Power distribution network multi-target robust extension planning method considering source load uncertainty
By constructing an uncertain set and a multi-objective robust extended programming model, and combining the ε-constraint algorithm and the C&CG algorithm, the problem of planning conservatism caused by the uncertainty of distributed power sources in the distribution network is solved, a balance between economic efficiency and carbon emissions is achieved, and a planning scheme for a low-carbon economy is provided.
Patent Information
- Application Number
- CN202410901674.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-05
- Publication Date
- 2024-11-22
AI Technical Summary
Existing power distribution network planning methods fail to effectively handle the randomness and uncertainty of distributed power sources, resulting in highly conservative planning schemes that cannot balance economic efficiency and carbon emissions.
A multi-objective robust extended programming method for distribution networks that takes into account source-load uncertainties is adopted. By constructing an uncertainty set, a multi-objective robust extended programming model is established, and the Pareto front solution set is generated by solving the model using the ε-constraint algorithm and the C&CG algorithm. Finally, the optimal compromise solution is determined by the TOPSIS method.
It effectively reduces the conservatism of power distribution network planning schemes, achieves a balance between economy and carbon emissions, and provides a low-carbon and economical planning scheme.
Smart Images

Figure CN119005560A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of distribution network expansion planning, and in particular relates to a distribution network multi-objective robust expansion planning method taking into account source-load uncertainty. Background Art
[0002] In response to the growing energy demand and increasingly severe environmental problems, replacing fossil energy with clean energy and reducing carbon emissions has become a development trend in the global energy system. Increasing the proportion of distributed power sources in the distribution system to reduce the use of non-renewable energy such as coal at the power generation end is an important measure. Increasing the investment in distributed power sources to achieve the goal of emission reduction will increase the comprehensive cost of the distribution network.
[0003] Existing research that considers carbon emissions in the distribution network planning process mostly considers emissions issues as economic factors, such as the carbon emission cost of the distribution network, which cannot reflect the specific effect of carbon emission reduction. On the other hand, the randomness and uncertainty of renewable distributed power generation also greatly affect the safe and stable operation of the distribution network. Traditional distribution network planning is mostly deterministic planning, and after the distributed power source is connected to the distribution network, its output is random, and the complexity of the power system is greatly increased. Therefore, the traditional distribution network planning method cannot adapt well to the grid connection of distributed power sources. Therefore, it is necessary to consider the carbon emissions and uncertainty issues of the distribution network in the planning process and conduct in-depth research. Summary of the invention
[0004] The present invention provides a multi-objective robust expansion planning method for a distribution network taking into account the uncertainty of source and load. The planning method can well describe the uncertainty of wind and solar loads, effectively reduce the conservatism of the planning scheme, and provide a planning scheme that balances economy and carbon emissions for the distribution network.
[0005] The technical solution adopted by the present invention is:
[0006] A multi-objective robust expansion planning method for a distribution network taking into account source and load uncertainty comprises the following steps:
[0007] Step 1: Obtain data information of the distribution network to be expanded;
[0008] Step 2: Consider the uncertainty of load power and distributed generation output, and construct an uncertainty set based on the budget;
[0009] Step 3: Taking the annual comprehensive cost and carbon emission minimization as the objective function, a multi-objective robust expansion planning model for the distribution network is established;
[0010] Step 4: Use the ε-constraint algorithm to transform the multi-objective robust problem into multiple single-objective equivalent problems;
[0011] Step 5: Solve the equivalent single-objective robust problem based on the C&CG algorithm, and multiple single-objective equivalent solutions constitute a robust Pareto solution set;
[0012] Step 6: Based on TOPSIS, the optimal compromise solution is determined from the solution set to obtain the distribution network expansion planning scheme.
[0013] In the step 1, obtaining data information of the distribution network to be expanded includes:
[0014] Obtain the original network topology of the distribution network to be expanded, the topology of the new lines to be built, load data, and the locations of the equipment to be planned;
[0015] Obtain the investment and operation and maintenance costs and equipment operating parameters of the equipment to be built;
[0016] Obtain the prices of electricity purchased and abandoned for the distribution network to be expanded.
[0017] In the step 2, the budget-based uncertainty set constructed includes:
[0018]
[0019] In formula (1), U is the uncertainty set; u is the uncertainty variable, including load active power, wind power, and photovoltaic active power output; are the uncertain variables of load active power, wind power active output, and photovoltaic active output respectively; α represents the fluctuation range of the uncertainty variable; are the predicted values of load active power, wind power active output, and photovoltaic active output of node j respectively; Ω B ,Ω WT ,Ω PV are the load node set, wind power node set, and photovoltaic node set respectively; Γ L , Γ WT , Γ PV Represents the uncertain adjustment parameters introduced for load active power, wind power active output, and photovoltaic active output.
