Low-voltage power distribution network distributed energy storage double-layer optimization method considering distributed power supply access

Abstract

A low-voltage power distribution network distributed energy storage double-layer optimization method considering distributed power supply access comprises the following steps: constructing a network equation of a distributed power supply and distributed energy storage based on the randomness of distributed power supply output; establishing a double-layer optimization model by considering distributed power supply access and a distributed energy storage configuration strategy; aiming at a distributed power supply and distributed energy storage collaborative double-layer optimization configuration problem, adopting a strategy of combining an improved particle swarm algorithm and a second-order cone-convex relaxation linear programming technology to solve an established double-layer optimization model; and carrying out comprehensive evaluation on the double-layer optimization model based on the distributed power supply and the distributed energy storage. The method can significantly improve the reliability and adaptability of the low-voltage power distribution network system, improve the reliability and flexibility of the low-voltage power distribution network system, ensure effective utilization of local power generation resources and energy storage capability in the low-voltage power distribution network, effectively solve the loss problem caused by access of new energy to the low-voltage power distribution network, and improve the reliability and flexibility of the low-voltage power distribution network system. Economical and stable operation of the low-voltage distribution network is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of low-voltage distribution network loss optimization, and in particular to a two-layer optimization method for distributed energy storage in a low-voltage distribution network taking into account access of distributed power sources. Background Art

[0002] With the integration of distributed generation (DG) into the grid, low-voltage distribution networks are facing unprecedented opportunities and challenges. The popularity of distributed generation (DG) means that more and more DGs are connected to low-voltage distribution networks. This transformation has greatly enriched the sources of power supply and improved the flexibility and reliability of the system. my country's photovoltaic and wind power installed capacity continues to grow, especially in low-voltage distribution networks, where the popularity of distributed photovoltaics is particularly significant. However, this also brings new technical requirements and challenges to the management and operation of low-voltage distribution networks. By optimizing the configuration and scheduling strategy of distributed energy storage systems (DESS), the integration of distributed energy storage can effectively reduce voltage over-limit losses, increase the utilization rate of renewable energy, and improve the flexibility and stability of the system. Therefore, solving the problem of over-limit losses in low-voltage distribution networks is of great significance to engineering practice.

[0003] Existing research addressing voltage overshoot losses in low-voltage distribution networks primarily focuses on load management, line loss analysis, and voltage control strategies. However, while optimizing energy storage configuration to mitigate losses caused by voltage fluctuations is a relatively limited approach, there is a lack of systematic research specifically focused on optimizing line losses in low-voltage distribution networks. While there has been some research on distributed generation and demand-side response, there is a lack of systematic research specifically on optimizing line losses in low-voltage distribution networks.

[0004] Existing technologies involving distributed energy storage dual-layer optimization include:

[0005] The Chinese patent "A two-layer optimization configuration method for distributed energy storage systems" (application number: 202411169087.5) first uses an improved particle swarm algorithm to solve the upper-layer model with the comprehensive annual cost and annual income of energy storage as the target, and obtains the optimal result of energy storage site selection and sizing. With the annual photovoltaic income as the objective function, a lower-layer optimization model is constructed to coordinate the output coordination between photovoltaics and energy storage, and solve the lower-layer optimization model to obtain the optimal operation strategy of energy storage charging and discharging.

[0006] The Chinese patent "A distributed hybrid cloud energy storage collaborative configuration method" (application number: 202411278270.9) discloses a distributed hybrid cloud energy storage collaborative configuration method, including grid cluster division and double-layer energy storage one-by-one planning and optimization configuration. Through power system simulation, cluster division based on electrical distance and power balance conditions is realized according to the grid structure. Based on the cluster division results, double-layer one-by-one configuration of multiple energy storage is realized for the overall distribution network.

[0007] The Chinese patent "A two-layer model optimization method for the site selection and sizing of energy storage in a distribution network containing distributed power sources" (application number: 202311713283.X) discloses a two-layer model optimization method for the site selection and sizing of energy storage in a distribution network containing distributed power sources. The upper layer model is the planning layer, which takes the energy storage investment cost as the goal, and the lower layer is the operation layer, which takes the minimization of energy storage operation and maintenance cost and network loss cost as the goal, and the energy storage output strategy as the variable. The improved multi-objective particle swarm algorithm and particle swarm algorithm are used to optimize and solve the two-layer model.

