Hybrid optimization algorithm for double-layer programming model of rotary power flow controller and energy router

Abstract

The invention provides a hybrid optimization algorithm for a double-layer planning model of a rotary power flow controller and an energy router. According to the invention, a hybrid optimization algorithm combining an improved particle swarm optimization algorithm and second-order cone programming is formed for a double-layer programming model of a rotation power flow controller and an energy router based on space-time characteristics. The algorithm comprises the following steps of: solving an upper model by adopting a time-varying coefficient improved particle swarm optimization (IPSO) algorithm; and second-order cone relaxation processing is adopted for the lower-layer model. And outputting an optimal operation result of the RPFC and the ER through a hybrid optimization algorithm according to the framework of the double-layer coordinated planning model and the solving process of the model in combination with the objective function and the constraint condition of the double-layer model. Finally, through the joint access of the RPFC and the ER, the voltage spatial and temporal distribution of the power distribution network can be effectively improved, and the annual investment and operation cost is reduced.

Description

Technical Field

[0001] The present invention belongs to the field of solving models of transmission and distribution power grids, and particularly relates to a hybrid optimization algorithm for a double-layer planning model of a rotating power flow controller and an energy router. Background Art

[0002] With the increasing proportion of new energy access year by year, the problems of energy imbalance in the distribution network system in the time and space dimensions and the complication of voltage / power flow distribution have become more prominent. The combined introduction of an electromagnetic rotating power flow controller (RPFC) and an energy router (ER) can improve the high-proportion accommodation and power supply reliability of a new type of active distribution network (ADN) from two dimensions of time and space.

[0003] Design a double-layer planning scheme according to the optimal positions and capacities of the RPFC and the optimal positions and capacities of the ER. The upper-layer model is the site selection and capacity determination layer, with the investment and construction costs of the RPFC and the ER as the optimization objectives; the lower layer is the optimal operation layer, with the lowest operation cost of the distribution network as the objective.

[0004] The particle swarm optimization algorithm is an evolutionary computing technology. Its basic idea is to find the optimal solution through cooperation and information sharing among individuals. The advantages are simple to implement and do not require adjustment of many parameters, and the convergence speed is fast. When solving the upper-layer model, a time-varying coefficient improved particle swarm algorithm (IPSO) is used to improve the calculation performance.

[0005] Considering that the optimal operation model of the distribution network is a non-linear and non-convex model, it is necessary to perform second-order cone relaxation processing on the equations of RPFC, ER constraints, and distribution network operation constraints.

[0006] Based on the above background, for the configuration problem of considering the access of RPFC and ER to an AC-DC hybrid active distribution network, a hybrid optimization algorithm combining an improved particle swarm algorithm (IPSO) and second-order cone programming (SOCP) is required to solve the double-layer planning model, and finally realize improving the high-proportion accommodation and power supply reliability of a new type of active distribution network from two dimensions of time and space. Summary of the Invention

[0007] The purpose of the present invention is to solve the double-layer planning model of RPFC and ER based on time and space characteristics. To achieve the above purpose, the present invention provides a hybrid optimization algorithm for a double-layer planning model of a rotating power flow controller and an energy router.

[0008] The technical solution of the present invention is to propose a hybrid optimization algorithm combining an improved particle swarm algorithm and second-order cone programming for the configuration problem of considering the access of RPFC and ER to an AC-DC hybrid active distribution network, and decompose the investment costs of the RPFC and the ER and the optimal operation cost of the distribution network into two-layer optimization problems.

[0009] The site selection planning of RPFC and ER is regarded as the upper-layer model. The upper-layer model aims to minimize the equivalent annual cost of investment and construction during the ADN planning period, including: the investment cost of RPFC, FRPFC, the investment cost of ER (including the construction cost of ER body, FERD, and the construction costs of supporting DG, FDG, ES, and V2G, FES, FV2G), FER, and the civil engineering cost, FCE.

[0010] The constraint conditions of the upper-layer model include: the installation location, capacity, and power constraints of RPFC, and the capacity and power constraints.

[0011] Aiming at the defect that the particle swarm algorithm is prone to falling into local optimum, a time-varying coefficient improved particle swarm algorithm (IPSO) is adopted to solve the upper-layer model, so as to obtain the location and capacity information of RPFC and ER in the distribution network.

[0012] Based on the IPSO algorithm, confirm the access location and capacity set of RPFC and ER, initialize the parameters of the particle swarm algorithm, and generate the initial population of RPFC and ER, and transfer the RPFC and ER schemes generated by the upper layer to the lower layer.