[0020] In the step 3, a multi-objective robust expansion planning model for the distribution network is established with the minimum annual comprehensive cost and carbon emissions as the objective function, specifically including:
[0021]
[0022] In formula (2), minf1 represents the minimization of annual comprehensive cost; minf2 represents the minimization of carbon emissions; annual investment cost C INV Includes: New line cost C LINE 、Annual investment cost of wind power C WT , photovoltaic annual investment cost C PV ; Annual operating cost C OPEIncluding: Main grid electricity purchase cost C buy , Network loss cost C loss 、Operation cost of distributed power generation C dg 2. Penalty costs for wind and solar power abandonment Carbon emissions from purchasing electricity from the main grid for the distribution network.
[0023] The multi-objective expansion planning model of the distribution network satisfies the following constraints:
[0024] 1) Equipment investment constraints:
[0025]
[0026] In the formula, N are 0-1 variables indicating whether the j-th wind power, photovoltaic power and new line are put into construction or not; wt 、N pv 、N line is the maximum number of operational constructions; line , pv , wt They are respectively the lines to be selected, photovoltaic power and wind power collection.
[0027] 2) Network radial and connectivity constraints:
[0028] n=m+1 (8);
[0029] Where n and m are the total number of nodes and lines in the planned distribution network, respectively.
[0030] 3) Power flow constraints:
[0031]
[0032] In formula (9), δ j represents the set of lines with j as the starting point, and k is the end point of the corresponding line; φ j represents the set of lines with j as the end, and i is the beginning of the corresponding line; P ij,t is the active transmission power of branch ij; I ij,t is the current of branch ij; r ij is the resistance of branch ij; Ω wt is the numbered node of wind power; is the active power output of the wind turbine at node j; Ω pv is the numbered node of photovoltaic; is the active power output of the photovoltaic power station at node j; Make contributions to the checkpoint;Ω load is the numbered node of the load; is the active load at node j; Q ij,t is the reactive transmission power of branch ij; xij is the reactance of branch ij; is the reactive power output of wind turbine at node j; is the reactive power output of the photovoltaic power station at node j; Make a contribution to the checkpoint; is the reactive load at node j; V i,t 、V j,t Respectively represent the voltage of nodes i and j; Ω l is a collection of lines; ij is the number in the collection of lines, representing the endpoint.
[0033] 4) Safety constraints:
[0034]
[0035] In formula (10), V max 、V min are the upper and lower limits of the node voltage respectively; is the upper limit of the current on line ij.
[0036] 5) Distributed power operation constraints:
[0037]
[0038] In formula (11), is the distributed power capacity of node i; Maximum configuration capacity for distributed power generation; Provide active power for distributed power sources; is the output coefficient of distributed power source; is the minimum power factor of the distributed power source; It is the reactive power output of distributed power source.
[0039] The voltage drop and power transmission constraints in the power flow constraints are relaxed by the large M method, and the following formula is obtained:
[0040]
[0041] In formula (12), are the square equivalent values of node i voltage and line ij current respectively; Instead of M is a sufficiently large positive number; P max , Q max I are the upper limits of active and reactive power of line transmission respectively; max is the upper limit of line current;
[0042] Then, for the formula Performing second-order cone relaxation, we obtain Transpose both sides of the equation:
[0043]
[0044] Then it is further equivalent to:
[0045]
[0046] Its second-order cone constraint form is:
[0047]
[0048] In the step 4, the ε-constraint algorithm is used to transform the multi-objective robust problem into multiple single-objective equivalent problems, which specifically includes the following steps:
[0049] The ε-constrained algorithm creates a single-objective equivalent optimization problem by optimizing one objective while imposing a set of upper bound constraints on other objectives. For the proposed dual-objective problem, the single-objective equivalent optimization problem can be modeled as follows.
[0050]
[0051] Where: x is the first stage optimization variable, x∈Z + or {0, 1}; u is an uncertain variable; y is a continuous variable in the second stage; by setting the upper bound coefficient ε to different values uniformly distributed from 0 to 1, multiple single-objective equivalent robust optimization problems can be obtained.
[0052] The step five specifically includes the following steps:
[0053] The C&CG algorithm for generating constraints splits the two-stage robust optimization problem of min-max-min structure into a main problem MP and a sub-problem SP for solution; the main problem MP after splitting is a mixed integer second-order cone programming problem; the main problem MP is solved to obtain the first-stage optimization variable x ★ , i.e., the distribution network structure, equipment capacity and selection, the solution result is fixed and passed to the sub-problem SP; the sub-problem SP solves the worst scenario u in operation according to the planning scheme obtained by the main problem MP l , and passed to the main problem MP.