[0008] The patents in the aforementioned prior art have made certain contributions to the field of distributed energy storage two-tier optimization, but most of them currently lack intelligent management of energy storage equipment in their scheduling strategies. After the integration of renewable energy, how to effectively utilize distributed energy storage to shave peaks and fill valleys and smooth out the power flowing into and out of the grid is an urgent problem that needs to be solved. Research on the coordinated optimization of distributed energy storage and loads in low-voltage distribution networks with DG integration is relatively limited. Furthermore, the impact of timing fluctuations and the grid's ability to accommodate renewable energy have not been fully explored and analyzed. Existing optimization models are mostly static models that fail to fully account for the inherent randomness and volatility of DG power generation. In practical applications, timing variations in load demand and power generation capacity can lead to unstable system performance. Therefore, further research in these areas will help promote the effective application of energy storage in low-voltage distribution networks, promote the wider use of renewable energy, and promote the sustainable development of power systems. Summary of the Invention

[0009] To solve the above technical problems, the present invention proposes a two-layer optimization method for distributed energy storage in a low-voltage distribution network taking into account the access of distributed power sources. This method can significantly improve the reliability and adaptability of the low-voltage distribution network system, enhance the reliability and flexibility of the low-voltage distribution network system, ensure the effective use of local power generation resources and energy storage capacity in the low-voltage distribution network, effectively solve the loss problem caused by the access of new energy to the low-voltage distribution network, and realize the economical and stable operation of the low-voltage distribution network.

[0010] The technical solution adopted by the present invention is:

[0011] A two-layer optimization method for distributed energy storage in a low-voltage distribution network taking into account the access of distributed power sources comprises the following steps:

[0012] Step 1: Based on the randomness of distributed power output, construct the network equations of distributed power and distributed energy storage;

[0013] Step 2: Considering the distributed generation access and distributed energy storage configuration strategies, a two-layer optimization model is established;

[0014] Step 3: For the coordinated two-level optimization configuration problem of distributed power generation and distributed energy storage, an improved particle swarm optimization (PSO) algorithm is combined with the second-order cone-convex relaxation linear programming technology to solve the two-level optimization model established in step 2.

[0015] In this two-layer optimization model, the upper layer optimizes the capacity configuration of distributed generation and distributed energy storage using an improved particle swarm optimization algorithm, ensuring the effective utilization of local generation resources and energy storage capacity within the low-voltage distribution network. The lower layer, using CPLEX, employs second-order cone programming to accurately solve the daily scheduling plan, optimizing the charging and discharging strategies of the energy storage units to meet changes in power load and fluctuations in grid demand. The results of the lower layer's daily scheduling strategy are fed back to the upper layer, enabling interaction between the upper and lower layers. This improves the overall performance of the low-voltage distribution system and effectively coordinates the relationship between distributed generation and energy storage systems, which has a certain impact on practical engineering.

[0016] In step 1, the network equations of distributed power sources and distributed energy storage are constructed, specifically including:

[0017] 1) Photovoltaic power generation model:

[0018] The active power output of photovoltaic power generation is mainly affected by the local light intensity. The light intensity r is usually assumed to follow the Beta distribution, and its probability density function is:

[0019]

[0020] In formula (1), f(pv) is the probability density function of photovoltaic power output power; Γ(α+β) is the gamma function of the sum of shape parameters used to describe the Beta distribution characteristics; Γ(α) is the value of the gamma function at parameter α; Γ(β) is the value of the gamma function at parameter β; I, I pv are the actual value and rated value of light intensity respectively; ɑ and β are the shape parameters of Bate; Γ is the gamma function;

[0021] Among them, the photovoltaic output power P pv The relationship between the actual value of the light intensity I is:

[0022]

[0023] In formula (2): P r is the rated power of the photovoltaic.

[0024] 2) Wind power generation model:

[0025] The change in wind turbine output is usually related to wind speed, which is considered to be closer to the Weibull distribution. Its probability density function is:

[0026]

[0027] In formula (3), f(v) is the probability density function of wind power output; c is the scale parameter, k is the shape parameter, and v is the average wind speed in the region.

[0028] The actual output power P of the fan WT The relationship between it and wind speed v can be expressed as:

[0029]

[0030] In formula (4): P W is the fan output power; P WT is the rated output power of the fan; v r , v in , v co They are respectively the rated wind speed, cut-in wind speed and cut-out wind speed of the fan.

[0031] 3) Energy storage system model:

[0032] The state of charge (SOC) of an energy storage system is an important indicator for measuring its charge and discharge capabilities. The change in the energy storage system's SOC at different time points is closely related to its current state and can be expressed using the following formula:

[0033]

[0034] In formula (5): SOC t+1 The state of charge at the next moment; SOC t is the state of charge at the current moment; P charge P is the charging power of the energy storage system at the current moment (positive value); discharge is the discharge power of the energy storage system at the current moment, a positive value indicates discharge, and a negative value indicates charging; C is the total capacity of the energy storage system; Δt is the time step.

[0035] SOC reflects the current charge level of the energy storage system, directly affecting the battery's charge-discharge strategy and available power. In SOC calculations, the difference between charging and discharging power determines the SOC's changing trend: if charging power exceeds discharging power, SOC increases; otherwise, it decreases. Total capacity (C) reflects the maximum energy that the energy storage system can store.

[0036] In step 2, a two-layer optimization model is established, which specifically includes:

[0037] A two-layer optimization model considering DG access is established, including Equations (6) to (19), to achieve the optimal configuration of distributed energy storage. The upper model of the two-layer optimization model aims to minimize the comprehensive investment cost of the distributed energy storage system (Equation (6)) and find the optimal capacity configuration of various energy storage devices, including Equations (6) to (13).