[0013] The lower-layer model is the operation model, aiming to meet the lowest cost under the safe power supply operation of the distribution network, including the main network power purchase cost, F buy , the operation and maintenance cost of RPFC, F O-RPFC , the operation and maintenance cost of ER, F O-ER , the penalty cost for light curtailment, F cur and the loss cost, F loss .

[0014] The constraint conditions of the lower-layer model, while satisfying the upper-layer constraint conditions, also need to satisfy the active distribution network safe operation constraints, including: ADN safe operation constraints, ER internal port ESS operation constraints, ER internal port V2G operation constraints, photovoltaic output constraints, and RPFC operation constraints.

[0015] Considering that the distribution network optimal operation model is a non-linear and non-convex model, the second-order cone programming is carried out on the models of RPFC and ER to obtain the optimal operation objective value. Based on the solution results of the lower-layer model, output the objective function of the lower-layer operation model, and at this time, a local optimal solution is obtained.

[0016] Update the particle velocity and position according to the upper-layer model theoretical analysis, and perform iterative calculation until the maximum number of iterations is satisfied. At this time, output the global optimal solution of the model, and evaluate the equivalent annual comprehensive cost of the distribution network at this time. Description of the Drawings

[0017] Figure 1 is the framework of the RPFC and ER two-layer coordinated planning model

[0018] Figure 2 Solving diagram for the double - layer programming model based on IPSO and SOCP

[0019] Figure 3 Movement of the classical particle swarm optimization algorithm

[0020] Figure 4 Flow chart for solving the double - layer model Specific implementation manner

[0021] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0022] Refer to Figure 1 The RPFC and ER double - layer coordinated planning model framework in []. The upper - layer model is the site - selection and capacity - determination layer, with the investment and construction costs of RPFC and ER as the optimization objectives; the lower - layer is the optimal operation layer, with the lowest operation cost of the distribution network as the objective.

[0023] The upper - layer objective function includes: the investment cost FRPFC of RPFC and the investment cost of ER, and the expressions are as follows:

[0024]

[0025] In the formula, θ is the equal - annual - value operator, which is used to convert the total cost into the equal - annual - value cost; where d and L T are the discount rate and the asset life cycle respectively.

[0026] The constraint conditions of the upper - layer model include: the installation location, capacity and power constraints of RPFC and the capacity and power constraints.

[0027] The installation location, capacity and power constraints of RPFC are:

[0028]

[0029] In the formula: and are the active power, reactive power and transmission loss flowing through nodes i and j at time t respectively; are the active power, reactive power and installed capacity transmitted by branch ij respectively; are the maximum active - power regulation and maximum reactive - power regulation ranges of RPFC; z RPFC is the node where RPFC is connected, and z ij is the line where RPFC is allowed to be connected.

[0030] The capacity and power constraints are as follows:

[0031]

[0032] In the formula: and are the AC active power and reactive power of the energy router access node i at time t; and are the active and reactive exchange powers of the internal energy storage, photovoltaic, V2G and network loss of the energy router respectively; and are the capacities of ESS, V2G, DG and ER respectively; and are the maximum active power regulation and maximum reactive power regulation ranges of ER respectively; z ER is the ER access node, z i is the distribution network allowed access node.

[0033] The expression of the lower-layer objective function is:

[0034]

[0035] In the formula: W u is the set of typical day scenarios; u represents the current typical day scenario; D u is the probability of the u typical day scenarios; N is the set of distribution network nodes; t is the corresponding time period of the typical day; S buy,i is the load size of node i; C EC,t is the peak-valley electricity price corresponding to time period t; c PVG is the unit penalty cost for DG light curtailment, is the light curtailment power; k 1 、k 2 、k 3 、k 4 、k 5 are the operation and maintenance cost coefficients of RPFC, ER, ESS, V2G and DG respectively; is the total network loss at time t; are the line, RPFC and ER operation losses at time t respectively; is the current of line ij at time t; r ij is the resistance of line ij.

[0036] The constraint conditions of the lower-layer model include: ADN safe operation constraint, ER internal port ESS operation constraint, ER internal port V2G operation constraint, photovoltaic output constraint and RPFC operation constraint.