[0054] 1) Solve the main problem MP:
[0055]
[0056] In formula (17), c is the coefficient of the decision variable x in the objective function; y l 、u lare the second-stage variables and uncertainty variables transferred from the subproblem, which are reflected in the main problem in the form of constraints; the constant coefficient matrix A is the coefficient matrix related to the decision variable x; the column vector d represents the coefficient vector related to the decision variable x; G, E, Q all represent the coefficient matrix related to the decision variable; h and a are constant vectors, representing the coefficient vector related to the decision variable y; the constant matrix M represents the coefficient matrix related to the uncertainty variable; η is the auxiliary variable introduced to solve the l-th subproblem, which is the C&CG cut; u l The worst scenario obtained for the subproblem.
[0057] At this time, the main problem MP is a mixed integer second-order cone programming problem, which can be solved directly using a linear solver.
[0058] 2) Solution of subproblem SP:
[0059] First, for the max-min subproblem, the strong duality theory is used to transform the inner minimization problem min into its dual form and merge it into a maximization problem, as follows:
[0060]
[0061] Where π, α, and θ are the dual variables of the subproblem constraints.
[0062] The transformed subproblem has a nonlinear term u T π, since the uncertainty variable u takes its endpoint value in the worst case, the fluctuation range of the uncertainty variable can be converted into Where u0 is the minimum value of the uncertainty variable, The 0-1 control variable is used to determine the value of the uncertain variable, and △u is the fluctuation range of the uncertain variable. At this time, the nonlinear term is in the form of the product of binary variables and continuous variables, and auxiliary variables and related constraints are introduced for linearization; the details are as follows:
[0063]
[0064] In the formula, μ is the product of the 0-1 variable and the continuous variable in the original nonlinear term; M is a large positive integer; z is the auxiliary variable introduced to control the value of the nonlinear term.
[0065] The relaxed subproblem is a linear problem and can be solved directly.
[0066] Then the C&CG algorithm iteratively solves the main problem MP and the subproblem SP until the gap between the next bound given by the main problem MP and the upper bound given by the subproblem SP is narrowed to a termination threshold k, and outputs the corresponding solution.
[0067] Multiple different robust problems can have different solutions, thus producing a Pareto front.
[0068] In step 6, the solutions in the Pareto front form a decision matrix according to the target value, and then make a decision according to the following steps:
[0069] 1) Obtain the target n positive ideal solution from the decision matrix With negative ideal solution
[0070]
[0071] In the formula, are all the elements in the nth column of the decision matrix.
[0072] 2) Calculate the closeness of each decision plan to the positive ideal solution Closeness to negative ideal solution
[0073]
[0074] 3) Calculate the evaluation index f m :
[0075]
[0076] The evaluation indicators are sorted and the maximum value is the optimal solution.
[0077] The present invention provides a multi-objective robust expansion planning method for a distribution network taking into account source-load uncertainty, and the technical effects are as follows:
[0078] 1) The present invention takes into account the impact of carbon emission targets during the distribution network expansion planning process, increases investment in new energy, reduces the purchase of electricity from the superior power grid, thereby reducing dependence on fossil energy such as thermal power and reducing the carbon emissions of the system.
[0079] 2) In view of the fact that traditional robustness is too conservative, the present invention adopts uncertainty sets to describe the uncertainty of wind power, photovoltaic power and load, and introduces uncertainty adjustment parameters to reduce the conservatism of the planning scheme.
[0080] 3) The Epsilon-constraint method of the present invention can transform the multi-objective robust optimization model into a single-objective equivalent robust optimization model without affecting the optimality and robustness of the solution, and the obtained Pareto frontier is more accurate.
[0081] 4) The multi-objective robust expansion planning method for distribution networks proposed in the present invention taking into account the uncertainty of sources and loads can effectively handle the uncertainty of the source and both sides, so that the output planning scheme has better economy, low carbon and conservatism. BRIEF DESCRIPTION OF THE DRAWINGS
[0082] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:
[0083] Figure 1 Schematic diagram of the multi-objective robust expansion planning method for distribution networks taking into account source and load uncertainties.
[0084] Figure 2 To improve the IEEE33 node system diagram.
[0085] Figure 3 This is the final planning solution result diagram.