[0038] The lower-level model aims to minimize the network loss of the low-voltage distribution network under the energy storage capacity constraints determined by the upper-level model (Equation (14)). Subsequently, the optimization results of the lower-level model are fed back to the upper-level model to guide investment decisions.

[0039] The difference between the upper and lower objective functions (including Equations (6) and (14)) is represented by the parameter TOL. The iteration stops when TOL is less than the set value. By continuously iterating the two-layer optimization model, the optimal capacity configuration scheme of distributed energy storage can be found.

[0040] (1) Upper model objective function:

[0041] minC total =C DG +C DESS +C BUY (6);

[0042] In formula (6): C total is the total investment cost; C DG New energy investment and operating costs; C DESS is the investment cost of distributed energy storage; C BUY The cost of purchasing electricity;

[0043] ①. New energy investment and operating costs:

[0044]

[0045] In formula (7): x i Indicates that DG is installed at node i; P DG，i C is the installed capacity of the DG at node i; in,i is the investment cost per unit capacity of DG installed at node i.

[0046] ②. Distributed energy storage investment cost:

[0047]

[0048] In formula (8): y i To install the node collection; c E ,c Pare the operating costs per unit capacity and per unit power of the DESS installed at node i; E DESS ,P DESS are the rated capacity and rated power of the DESS installed at node i, respectively.

[0049] ③. Electricity purchase cost:

[0050]

[0051] In formula (9): C b The price of electricity purchased from the large power grid; P load is the load active power; P WT 、P PV is the active output power of wind power generation and photovoltaic power generation; P DESS is the active power output of DESS; △t is the length of each time period.

[0052] (2) Upper model constraints:

[0053] ①.DESS power and capacity constraints:

[0054]

[0055] In formula (10): P DESS 、P DESS，min The upper and lower limits of the energy storage rated power; E DESS,max 、E DESS，min is the maximum and minimum value of the energy storage rated capacity; n DESS is the maximum installed quantity of energy storage; P DESS is the active power value of DESS; E DESS is the capacity value of DESS; n max The upper limit of the number of DESSs allowed to be deployed in the system.

[0056] ②. Distributed power output constraints:

[0057]

[0058] In formula (11): P wind(t) is the wind power output at time t; P wind,max(t) Indicates the maximum available output of the wind power system at that moment;

[0059]

[0060] In formula (12): P solar(t) is the photovoltaic output at time t; P solar,max(t) It is the maximum output of photovoltaic power generation at that moment.

[0061]

[0062] In formula (13): P total,max(t) The maximum output limit of renewable energy is set according to the demand and dispatch capacity of the power grid.

[0063] (3) Lower model objective function:

[0064]

[0065] In formula (14): f is the network loss cost; ψ b is the system branch set; I ij,t is the effective value of the current in branch ij at time t; R ij is the resistance value of branch ij.

[0066] (4) Lower model constraints:

[0067] ①. System power flow balance constraints:

[0068]

[0069] In the above formula: P ij,t , Q ij,t are the active power and reactive power from node i to node j at time t; P i,t , Q i,t are the active load and reactive load at node i at time t, respectively.

[0070] ②.Node voltage constraints:

[0071] V i,max ≤V i,t ≤V i,max (17);

[0072] In formula (17): V i,max and V i,min are the maximum and minimum voltage amplitudes at node i respectively; V i,t is the voltage of node i at time t.

[0073] ③.Node current constraints:

[0074]

[0075] In formula (18): I ij，t is the effective value of the current flowing through ij at time t; I ij，max is the maximum current amplitude at branch ij.

[0076] ④. Energy storage charging and discharging and SOC constraints:

[0077]

[0078] In formula (19): SOCmin , SOC max are the lower and upper limits of the DESS state of charge, respectively; SOC(t) is the state of charge of the DESS at time t; P charge(t) , P discharge(t) are the charging power and discharging power of energy storage at time t respectively; P charge,max , P discharge,max are the maximum charging power and maximum discharging power of distributed energy storage DESS respectively.

[0079] In step 3, the improved particle swarm algorithm PSO specifically includes:

[0080] ①. Dynamic inertia weight:

[0081] The inertia weight w adopts a linear decreasing strategy, namely:

[0082]

[0083] In formula (20): w max 、w min are the maximum and minimum inertia weight values, t n is the current iteration number, T max is the maximum number of iterations.

[0084] ②. Adaptive learning factor:

[0085] The individual learning factor c1 and the social learning factor c2 are dynamically adjusted according to the fitness value of the particle, namely:

[0086]

[0087] In formula (21): f best 、f worst are the optimal fitness value and the worst fitness value in the particle swarm, respectively, f i is the fitness value of the current particle; c 1,min 、c 1,max are the minimum and maximum values of the individual learning factor c1; c 2,min 、c 2,max are the minimum and maximum values of the social learning factor c2, respectively.

