables and operation continuous variables, respectively; b, c, d, e, A, B, and D represent coefficient matrices; the superscript T represents the transposed matrix; and the subscript s represents the number of the typical scenario day. is the intermediate variable, μ s and λ s represents the dual variable in the iteration process, μ s Used to generate the optimal dual cut, λ s Used to generate feasible dual cuts. The superscript * indicates the optimal value during the iteration. s Indicates the relaxation of constraints when the subproblem is infeasible, d s represents the coefficient matrix of a typical scenario day s, x s represents the investment 0-1 variable on the typical scenario day s, y s represents the running continuous variable of the typical scenario day s, M is a positive integer, v s represents the slack variable, A s represents the coefficient matrix of a typical scenario day s, B s represents the coefficient matrix of the typical scenario day s, b s Represents the coefficient matrix of the typical scenario day s, π s Represents the dual variable of the unified dual cut under the typical scenario day s.

[0162] Next, the improved Benders decomposition algorithm is used to solve the model. First, the main planning problem is solved and the planning variables are recorded. Then, the optimal values ​​of the planning variables are fed back to the relaxed running subproblem. Next, the dual variables of the running subproblem are recorded to form a unified dual cut, and it is determined whether the upper and lower bounds formed by the main subproblem converge. If so, stop. Otherwise, solve the main problem again and continue the iterative process.

[0163] The distribution network probabilistic planning method and device considering multi-time scale uncertainty provided by the above embodiment at least include the following beneficial effects:

[0164] By considering uncertainties at multiple time scales, long-term planning uncertainties and short-term operating uncertainties can be coordinated and combined; multi-scenario hybrid trees can cover a wider range of future scenarios, thereby improving the comprehensiveness and reliability of planning results; the planning scheme obtained by the multi-time scale probabilistic planning model can better cope with the multiple uncertainties faced by the high-voltage distribution network during the planning period, and improve the economy, flexibility and adaptability of the planning scheme; the improved solution algorithm can improve the solution efficiency, shorten the calculation time, and make it possible to solve large-scale problems.

[0165] refer to Figure 5 In some embodiments, a distribution network probabilistic planning device considering multi-time scale uncertainties is also provided, comprising:

[0166] A scenario building module 201 builds a multi-stage hybrid scenario tree for multi-time scale uncertainty factors based on the uncertainty factors of the distribution network at different times;

[0167] A target establishment module 202 is used to establish an objective function with the goal of minimizing the comprehensive cost under multiple scenarios within the planning period according to the multi-stage hybrid scenario tree;

[0168] The planning module 203 is used to establish the constraints of distribution network planning investment and operation, and obtain the distribution network probabilistic planning model according to the objective function and the constraints;

[0169] The analysis module 204 solves the distribution network probabilistic planning model based on the improved Benders decomposition algorithm to obtain a probabilistic planning solution.

[0170] Furthermore, the uncertainties at different times include uncertainties in the planning stage and uncertainties within a preset time; the uncertainties in the planning stage include load growth and electricity price status; and the uncertainties within the preset time include wind power load and photovoltaic load.

[0171] Furthermore, the scenario establishment module 201 establishes a multi-stage hybrid scenario tree for multi-time scale uncertainty factors based on the uncertainty factors of the distribution network at different times, including:

[0172] Determine the planning stage of the distribution network;

[0173] The current state of the distribution network is taken as the root node, the load growth and electricity price state in each planning stage are taken as nodes, the planning path between the planning stages are taken as line segments, and the root node and each node are connected through the line segments;

[0174] The Markov chain method is used to obtain the probability transfer characteristics of the connected nodes between different stages and the occurrence probability of the planned path;

[0175] At each node, scenarios are clustered based on wind power load and photovoltaic load to form a typical scenario set.

[0176] Furthermore, the typical scene set includes wind power load and photovoltaic load on a typical scene day; the scene establishment module 201 performs scene clustering based on wind power load and photovoltaic load at each node to form a typical scene set, including:

[0177] At each node, the Weibull model parameters are obtained based on the historical wind speed data to generate simulated wind speed, and the photovoltaic Beta model parameters are obtained based on the historical photovoltaic data to generate simulated photovoltaic power;

[0178] The simulated wind speed and the simulated photovoltaic power within a preset time period are taken as a sample set, and a clustering algorithm is used to perform a clustering operation on the sample set to obtain a typical scene set.