[0086] Figure 4 It is the Pareto front curve. DETAILED DESCRIPTION
[0087] A multi-objective robust expansion planning method for a distribution network taking into account source and load uncertainty includes the following implementation steps:
[0088] Step 1: Obtain data information of the distribution network to be expanded; specifically includes the following steps:
[0089] Obtain the original network topology of the distribution network to be expanded, the line topology to be built, load data and the location of the equipment to be planned;
[0090] Obtain the investment and operation and maintenance costs and equipment operating parameters of the equipment to be built;
[0091] Obtain the power purchase and abandonment prices of the distribution network to be expanded;
[0092] Obtain the distributed generation output coefficient and its fluctuation range of the distribution network to be expanded.
[0093] Step 2: Consider the uncertainty of load power and distributed generation output, and construct an uncertainty set based on the budget:
[0094] The specific steps include the following: Load, wind speed and solar radiation all have uncertain characteristics. Therefore, a typical day simulation operation cycle of 24 hours is adopted, and the value of each hour is calculated through the historical data of the whole year to obtain the average data of 24 hours. By setting a fluctuation range to describe the uncertainty of load power and distributed power output, and adding an uncertain parameter to control the conservatism of the solution, the constructed uncertainty set is as follows:
[0095]
[0096] In the formula, are the average values of the nodes corresponding to each uncertainty variable at time t; α is the fluctuation range of the uncertainty variable; Γ L , Γ WT , Γ PVUncertain adjustment parameters are introduced for load active power, wind power and photovoltaic active output.
[0097] Step 3: Taking the minimum annual comprehensive cost and carbon emissions as the objective function, a multi-objective robust expansion planning model for the distribution network is established, which specifically includes the following steps:
[0098] The annual comprehensive cost f1 includes the annual investment cost C INV and annual operating cost C OPE ,
[0099] minf1=C INV +C OPE
[0100] In the formula, C INV is the distribution network planning cost and C INV =C LINE +C PV +C WT ,C LINE is the line investment cost equivalent to one year and τ is the planned annual value coefficient of the equipment r is the discount rate, T is the fixed investment payback period of the equipment, c line is the investment cost of the line, Ψ L is the set of candidate routes, x line is the investment decision variable; C PV is equivalent to one year's photovoltaic investment cost and c pv is the photovoltaic investment cost per unit capacity, x pv is the photovoltaic investment decision variable, is the installed capacity of distributed photovoltaics, pv is the set of distributed photovoltaic nodes to be connected; C WT is equivalent to one year's wind power investment cost and c wt is the photovoltaic investment cost per unit capacity, x wt is the photovoltaic investment decision variable, is the installed capacity of distributed photovoltaics, wt is the set of wind power nodes to be connected; C OPE is the annual operating cost of the distribution network and C OPE =C buy +C loss +C dg , C buy The annual electricity purchase fee from the superior is the unit electricity purchase cost of the main grid at time t, P t sp To purchase electricity from the superior, C loss is the annual network loss cost and Closs =365·∑ t∈T △t·c loss P loss , c loss is the unit network loss cost, P loss is the network loss per unit time, C dg is the operation and maintenance cost of distributed power generation and c pv 、c wt are the unit photovoltaic and wind power output operation and maintenance costs, It is the active power output of distributed photovoltaic and wind power at time t.
[0101] The carbon emissions of the distribution network are the indirect carbon emissions generated by purchasing electricity from the main grid:
[0102]
[0103] In the formula, is the carbon emissions of the distribution network, and η is the carbon emission factor of the distribution network purchasing electricity from the main grid.
[0104] The multi-objective expansion planning model of the distribution network satisfies the following constraints:
[0105] 1) Equipment investment constraints:
[0106]
[0107] x wt 、x pv 、x line N are 0-1 variables indicating whether wind power, photovoltaic power and new lines are put into construction or not; wt 、N pv 、N line The maximum number of running builds.
[0108] 2) Network radial and connectivity constraints:
[0109] n=m+1 (8);
[0110] Where n and m are the total number of nodes and lines in the planned distribution network, respectively.
[0111] 3) Power flow constraints:
[0112]
[0113] Where P ij,t is the active transmission power of branch ij; I ij,t is the current of path ij; r ij is the resistance of branch ij; Q ij,t is the reactive transmission power of branch ij; xij is the reactance of branch ij; is the active power output of the wind turbine at node j; is the active power output of the photovoltaic power station at node j; Make contributions to the checkpoint; is the active load at node j; is the reactive power output of wind turbine at node j; is the reactive power output of the photovoltaic power station at node j; Make a contribution to the checkpoint; is the reactive load at node j; Ω wt ,Ω pv ,Ω load They are the numbered nodes of wind power, photovoltaic power and load respectively.