[0088] ③. Mutation operation:

[0089] In each iteration, the position of the particle is randomly perturbed with a certain probability, that is:

[0090] x i =x i +η·randn(0,1) (22);

[0091] In formula (22): xi is the current position of the i-th particle; η is the disturbance amplitude, and randn(0,1) is a random number from the standard normal distribution.

[0092] Solving the two-level optimization model based on second-order cone programming:

[0093] By introducing intermediate variables, the key parameters in the two-level optimization model are defined. The replacement formula is:

[0094]

[0095] In formula (23): V i is the quadratic relationship of node i voltage after introducing the intermediate variable; l ij is the quadratic relationship of branch ij current after the introduction of intermediate variables; U is the node voltage value; I is the effective value of branch current.

[0096] In the two-level optimization model, constraint transformation is the key to improving the efficiency of model solution. By replacing variables, the model solution can be effectively simplified. There are nonlinear constraints in the original two-level optimization model, which are convexly relaxed by formula (23);

[0097]

[0098] In formula (24): ij is the quadratic relationship of branch ij current after introducing the intermediate variable; V i is the quadratic relationship of node i voltage after the introduction of intermediate variables; P ij , Q ij are the active power and reactive power of branch ij respectively.

[0099] The present invention transforms equations (14) to (19) in the two-level optimization model into mixed integer second-order cone programming models (23) to (25), and transforms the original nonlinear constraints into a second-order cone form that is easier to solve through relaxation technology.

[0100]

[0101] i ij,t ≤i ij,max (26);

[0102] Where: i ij,t is the current value of branch ij at time t; i ij,max is the maximum current flowing through branch ij.

[0103] Step 4: Comprehensively evaluate the two-tier optimization model based on distributed generation and distributed energy storage. Specifically, use multiple evaluation indicators, including power loss, DESS daily operation and dispatch, and DG absorption capacity, to fully reflect the comprehensive effect of the optimization scheme.

[0104] The present invention provides a two-tiered optimization method for distributed energy storage in low-voltage distribution networks that takes into account the access of distributed power sources. The technical effects are as follows: 1) By constructing a two-tiered optimization model that comprehensively considers distributed power sources (DG), distributed energy storage (DESS), and load scheduling, the present invention systematically explores effective strategies for optimizing line losses in low-voltage distribution networks. Within this framework, an improved particle swarm optimization algorithm is first used to optimize the configuration of distributed power sources and energy storage to improve the overall performance of the power grid. Secondly, a second-order cone programming method is used to optimize the scheduling strategy of the energy storage system to ensure that while meeting power demand, line losses caused by voltage anomalies are effectively reduced.

[0105] 2) This invention combines an improved particle swarm optimization algorithm with second-order cone programming. The upper-level model optimizes the capacity configuration of DG and DESS using the improved particle swarm optimization algorithm, ensuring the effective utilization of local power generation resources and energy storage capacity in the low-voltage distribution system. Meanwhile, the lower-level model uses second-order cone programming to accurately solve the daily scheduling strategy of the energy storage system, optimizing the charging and discharging strategies of the energy storage units to meet changes in power load and fluctuations in grid demand. This improves the overall performance of the low-voltage distribution network.

[0106] 3) Through the collaborative optimization strategy of the present invention, the low-voltage distribution network can better cope with load fluctuations, improve the new energy absorption rate, reduce network losses, achieve optimal energy configuration, improve the reliability and flexibility of the low-voltage distribution system, and ensure the effective use of local power generation resources and energy storage capacity in the low-voltage distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0107] The present invention will be further described below with reference to the accompanying drawings and examples:

[0108] Figure 1 This is a structural diagram of the improved IEEE1333 node system of the present invention.

[0109] Figure 2 This is a diagram of a two-layer optimization model of distributed energy storage in a low-voltage distribution network with distributed power access according to the present invention.

[0110] Figure 3 This is the improved particle swarm algorithm solution flow chart of the present invention.

[0111] Figure 4 This is a comparison chart of the timing network loss changes of Scheme 1 and Scheme 2 of the present invention.

[0112] Figure 5 This is a diagram of the daily operation scheduling changes of DG and DESS according to the present invention.

[0113] Figure 6 It is an iterative optimization configuration diagram of the improved particle swarm algorithm of the present invention.

[0114] Figure 7 This is a comparison diagram of node voltage levels at each moment in Scheme 1 and Scheme 2 of the present invention.

[0115] Figure 8 It is a comparison chart of the new energy consumption rate of the present invention.

[0116] Figure 9 This is a comparison diagram of the power quality of Scheme 1 and Scheme 2 of the present invention. DETAILED DESCRIPTION

[0117] Using the improved IEEE33 node system, such as Figure 1 The present invention provides a two-layer optimization method for distributed energy storage in a low-voltage distribution network taking into account the access of distributed power sources, which specifically includes the following steps:

[0118] Step 1: Based on the randomness of distributed power output, construct the network equations of distributed power and distributed energy storage;

[0119] Step 2: Considering the distributed generation access and distributed energy storage configuration strategies, a two-layer optimization model is established;

[0120] Step 3: For the coordinated two-layer optimization configuration problem of distributed power generation and distributed energy storage, an improved particle swarm optimization (PSO) algorithm is combined with the second-order cone-convex relaxed linear programming technology to solve the model;

[0121] Step 4: Conduct comprehensive evaluation based on the DG and DESS two-layer optimization model.