[0179] Furthermore, the objective function of the multi-stage probabilistic programming model is as follows:

[0180]

[0181] Wherein, the subscript l represents the planning path between planning stages, and Ω l represents the set of planned paths, Ω t represents the set of planning stages, ρ l represents the occurrence probability of the planned path l, f l represents the planning cost under planning path l, γt represents the initial year of planning stage t, ψ represents the annualized interest rate, They represent the equipment investment cost, equipment maintenance cost, daily operation cost and power failure cost of the high-voltage distribution network in stage t respectively, and f represents the comprehensive cost.

[0182] Furthermore, the constraints on distribution network planning, investment and operation include equipment construction constraints, power flow constraints, grid radial constraints and upper and lower limit constraints of operating variables.

[0183] Furthermore, the analysis module 204 solves the distribution network probability planning model based on the improved Benders decomposition algorithm to obtain a probability planning solution, including:

[0184] converting the distribution network probabilistic programming model into a compact form;

[0185] Based on the compact form, the distribution network probabilistic programming model is decoupled into an investment main problem and an operation sub-problem;

[0186] Introduce slack variables into the subproblem as penalty terms, obtain the relaxed running subproblem, and form a unified dual cut;

[0187] Solve the main problem and record the planning variables;

[0188] Feeding back the planning variables to the relaxed running subproblem and recording the unified dual cut;

[0189] It is determined whether the upper bound and the lower bound formed by the main problem and the relaxed running subproblem converge. If so, the iteration is stopped; otherwise, the main problem is solved again.

[0190] Further, the compact form is as follows:

[0191] min x∈{0,1},y {c T x+d T y:Dx≤e,Ax+By≤b};

[0192] The main problem is as follows:

[0193]

[0194] The relaxed running subproblem is as follows:

[0195]

[0196] The unified dual cut is as follows:

[0197]

[0198] Where x and y represent investment 0-1 variables and operation continuous variables, respectively; b, c, d, e, A, B, and D represent coefficient matrices; the superscript T represents the transposed matrix; and the subscript s represents the number of the typical scenario day. is the intermediate variable, μ s and λ s represents the dual variable in the iteration process, μ s Used to generate the optimal dual cut, λ s Used to generate feasible dual cuts. The superscript * indicates the optimal value during the iteration. s Indicates the relaxation of constraints when the subproblem is infeasible, d s represents the coefficient matrix of a typical scenario day s, x s represents the investment 0-1 variable on the typical scenario day s, y s represents the running continuous variable of the typical scenario day s, M is a positive integer, v s represents the slack variable, A s represents the coefficient matrix of a typical scenario day s, B s represents the coefficient matrix of the typical scenario day s, b s Represents the coefficient matrix of the typical scenario day s, π s Represents the dual variable of the unified dual cut under the typical scenario day s.

[0199] In some embodiments, a computer storage medium is also provided, wherein the computer storage medium stores computer instructions, and when the computer instructions are called, they are used to execute the above method.

[0200] Although preferred embodiments of the present invention have been described, additional changes and modifications may be made to these embodiments by those skilled in the art once the basic inventive concepts are known. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention. Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A probabilistic planning method for distribution networks considering multi-time scale uncertainties, It is characterized in that include: Based on the uncertainties of the distribution network at different times, a multi-stage hybrid scenario tree of multi-time scale uncertainties is established. According to the multi-stage hybrid scenario tree, an objective function is established with the goal of minimizing the comprehensive cost under multiple scenarios within the planning period; Establishing constraints on distribution network planning, investment and operation, and obtaining a distribution network probabilistic planning model based on the objective function and the constraints; The distribution network probabilistic planning model is solved based on the improved Benders decomposition algorithm to obtain a probabilistic planning solution.

2. The method according to claim 1, It is characterized in that The uncertainties at different times include uncertainties in the planning stage and uncertainties within the preset time; the uncertainties in the planning stage include load growth and electricity price status; the uncertainties within the preset time include wind power load and photovoltaic load.

3. The method according to claim 2, It is characterized in that Based on the uncertainties of the distribution network at different times, a multi-stage hybrid scenario tree of multi-time scale uncertainties is established, including: Determine the planning stage of the distribution network; The current state of the distribution network is taken as the root node, the load growth and electricity price state in each planning stage are taken as nodes, the planning path between the planning stages are taken as line segments, and the root node and each node are connected through the line segments; The Markov chain method is used to obtain the probability transfer characteristics of the connected nodes between different stages and the occurrence probability of the planned path; At each node, scenarios are clustered based on wind power load and photovoltaic load to form a typical scenario set.