[0114] 4) Safety constraints:
[0115]
[0116] Where V max 、V min are the upper and lower bounds of the node voltage respectively; is the upper bound of the current on line ij.
[0117] 5) Distributed power operation constraints
[0118]
[0119] In the formula, is the distributed generation capacity of node i; Maximum configuration capacity for distributed power generation; Provide active power for distributed power sources; is the output coefficient of distributed power source; is the minimum power factor of the distributed power source; It is the reactive power output of distributed power source.
[0120] The voltage drop and power transmission constraints in the power flow constraints are relaxed by the large M method, and the following formula is obtained:
[0121]
[0122] In the formula, the variable Instead of is the square equivalent value of voltage and current; M is a sufficiently large positive number; P max , Q max I are the upper limits of active and reactive power of line transmission respectively; max is the upper limit of line current;
[0123] Then, for the formula Performing second-order cone relaxation, we obtain Transpose both sides of the equation:
[0124]
[0125] Then it is further equivalent to:
[0126]
[0127] Its second-order cone constraint form is:
[0128]
[0129] Step 4: Use the ε-constraint algorithm to transform the multi-objective robust problem into multiple single-objective equivalent problems:
[0130] The Epsilon-constrained method creates a single-objective equivalent optimization problem by optimizing one objective while imposing a set of upper bound constraints on other objectives. The specific steps are as follows:
[0131] The Epsilon-constrained algorithm creates a single-objective equivalent optimization problem by optimizing one objective while imposing a set of upper bound constraints on other objectives. For the proposed dual-objective problem, the single-objective equivalent optimization problem can be modeled as follows.
[0132]
[0133] By setting the upper bound coefficient ε to different values uniformly distributed from 0 to 1, multiple single-objective equivalent robust optimization problems can be obtained.
[0134] Step 5: Solve the equivalent single-objective robust problem based on the C&CG algorithm. Multiple single-objective equivalent solutions constitute a robust Pareto solution set. The C&CG algorithm decomposes the two-stage planning problem of the min-max-min structure into main and sub-problems; the main problem MP uses the cutting plane obtained by the sub-problem SP and the identified bad scenarios to solve the planning scheme, and the sub-problem SP determines the worst scenario based on the optimal planning scheme obtained by the main problem MP. The main and sub-problems iterate each other until convergence to obtain the planning scheme under the constraint.
[0135] The specific steps include:
[0136] The column and constraint generation algorithm splits the two-stage robust optimization problem of min-max-min structure into a master problem (MP) and a subproblem SP for solution; the master problem MP after splitting is a mixed integer second-order cone programming problem; the master problem is solved to obtain the first-stage optimization variable x ★ , that is, the distribution network structure, equipment capacity and selection, fix the solution and pass it to the sub-problem; the sub-problem solves the worst scenario u in operation according to the planning scheme obtained by the main probleml , and passed to the main problem.
[0137] 1) Solve the main problem:
[0138]
[0139] In the formula, x is the first stage optimization variable, x∈Z + or {0, 1}; u is an uncertain variable; y is a second-stage continuous variable; the constant coefficient matrix A is the coefficient matrix associated with the decision variable x; the column vector d represents the coefficient vector associated with the decision variable x; G, E, and Q all represent the coefficient matrix associated with the decision variable; h and a are constant vectors, representing the coefficient vector associated with the decision variable y; the constant matrix M represents the coefficient matrix associated with the uncertain variable; η is the auxiliary variable introduced to solve the l-th subproblem, which is the C&CG cut; u l The worst scenario obtained for the subproblem.
[0140] At this time, the main problem MP is a mixed integer second-order cone programming problem that can be directly solved using a linear solver.
[0141] 2) Solving subproblems: First, for the max-min subproblem, the strong duality theory is used to transform the inner minimization problem min into its dual form and merge it into a maximization problem, as follows:
[0142]
[0143] Where π, α, and θ are the dual variables of the subproblem constraints.
[0144] The transformed subproblem has nonlinear terms, which are relaxed using the big M method. Then the C&CG algorithm iteratively solves the main problem and the subproblem until the gap between the lower bound given by the main problem and the upper bound given by the subproblem is reduced to a termination threshold k, and the corresponding solution is output.
[0145] Multiple different robust problems can have different solutions, thus generating a Pareto front.
[0146] Step 6: Based on TOPSIS, the optimal compromise solution is determined from the solution set to obtain the distribution network expansion planning scheme. The decision-making method using the TOPSIS method is to determine the best sorting order by comparing the distance between the alternative solutions and the ideal solution. The ideal solution includes positive ideal solution and negative ideal solution. The positive ideal solution is the best solution in each attribute, while the negative ideal solution is the worst solution in each attribute.