[0122] In step 4, a comprehensive evaluation is performed based on the DG and DESS two-layer optimization model. In this specific example, the specific evaluation is as follows:

[0123] To analyze the rationale for DESS integration into low-voltage distribution networks, this paper presents two comparative scenarios. This consideration reflects the low utilization of renewable energy and the rising cost of electricity purchases in the absence of energy storage support. By rationally configuring energy storage systems, efficient utilization of renewable energy can be achieved, power curtailment rates can be reduced, supply and demand in the distribution network can be balanced, and reliance on the main grid can be minimized.

[0124] In this two-layer optimization model, the upper layer optimizes the capacity configuration of distributed generation and distributed energy storage using an improved particle swarm optimization algorithm, ensuring the effective utilization of local generation resources and energy storage capacity within the low-voltage distribution network. The lower layer, using CPLEX, employs second-order cone programming to accurately solve the daily scheduling plan, optimizing the charging and discharging strategies of the energy storage units to meet changes in power load and fluctuations in grid demand. The results of the lower layer's daily scheduling strategy are fed back to the upper layer, enabling interaction between the upper and lower layers. This improves the overall performance of the low-voltage distribution system and effectively coordinates the relationship between distributed generation and energy storage systems, which has a certain impact on practical engineering.

[0125] In step 1, the network equations of distributed power sources and distributed energy storage are constructed, specifically including:

[0126] 1) Photovoltaic power generation model:

[0127] The active power output of photovoltaic power generation is mainly affected by the local light intensity. The light intensity r is usually assumed to follow the Beta distribution, and its probability density function is:

[0128]

[0129] In formula (1), f(pv) is the probability density function of photovoltaic power output power; Γ(α+β) is the gamma function of the sum of shape parameters used to describe the Beta distribution characteristics; Γ(α) is the value of the gamma function at parameter α; Γ(β) is the value of the gamma function at parameter β; I, I pv are the actual value and rated value of light intensity respectively; ɑ and β are the shape parameters of Bate; Γ is the gamma function;

[0130] Among them, the photovoltaic output power P pv The relationship between the actual value of the light intensity I is:

[0131]

[0132] In formula (2): P r is the rated power of the photovoltaic.

[0133] 2) Wind power generation model:

[0134] The change in wind turbine output is usually related to wind speed, which is considered to be closer to the Weibull distribution. Its probability density function is:

[0135]

[0136] In formula (3), f(v) is the probability density function of wind power output; c is the scale parameter, k is the shape parameter, and v is the average wind speed in the region.

[0137] The actual output power P of the fan WT The relationship between it and wind speed v can be expressed as:

[0138]

[0139] In formula (4): P W is the fan output power; P WT is the rated output power of the fan; v r , v in , v co They are respectively the rated wind speed, cut-in wind speed and cut-out wind speed of the fan.

[0140] 3) Energy storage system model:

[0141] During the day, photovoltaic power generation systems often generate more power than the grid can absorb. During this time, energy storage systems can absorb the excess energy. In the evening, when photovoltaic power generation decreases and electricity demand increases, the energy storage system releases energy to compensate for the grid shortfall. The state of charge (SOC) of an energy storage system is a key indicator of its charge and discharge capabilities. The changes in the energy storage system's SOC at different points in time are closely related to its current state and can be expressed using the following formula:

[0142]

[0143] In formula (5): SOC t+1 The state of charge at the next moment; SOC t It is the current state of charge, usually expressed as a percentage, ranging from 0% to 100%; P charge P is the charging power of the energy storage system at the current moment (positive value); discharge is the discharge power of the energy storage system at the current moment, a positive value indicates discharge, and a negative value indicates charging; C is the total capacity of the energy storage system; Δt is the time step.

[0144] SOC reflects the current charge level of the energy storage system, which directly affects the battery charge and discharge strategy and available power. In the SOC calculation, the difference between the charging power and the discharging power determines the change trend of the SOC, that is, if the charging power is greater than the discharging power, the SOC will increase; otherwise, it will decrease. The total capacity C reflects the maximum energy that can be stored in the energy storage system, which is crucial for the management and optimization of battery use. The change in SOC is particularly important for the scheduling and optimization of energy storage systems, especially in an environment where renewable energy is integrated. By reasonably controlling the charging and discharging process of energy storage, the stability and reliability of the power grid can be effectively improved, and more efficient energy utilization can be achieved.

[0145] In step 2, a two-layer optimization model is established, such as Figure 2 As shown, specifically including:

[0146] A two-layer optimization model considering DG access is established, including Equations (6) to (19), to achieve the optimal configuration of distributed energy storage. The upper model of the two-layer optimization model aims to minimize the comprehensive investment cost of the distributed energy storage system (Equation (6)) and find the optimal capacity configuration of various energy storage devices, including Equations (6) to (13).