[0037] The ADN safe operation constraint is:

[0038]

[0039]

[0040] Where: P ij,t and Q ij,t are the powers flowing through node ij of line ij in period t respectively; x ij is the reactance of line ij; W N is the branch set; U i,t , U j,t are the voltages of nodes i and j at time t respectively; U i.min and U i.max are the lower and upper limit values of the voltage of node i respectively; I ij.max is the maximum value of the line current.

[0041] ESS operation constraints for the internal port of ER:

[0042]

[0043] Where: and are the ESS capacities in periods t and t + 1 respectively; h ESSc is the ESS charging efficiency; h ESSd is the ESS discharging efficiency; is the charging power; is the ESS discharging power; ε ESS is the ESS self-discharge rate; and are the minimum and maximum ESS capacities respectively; and are the minimum and maximum charging powers of the ESS respectively; and are the minimum and maximum discharging powers of the ESS respectively; δ ESSc and δ ESSd represent the ESS discharging and charging states (state variables: 0 / 1).

[0044] V2G operation constraints for the internal port of ER:

[0045]

[0046] Where: and are the V2G capacities in periods t and t + 1 respectively; h V2Gc is the V2G charging efficiency; h V2Gd is the V2G discharging efficiency; is the charging power; is the V2G discharging power; ε V2G is the V2G self-discharge rate; and are the minimum and maximum capacities of V2G, respectively; and are the minimum and maximum charging powers of V2G, respectively; and are the minimum and maximum discharging powers of V2G, respectively; δ V2Gc and δ V2Gd represent the discharging and charging states of V2G (state variables: 0 / 1).

[0047] The photovoltaic output constraint is:

[0048] 0 ≤ P DG (t) ≤ P DG,max

[0049] where: P DG (t) is the real-time photovoltaic output, and P DG,max is the photovoltaic access capacity.

[0050] The RPFC operation constraint is:

[0051]

[0052] where: is the rated regulation capacity of RPFC.

[0053] Refer to Figure 2 The solution diagram of the bi-level programming model based on IPSO and SOCP. The improved particle swarm algorithm is used to solve the active distribution network morphological structure with the lowest investment cost, including the positions and capacities of RPFC and ER; the second-order cone programming is used to solve the lowest operation cost under the current distribution structure morphology, including the active and reactive powers of RPFC, ESS, and V2G.

[0054] Refer to Figure 3 is the movement of the classical particle swarm algorithm. In the classical particle swarm algorithm, an individual particle is mainly composed of a position parameter and a velocity parameter:

[0055]

[0056] where: v i , ω, c i , r i , v i , pbest i , x i , gbest i are the velocity vector of the i-th particle, the inertia weight, the acceleration factor, a random number in the interval [0,1], the velocity vector of the i-th particle, the local optimal position of the i-th particle individual, the position vector of the i-th particle, and the global optimal position of the i-th particle individual, respectively.

[0057] Since the particle swarm optimization algorithm is prone to falling into local optimum solution, a time-varying coefficient improved particle swarm optimization algorithm (IPSO) is adopted to solve the upper-layer model. The key factors of the particle swarm optimization algorithm, c i are added with time-varying characteristics, and a new iteration optimal index is added to the calculation of the v i velocity vector to improve the calculation performance, as shown below:

[0058]

[0059] In the formula, w min and w max , c 1 , c 2 , c 3 , c min , c max are the minimum and maximum values of the inertia weight, the current value, the minimum value, and the maximum value of the learning factor respectively; i is the current iteration number; i max is the maximum iteration number; Ibest i represents the historical optimal solution of particle i.

[0060] Considering that the optimal operation model of the distribution network is a non-linear non-convex model, a large number of loop calculations are required to solve it using a heuristic algorithm, and the solution time is long. Therefore, second-order cone relaxation processing is performed on the RPFC, ER constraints, and distribution network operation constraints:

[0061]

[0062] Refer to Figure 4 as the flow chart for solving the bi-level model. First, basic parameters such as network parameters, operation data of photovoltaic and load, equipment parameters of RPFC and ER, energy storage parameters, and V2G prediction data are obtained to determine the initial conditions for model establishment and solution.

[0063] Based on the IPSO algorithm, confirm the access positions and capacity sets of RPFC and ER, initialize the parameters of the particle swarm optimization algorithm, and generate the initial populations of RPFC and ER. Transmit the RPFC and ER schemes generated by the upper layer to the lower layer, and calculate the objective function of the upper-layer planning model at this time.