[0147] The specific steps include:
[0148] The solutions in the above Pareto front form a decision matrix according to the target value, and then make a decision according to the following steps.
[0149] 1) Obtain the target n positive ideal solution from the decision matrix With negative ideal solution
[0150]
[0151] In the formula, c mn are all the elements in the nth column of the decision matrix.
[0152] 2) Calculate the closeness of each decision plan to the positive ideal solution Closeness to negative ideal solution
[0153]
[0154] 3) Calculate the evaluation index f m :
[0155]
[0156] The evaluation indicators are sorted and the maximum value is the optimal solution.
[0157] Verification example:
[0158] To verify the effectiveness of the method proposed in the present invention, the modified IEEE33 node system is used as an example and verified using the matlab platform. The system reference voltage is 12.66 kV, and the system topology is as follows: Figure 2 As shown: the numbers represent the node numbers, the solid lines represent the constructed lines, and the dotted lines represent the lines to be newly constructed, with a total of 14 lines. The system model includes 46 lines and 43 nodes, of which the newly added load nodes are 34-43.
[0159] The line parameters are: unit cost of 80,000 yuan / km, unit impedance of 0.6+j0.59Ω / km. The discount rate is 10%, and the planning period is 10 years. The investment cost of PVG per unit capacity is 6,000 yuan / kW, the investment cost of WTG per unit capacity is 5,800 yuan / kW, the unit power generation cost of PVG and WTG is 0.2 yuan / kWh, and the capacity of a single PVG and WTG is 100kW. The main grid power purchase cost is 0.8 yuan / kWh, the line network loss cost is 0.6 yuan / kWh, and the penalty cost for wind and solar abandonment is 0.6 yuan / kWh. The total installed capacity of wind power and photovoltaic power is capped at 2,000kW respectively, the PVG nodes to be installed are 20, 24, and 29, and the WTG nodes to be installed are 8, 17, and 31. Uncertain parameter settings: the fluctuation range α is 0.2, and the wind power uncertainty budget Γ wt =2, the photovoltaic uncertainty budget Γ pv =2, load uncertainty budget Γ L is 21.
[0160] The modified 33-node distribution network system expansion result is obtained as follows: Figure 3 As shown in Figure 1, the original line was expanded and 10 lines were added to supply power to 10 new load nodes. The three solutions of minimizing the annual comprehensive cost, the optimal compromise solution and minimizing the annual carbon emissions were compared in the solution process. The Pareto frontier curve is shown in Figure 1. Figure 4 As shown, annual comprehensive cost and carbon emissions are a set of conflicting goals. Lower carbon emissions will bring about greater carbon emissions, so a balance needs to be struck between the two goals.
[0161] The rest of the results are shown in Table 1.
[0162] Table 1 Planning results
[0163]
[0164] Comparing the three schemes, the planning scheme under the minimum cost can achieve lower planning costs, but it will bring higher carbon emissions, and minimizing carbon emissions will increase the investment in distributed power sources in the distribution network. Although the power generation cost of distributed power sources is more economical than the main grid power purchase, the overall comprehensive cost is too high. The optimal compromise scheme obtained by this method can reduce carbon emissions by 15.7% compared with the scheme of minimizing investment costs, and can reduce the annual comprehensive cost by 5.5% compared with minimizing carbon emissions. Therefore, fully considering the two goals can bring better benefits.
[0165] In summary, the multi-objective robust expansion planning method for distribution networks taking into account source-load uncertainty designed by the present invention takes into account the uncertainty of load and distributed power output, line expansion, and distributed power planning. The results of the example show that considering multiple objectives in dealing with uncertainty issues can not only achieve better economic efficiency, but also obtain a planning scheme with lower carbon emissions. The multi-objective robust expansion planning model for distribution networks taking into account source-load uncertainty is more load-based and actual working conditions, and considering carbon emissions has certain significance for environmental protection.
Claims
1. A multi-objective robust expansion planning method for distribution networks taking into account source and load uncertainty, characterized in that The following steps are involved: Step 1: Obtain data information of the distribution network to be expanded; Step 2: Consider the uncertainty of load power and distributed generation output, and construct an uncertainty set based on the budget; Step 3: Taking the minimization of annual comprehensive cost and carbon emissions as the objective function, a multi-objective robust expansion planning model for the distribution network is established; Step 4: Use the ε-constraint algorithm to transform the multi-objective robust problem into multiple single-objective equivalent problems; Step 5: Solve the equivalent single-objective robust problem based on the C&CG algorithm, and multiple single-objective equivalent solutions constitute a robust Pareto solution set; Step 6: Based on TOPSIS, the optimal compromise solution is determined from the solution set to obtain the distribution network expansion planning scheme.