[0147] The lower-level model aims to minimize the network loss of the low-voltage distribution network under the energy storage capacity constraints determined by the upper-level model (Equation (14)). Subsequently, the optimization results of the lower-level model are fed back to the upper-level model to guide investment decisions.

[0148] The difference between the upper and lower objective functions (including Equations (6) and (14)) is represented by the parameter TOL. The iteration stops when TOL is less than the set value. By continuously iterating the two-layer optimization model, the optimal capacity configuration scheme of distributed energy storage can be found.

[0149] (1) Upper model objective function:

[0150] minC total =C DG +C DESS +C BUY (6);

[0151] In formula (6): C total is the total investment cost; C DG New energy investment and operating costs; C DESS is the investment cost of distributed energy storage; C BUY The cost of purchasing electricity;

[0152] ①. New energy investment and operating costs:

[0153]

[0154] In formula (7): x i Indicates that DG is installed at node i; P DG，i C is the installed capacity of the DG at node i; in,i is the investment cost per unit capacity of DG installed at node i.

[0155] ②. Distributed energy storage investment cost:

[0156]

[0157] In formula (8): y i To install the node collection; c E ,c P are the operating costs per unit capacity and per unit power of the DESS installed at node i; E DESS ,PDESS are the rated capacity and rated power of the DESS installed at node i, respectively.

[0158] ③. Electricity purchase cost:

[0159]

[0160] In formula (9): C b The price of electricity purchased from the large power grid; P load is the load active power; P WT 、P PV is the active output power of wind power generation and photovoltaic power generation; P DESS is the active power output of DESS; △t is the length of each time period.

[0161] (2) Upper model constraints:

[0162] ①.DESS power and capacity constraints:

[0163]

[0164] In formula (10): P DESS 、P DESS，min The upper and lower limits of the energy storage rated power; E DESS,max 、E DESS，min is the maximum and minimum value of the energy storage rated capacity; n DESS is the maximum installed quantity of energy storage; P DESS is the active power value of DESS; E DESS is the capacity value of DESS; n max The upper limit of the number of DESSs allowed to be deployed in the system.

[0165] ②. Distributed power output constraints:

[0166]

[0167] In formula (11): P wind(t) is the wind power output at time t; P wind,max(t) Indicates the maximum available output of the wind power system at that moment, usually determined by wind speed and wind turbine characteristics;

[0168]

[0169] In formula (12): P solar(t) is the photovoltaic output at time t; P solar,max(t) It is the maximum output of photovoltaic power generation at that moment, which is usually affected by the intensity of solar radiation and the efficiency of photovoltaic modules.

[0170]

[0171] In formula (13): P total,max(t) The maximum output limit of renewable energy is set according to the demand and dispatch capacity of the power grid.

[0172] (3) Lower model objective function:

[0173]

[0174] In formula (14): f is the network loss cost; ψ b is the system branch set; I ij,t is the effective value of the current in branch ij at time t; R ij is the resistance value of branch ij.

[0175] (4) Lower model constraints:

[0176] ①. System power flow balance constraints:

[0177]

[0178] In the above formula: P ij,t , Q ij,t are the active power and reactive power from node i to node j at time t; P i,t , Q i,t are the active load and reactive load at node i at time t, respectively.

[0179] ②.Node voltage constraints:

[0180] V i,max ≤V i,t ≤V i,max (17);

[0181] In formula (17): V i,max and V i,min are the maximum and minimum voltage amplitudes at node i respectively; V i,t is the voltage of node i at time t.

[0182] ③.Node current constraints:

[0183]

[0184] In formula (18): I ij，t is the effective value of the current flowing through ij at time t; I ij，max is the maximum current amplitude at branch ij.

[0185] ④. Energy storage charging and discharging and SOC constraints:

[0186]

[0187] In formula (19): SOC is an important indicator to measure the efficiency of energy storage charge and discharge; SOC min , SOC max are the lower and upper limits of the DESS state of charge, respectively; SOC(t) is the state of charge of the DESS at time t; P charge(t) , P discharge(t) are the charging power and discharging power of energy storage at time t respectively; P charge,max , P discharge,max are the maximum charging power and maximum discharging power of distributed energy storage DESS respectively.

[0188] In step 3, Figure 3 The improved PSO solution flow chart shown in the figure makes the following improvements to the particle swarm algorithm PSO:

[0189] 1) Introducing a dynamic inertia weight strategy, by dynamically changing the inertia weight as the iteration progresses, the algorithm balances the global and local search capabilities, gradually enhancing the local search capability as the optimal solution is approached. This balancing mechanism helps particles efficiently search the solution space and avoid being trapped in a local optimum.