4. The method according to claim 3, It is characterized in that The typical scenario set includes wind power load and photovoltaic load under typical scenario days; At each node, scenarios are clustered based on wind power load and photovoltaic load to form a typical scenario set, including: At each node, the Weibull model parameters are obtained based on the historical wind speed data to generate simulated wind speed, and the photovoltaic Beta model parameters are obtained based on the historical photovoltaic data to generate simulated photovoltaic power; The simulated wind speed and the simulated photovoltaic power within a preset time period are taken as a sample set, and a clustering algorithm is used to perform a clustering operation on the sample set to obtain a typical scene set.

5. The method according to claim 1, It is characterized in that The objective function is as follows: Wherein, the subscript l represents the planning path between planning stages, and Ω l represents the set of planned paths, Ω t represents the set of planning stages, ρ l represents the occurrence probability of the planned path l, f l represents the planning cost under planning path l, γt represents the initial year of planning stage t, ψ represents the annualized interest rate, They represent the equipment investment cost, equipment maintenance cost, daily operation cost and power failure cost of the high-voltage distribution network in the planning stage t respectively, and f represents the comprehensive cost.

6. The method according to claim 1, It is characterized in that The constraints on distribution network planning, investment and operation include equipment construction constraints, power flow constraints, grid radial constraints and upper and lower limit constraints of operating variables.

7. The method according to claim 1, It is characterized in that The distribution network probabilistic planning model is solved based on the improved Benders decomposition algorithm to obtain a probabilistic planning solution, including: converting the distribution network probabilistic programming model into a compact form; Based on the compact form, the distribution network probabilistic programming model is decoupled into an investment main problem and an operation sub-problem; Introduce slack variables into the subproblem as penalty terms, obtain the relaxed running subproblem, and form a unified dual cut; Solve the main problem and record the planning variables; Feeding back the planning variables to the relaxed running subproblem and recording the unified dual cut; It is determined whether the upper bound and the lower bound formed by the main problem and the relaxed running subproblem converge. If so, the iteration is stopped; otherwise, the main problem is solved again.

8. The method according to claim 7, It is characterized in that The compact form is as follows: min x∈{0,1},y {c T x+d T y:Dx≤e,Ax+By≤b}; The main problem is as follows: The relaxed running subproblem is as follows: The unified dual cut is as follows: Where x and y represent investment 0-1 variables and operation continuous variables, respectively; b, c, d, e, A, B, and D represent coefficient matrices; the superscript T represents the transposed matrix; the subscript s represents the number of the typical scenario day; and θ s is the intermediate variable, μ s and λ s represents the dual variable in the iteration process, μ s Used to generate the optimal dual cut, λ s Used to generate feasible dual cuts. The superscript * indicates the optimal value during the iteration. s Indicates the relaxation of constraints when the subproblem is infeasible, d s represents the coefficient matrix of a typical scenario day s, x s represents the investment 0-1 variable on the typical scenario day s, y s represents the running continuous variable of the typical scenario day s, M is a positive integer, v s represents the slack variable, A s represents the coefficient matrix of a typical scenario day s, B s represents the coefficient matrix of the typical scenario day s, b s Represents the coefficient matrix of the typical scenario day s, π s Represents the dual variable of the unified dual cut under the typical scenario day s.

9. A distribution network probabilistic planning device considering multi-time scale uncertainties, It is characterized in that include: The scenario building module builds a multi-stage hybrid scenario tree for multi-time scale uncertainty factors based on the uncertainty factors of the distribution network at different times; A target establishment module, used to establish an objective function with the goal of minimizing the comprehensive cost under multiple scenarios within the planning period according to the multi-stage hybrid scenario tree; A planning module, used to establish the constraints of distribution network planning investment and operation, and obtain a distribution network probabilistic planning model according to the objective function and the constraints; The analytical module solves the distribution network probability planning model based on the improved Benders decomposition algorithm to obtain a probability planning solution.

10. A computer storage medium, It is characterized in that The computer storage medium stores computer instructions, and when the computer instructions are called, they are used to execute the method according to any one of claims 1 to 8.

Images