2. A multi-objective robust expansion planning method for a distribution network taking into account source-load uncertainty according to claim 1, characterized in that: In the step 1, obtaining data information of the distribution network to be expanded includes: Obtain the original network topology of the distribution network to be expanded, the topology of the new lines to be built, load data, and the location of the equipment to be planned; obtain the investment and operation and maintenance costs of the equipment to be built and the equipment operating parameters; Obtain the prices of electricity purchased and abandoned for the distribution network to be expanded.
3. A multi-objective robust expansion planning method for a distribution network taking into account source-load uncertainty according to claim 1, characterized in that: In the step 2, the budget-based uncertainty set constructed includes: In formula (1), U is the uncertainty set; u is the uncertainty variable, including load active power, wind power, and photovoltaic active power output; are the uncertain variables of load active power, wind power active output, and photovoltaic active output respectively; α represents the fluctuation range of the uncertainty variable; are the predicted values of load active power, wind power active output, and photovoltaic active output of node j respectively; Ω B ,Ω WT ,Ω PV are load node set, wind power node set, and photovoltaic node set respectively; Γ L , Γ WT , Γ PV Represents the uncertain adjustment parameters introduced for load active power, wind power active output, and photovoltaic active output.
4. A multi-objective robust expansion planning method for a distribution network taking into account source-load uncertainty according to claim 1, characterized in that: In the step 3, a multi-objective robust expansion planning model for the distribution network is established with the minimum annual comprehensive cost and carbon emissions as the objective function, specifically including: In formula (2), minf1 represents the minimization of annual comprehensive cost; minf2 represents the minimization of carbon emissions; annual investment cost C INV Includes: New line cost C LINE 、Annual investment cost of wind power C WT , photovoltaic annual investment cost C PV ; Annual operating cost C OPE Including: Main grid electricity purchase cost C buy , Network loss cost C loss 、Operation cost of distributed power generation C dg 2. Penalty costs for wind and solar power abandonment Carbon emissions from purchasing electricity from the main grid for the distribution network.
5. A multi-objective robust expansion planning method for a distribution network taking into account source-load uncertainty according to claim 4, characterized in that: The multi-objective expansion planning model of the distribution network satisfies the following constraints: 1) Equipment investment constraints: In the formula, N are 0-1 variables indicating whether the j-th wind power, photovoltaic power and new line are put into construction or not; wt 、N pv 、N line is the maximum number of operational constructions; line , pv ,Ψ wt They are the lines to be selected, photovoltaic power and wind power collection respectively; 2) Network radial and connectivity constraints: n=m+1 (8); Where n and m are the total number of nodes and lines in the planned distribution network, respectively; 3) Power flow constraints: In formula (9), δ j represents the set of lines with j as the starting point, and k is the end point of the corresponding line; φ j represents the set of lines with j as the end, and i is the beginning of the corresponding line; P ij,t is the active transmission power of branch ij; I ij,t is the current of branch ij; r ij is the resistance of branch ij; Ω wt is the numbered node of wind power; is the active power output of the wind turbine at node j; Ω pv is the numbered node of photovoltaic; is the active power output of the photovoltaic power station at node j; Make contributions to the checkpoint;Ω load is the numbered node of the load; is the active load at node j; Q ij,t is the reactive transmission power of branch ij; x ij is the reactance of branch ij; is the reactive power output of wind turbine at node j; is the reactive power output of the photovoltaic power station at node j; Make a contribution to the checkpoint; is the reactive load at node j; V i,t 、V j,t Respectively represent the voltage of nodes i and j; Ω l is a collection of lines; ij is the number in the collection of lines, indicating the endpoint; 4) Safety constraints: In formula (10), V max 、V min are the upper and lower limits of the node voltage respectively; is the upper limit of the current on line ij; 5) Distributed power operation constraints: In formula (11), is the distributed generation capacity of node i; Maximum configuration capacity for distributed power generation; Provide active power for distributed power sources; is the output coefficient of distributed power source; is the minimum power factor of the distributed power source; Reactive power output of distributed power sources; The voltage drop and power transmission constraints in the power flow constraints are relaxed by the large M method, and the following formula is obtained: In formula (12), are the square equivalent values of node i voltage and line ij current respectively; Instead of M is a sufficiently large positive number; P max , Q max They are respectively the upper limits of active and reactive power transmitted by the line; I max is the upper limit of line current; Then, for the formula Performing second-order cone relaxation, we obtain Transpose both sides of the equation: Then it is further equivalent to: Its second-order cone constraint form is:
6. A multi-objective robust expansion planning method for a distribution network taking into account source-load uncertainty according to claim 1, characterized in that: In the step 4, the ε-constraint algorithm is used to transform the multi-objective robust problem into multiple single-objective equivalent problems, which specifically includes the following steps: The ε-constrained algorithm creates a single-objective equivalent optimization problem by optimizing one objective while imposing a set of upper bound constraints on other objectives. For the proposed dual-objective problem, the single-objective equivalent optimization problem can be modeled as follows: Where x is the optimization variable in the first stage; u is the uncertain variable; y is the continuous variable in the second stage; by setting the upper bound coefficient ε to different values uniformly distributed from 0 to 1, multiple single-objective equivalent robust optimization problems can be obtained.