[0190] 2) Adopting adaptive learning factors, during the particle search process, the individual learning factor and social learning factor are dynamically adjusted according to the current optimal state of the particle and the optimal state of the entire group. When the fitness of the particle improves, the weight of the individual learning factor can be increased, so that the particle can rely more on its own experience; when the fitness does not improve significantly, the weight of the social learning factor can be increased, so that the particle pays more attention to the overall search direction of the group. Such an adaptive mechanism can more effectively guide the particles to converge to the optimal area;

[0191] 3) Introducing mutation operations to enhance the diversity of the particle swarm and avoid premature convergence. During the iteration process, a certain proportion of particles are randomly selected for mutation, changing their position or speed, so that the particle swarm maintains diversity in the solution space and prevents the algorithm from stagnating at a local optimal solution. Through mutation, the algorithm can continue to explore areas that have not been fully explored, increasing the probability of finding the global optimal solution;

[0192] The two-level optimization model uses CPLEX to accurately solve the daily scheduling plan. In the present invention, CPLEX will be used to accurately solve the optimization model obtained by the improved PSO and second-order cone convex relaxed linear programming (SOCP) to ensure the optimality and feasibility of the final scheduling plan.

[0193] The improved particle swarm algorithm PSO specifically includes:

[0194] ①. Dynamic inertia weight:

[0195] The inertia weight w adopts a linear decreasing strategy, namely:

[0196]

[0197] In formula (20): w max 、w min are the maximum and minimum inertia weight values, t n is the current iteration number, T max is the maximum number of iterations.

[0198] ②. Adaptive learning factor:

[0199] The individual learning factor c1 and the social learning factor c2 are dynamically adjusted according to the fitness value of the particle, namely:

[0200]

[0201] In formula (21): f best 、f worst are the optimal fitness value and the worst fitness value in the particle swarm, respectively, f i is the fitness value of the current particle; c 1,min 、c 1,max are the minimum and maximum values of the individual learning factor c1; c 2,min 、c 2,max are the minimum and maximum values of the social learning factor c2, respectively.

[0202] ③. Mutation operation:

[0203] In each iteration, the position of the particle is randomly perturbed with a certain probability, that is:

[0204] x i =x i +η·randn(0,1) (22);

[0205] In formula (22): x i is the current position of the i-th particle; η is the disturbance amplitude, and randn(0,1) is a random number from the standard normal distribution.

[0206] Solving the two-level optimization model based on second-order cone programming:

[0207] By introducing intermediate variables, the key parameters in the two-level optimization model are defined. The replacement formula is:

[0208]

[0209] In formula (23): V i is the quadratic relationship of node i voltage after introducing the intermediate variable; l ijis the quadratic relationship of branch ij current after the introduction of intermediate variables; U is the node voltage value; I is the effective value of branch current.

[0210] In the two-level optimization model, constraint transformation is the key to improving the efficiency of model solution. By replacing variables, the model solution can be effectively simplified. There are nonlinear constraints in the original two-level optimization model, which are convexly relaxed by formula (23);

[0211]

[0212] In formula (24): ij is the quadratic relationship of branch ij current after introducing the intermediate variable; V i is the quadratic relationship of node i voltage after the introduction of intermediate variables; P ij , Q ij are the active power and reactive power of branch ij respectively.

[0213] The present invention transforms equations (14) to (19) in the two-level optimization model into mixed integer second-order cone programming models (23) to (25), and transforms the original nonlinear constraints into a second-order cone form that is easier to solve through relaxation technology.

[0214]

[0215] Where: i ij,t is the current value of branch ij at time t; i ij,max is the maximum current flowing through branch ij.

[0216] Verification example:

[0217] Option 1: No energy storage is configured, and the excess power from renewable energy is directly abandoned as wind power. When the power in the distribution network is unbalanced, electricity is directly purchased from the main grid.

[0218] Solution 2: According to the two-level optimization strategy proposed in this invention, the coordinated optimization configuration of DG and DESS is considered.

[0219] The overall configuration scheme of DG and DESS is obtained as shown in Table 1.

[0220] Table 1 DG and DESS optimization configuration scheme

[0221]

[0222]

[0223] Preferably, in step 4, when selecting the most appropriate solution, multiple factors such as comprehensive cost need to be further considered. The comparison of economic indicators of Solution 1 and Solution 2 is shown in Table 2.

[0224] Table 2 Economic benefits under different schemes

[0225]

[0226] In Option 1, since DESS is not configured, the DESS investment cost is zero, and the new energy investment cost is 446,100 yuan. Although the initial investment is low, the lack of an energy storage system during operation leads to higher electricity purchase costs from the upstream network and higher system network loss costs, indicating that there is wind and solar curtailment. In Option 2, DESS is introduced, the investment cost is 261,900 yuan, and the electricity purchase costs and system network loss costs are reduced by 461,700 yuan and 36,500 yuan respectively, reducing the overall system operating cost by 361,000 yuan. This operation strategy is profitable.

[0227] The DESS is optimized for the low-voltage distribution network connected to DG. The time series network loss changes of Scheme 1 and Scheme 2 are compared. Figure 4 As shown, the annual network loss for Scenario 1 is 34 kWh, while for Scenario 2 it is 171.44 kW, a 14.6% reduction. Further analysis of the temporal distribution of network losses over a 24-hour period reveals that once renewable energy generation in Scenario 1 exceeds the actual load, the excess power is discarded. Furthermore, when power imbalances occur in the distribution network, power must be purchased from the main grid to meet load demand. In this scenario, network losses are primarily due to wind and solar curtailment, making them more significant.