[0064] Based on the distribution network structure generated by the upper-layer model, use the SOCP model to solve the lower-layer optimal power flow considering the output of RPFC and ER. Output the objective function of the lower-layer operation model based on the solution results of the lower-layer model, and a local optimal solution is obtained at this time.

[0065] Update the particle velocity and position according to the objective function and constraint conditions of the upper-layer model, and perform iterative calculation until the maximum iteration number is satisfied. At this time, output the global optimal solution of the model.

[0066] As described above, the present invention has been described in detail. Obviously, any modifications that are substantially without departing from the inventive points and effects of the present invention and are obvious to those skilled in the art are also included within the protection scope of the present invention.

Claims

1. A hybrid optimization algorithm for a rotating power flow controller and an energy router double-level programming model, characterized in that: A hybrid optimization algorithm based on the solution of a two-level programming model is provided; the invention patent includes: for the two-level programming model of the rotating power flow controller and the energy router based on the time-space characteristics, a hybrid optimization algorithm combining an improved particle swarm algorithm and a second-order cone programming is formed, and the algorithm is a time-varying coefficient improved particle swarm algorithm (IPSO) for solving the upper model and a second-order cone relaxation process for the lower model; finally, through the joint access of RPFC and ER, the application can effectively improve the time-space distribution of the distribution network voltage and reduce the annual investment and operating costs; The upper model of the two-layer model is the site selection and capacity determination layer, which takes the investment and construction costs of RPFC and ER as the optimization target; the lower layer is the operation optimization layer, which takes the lowest distribution network operation cost as the target; The upper model takes the minimum annual cost of investment and construction during the ADN planning period as the objective function, including: the investment cost of RPFC FRPFC, the investment cost of ER (including the ER main construction cost FERD and the supporting DG construction cost FDG, ES construction cost FES, V2G construction cost FV2G) FER, and the civil construction cost FCE; The constraints of the upper model include: RPFC installation location, capacity and power constraints, and capacity and power constraints; The upper model is a planning model. The site selection planning of RPFC and ER is used as the upper model. Aiming at the defect that the particle swarm algorithm is easy to fall into the local optimum, an improved particle swarm algorithm with time-varying coefficient (IPSO) is used to solve the upper model to obtain the location and capacity information of RPFC and ER in the distribution network. The IPSO algorithm is used to confirm the access location and capacity set of RPFC and ER, initialize the particle swarm algorithm parameters, generate the initial population of RPFC and ER, and transmit the RPFC and ER solutions generated by the upper layer to the lower layer; The lower model aims to meet the lowest cost under the safe power supply operation of the distribution network, including the main grid power purchase cost F buy , RPFC operation and maintenance cost F O-RPFC , ER operation and maintenance cost F O-ER , Abandonment penalty fee F cur and loss cost F loss； The constraints of the lower model include: ADN safety operation constraints, ER internal port ESS operation constraints, ER internal port V2G operation constraints, photovoltaic output constraints and RPFC operation constraints; The lower model is the operation model. Considering that the distribution network optimization operation model is a nonlinear non-convex model, the RPFC and ER models are processed by second-order cone programming to obtain the optimized operation target value; The generated distribution network structure uses the SOCP model to solve the lower optimal power flow considering the RPFC and ER outputs, and outputs the lower operation model objective function based on the solution result of the lower model, thereby obtaining a local optimal solution; The particle speed and position are updated according to the theoretical analysis of the upper model, and the iterative calculation is performed until the maximum number of iterations is met. At this time, the global optimal solution of the model is output, and the annual comprehensive cost of the distribution network at this time is evaluated.

2. According to the solution process of the double-layer model described in claim 1, combined with the objective function and constraints of the double-layer model, a mathematical analysis is performed on the hybrid optimization algorithm; In view of the defect that the particle swarm algorithm is easy to fall into the local optimum, an improved particle swarm algorithm (IPSO) with time-varying coefficients is used to solve the upper model, and its expression is: In the formula, w min 、w max , c1, c2, c3, c min 、c max They are the minimum and maximum values ​​of the inertia weight, the current value, minimum value, and maximum value of the learning factor respectively; i is the current iteration number; i max is the maximum number of iterations; Ibest i represents the historical optimal solution of particle i; The distribution network optimization operation model is a nonlinear non-convex model. The heuristic algorithm needs a lot of loop calculations to solve it, which takes a long time. Therefore, the RPFC, ER constraints and distribution network operation constraints are subjected to second-order cone relaxation, and the expression is:

Images