7. A multi-objective robust expansion planning method for a distribution network taking into account source-load uncertainty according to claim 1, characterized in that: The step five specifically includes the following steps: The C&CG algorithm splits the two-stage robust optimization problem of min-max-min structure into the main problem MP and the sub-problem SP for solution; the main problem MP after the split is a mixed integer second-order cone programming problem; the main problem MP is solved to obtain the first-stage optimization variable x * , i.e., the distribution network structure, equipment capacity and selection, the solution is fixed and transferred to the sub-problem SP; The sub-problem SP solves the worst scenario u in operation according to the planning solution obtained from the main problem MP l , and pass it to the main problem MP; 1): Solve the main problem MP: In formula (17), c is the coefficient of the decision variable x in the objective function; y l 、u l are the second-stage variables and uncertainty variables transferred from the subproblem, which are reflected in the main problem in the form of constraints; the constant coefficient matrix A is the coefficient matrix related to the decision variable x; the column vector d represents the coefficient vector related to the decision variable x; G, E, Q all represent the coefficient matrix related to the decision variable; h and a are constant vectors, representing the coefficient vector related to the decision variable y; the constant matrix M represents the coefficient matrix related to the uncertainty variable; η is the auxiliary variable introduced to solve the l-th subproblem, which is the C&CG cut; u l The worst-case scenario obtained for the subproblem; At this time, the main problem MP is a mixed integer second-order cone programming problem, which can be solved directly using a linear solver; 2) Solution of subproblem SP: First, for the max-min subproblem, the strong duality theory is used to transform the inner minimization problem min into its dual form and merge it into a maximization problem, as follows: In the formula, π, α, θ are the dual variables of the subproblem constraints; The transformed subproblem has a nonlinear term u T π, since the uncertainty variable u takes its endpoint value in the worst case, the fluctuation range of the uncertainty variable can be converted into Where u0 is the minimum value of the uncertainty variable, To determine the 0-1 control variable of the uncertain variable, △u is the fluctuation range of the uncertain variable; at this time, the nonlinear term is in the form of the product of binary variables and continuous variables, and auxiliary variables and related constraints are introduced for linearization; the details are as follows: In the formula, μ is the product of the 0-1 variable and the continuous variable in the original nonlinear term; M is a positive integer; z is the auxiliary variable introduced to control the value of the nonlinear term; Then the C&CG algorithm iteratively solves the main problem MP and the subproblem SP until the gap between the next bound given by the main problem MP and the upper bound given by the subproblem SP is reduced to a termination threshold k, and outputs the corresponding solution; Multiple different robust problems can have different solutions, thus producing a Pareto front.
8. A multi-objective robust expansion planning method for a distribution network taking into account source-load uncertainty according to claim 7, characterized in that: In step 6, the solutions in the Pareto front form a decision matrix according to the target value, and then make a decision according to the following steps: 1) Obtain the target n positive ideal solution from the decision matrix With negative ideal solution In the formula, are all the elements in the nth column of the decision matrix; 2) Calculate the closeness of each decision plan to the positive ideal solution Closeness to negative ideal solution 3) Calculate the evaluation index f m : The evaluation indicators are sorted and the maximum value is the optimal solution.
Citation Information
Cited By
Method and system for quickly solving dynamic partition model of power grid based on single-stage robustness
CN119358295A
Power distribution network multi-stage extension planning method and device considering carbon emission reduction
CN119482457A
Distributed photovoltaic energy storage bundling planning method considering photovoltaic uncertainty
CN119582323A
Power distribution network robust optimization scheduling method based on line loss and new energy output uncertainty
CN119674936A
Data-driven urban power grid multi-stage operation scheduling method and system
CN119813194A