[0228] In contrast, Option 2 utilizes a dual-tier optimization configuration, combining DG and DESS. By introducing DESS, when renewable energy generation exceeds the load, the energy storage system can promptly absorb the excess power, avoiding wind and solar curtailment and improving the efficiency of renewable energy use. This reduces reliance on power purchased from the main grid, minimizing network losses and demonstrating lower network losses across all time periods, significantly reducing overall annual network losses.

[0229] Figure 5 The diagram of the change of DG and DESS charging and discharging scheduling is shown in Figure 2. Figure 5As can be seen, wind power output gradually increases from early morning to midday, reaching a peak before gradually declining. This trend reflects the diurnal variability of wind energy, which typically exhibits higher output during daytime periods with higher wind speeds. Wind and solar power output peaks between 10:00 AM and 7:00 PM. The DESS effectively stores excess power, optimizing wind and solar power utilization and maximizing the DESS's valley-filling function during this period. From 8:00 PM to midnight, wind and solar power output gradually decreases. To meet load demand, the system purchases power from the upstream power grid to ensure grid stability. From midnight to 10:00 AM, as wind and solar power output increases, power purchases from the upstream power grid decrease. Photovoltaic power output also gradually increases. During this period, the system stores excess wind and photovoltaic power in the DESS to further ensure efficient energy utilization.

[0230] like Figure 6 As shown in Figure 3, the fitness value of the particle swarm algorithm changes during the iteration process. As the number of iterations increases, the fitness value gradually decreases, and the upper-level model optimizes the capacity configuration of DG and DESS and gradually finds a better configuration solution during the optimization process.

[0231] By comparing Scheme 1 and Scheme 2 before and after the introduction of DESS double-layer optimization, the voltage levels of each node are compared. Figure 7 As shown in the figure. In Scheme 1, the node voltage level fluctuates greatly, and the voltage of some nodes is low during specific periods, close to 0.95pu. In particular, during peak load periods or when wind and solar output is insufficient, the voltage stability is poor. This shows that the distribution network without an energy storage system is unable to effectively cope with load fluctuations and the uncertainty of renewable energy output, resulting in a decline in voltage quality. In contrast, the node voltage level of Scheme 2 is significantly improved. After the energy storage system is configured, the voltage fluctuation is significantly reduced, and the voltage of most nodes is maintained above 0.96pu, with a more uniform overall distribution. The energy storage system balances the supply and demand relationship of the distribution network through charge and discharge regulation, alleviates the impact of wind and solar output fluctuations on voltage, and significantly improves the voltage stability of the system.

[0232] like Figure 8 Figure 2 shows the changes in the renewable energy absorption rate over a 24-hour period for Scenarios 1 and 2. The renewable energy absorption rate increased from 82.3% in Scenarios 1 to 95.0% in Scenarios 2, a relative increase of 12.7 percentage points. In Scenarios 1, the renewable energy absorption rate fluctuated significantly, particularly during periods of high wind and solar output, when the absorption rate dropped significantly, even reaching 0%. This indicates that without energy storage, the low-voltage distribution network struggles to effectively absorb excess renewable energy output, leading to widespread wind and solar curtailment and low renewable energy utilization.

[0233] In contrast, Option 2 significantly improves the renewable energy absorption rate, maintaining a high level throughout the day, particularly during periods of high wind and solar output, where the absorption rate approaches 100%. By absorbing excess renewable energy output and releasing power when load demand is high, the energy storage system effectively balances the supply and demand of the distribution network and reduces wind and solar curtailment. This not only increases the utilization rate of renewable energy but also enhances the economic efficiency and reliability of the distribution network. By optimizing energy storage configuration and scheduling strategies, the operational performance of the low-voltage distribution network has been significantly improved, providing strong support for the efficient utilization of renewable energy and the stable operation of the distribution network.

[0234] Figure 9 The results compare key power quality indicators for Option 1 (no energy storage) and Option 2 (distributed energy storage configured based on a two-tier optimization strategy), including voltage harmonic distortion (THD), three-phase voltage imbalance, and system frequency deviation. This graph visually demonstrates the improvement in low-voltage distribution network power quality achieved through the coordinated optimization of distributed energy storage systems (DESS) and distributed generation (DG), validating the effectiveness of the proposed approach.

[0235] In terms of voltage harmonic distortion (THD), Figure 9 As shown in sub-figure a, the average THD for Scheme 1 is 4.2%, with a maximum instantaneous value of 5.8%, exceeding the IEC 61000-3-6 standard limit (3% dashed line) for some periods. This is due to power fluctuations caused by DG integration and harmonic injection caused by the nonlinear characteristics of the inverter. Scheme 2, on the other hand, reduces the average THD to 1.8% through the power smoothing effect of the DESS, reducing the fluctuation amplitude by 62% and meeting the standard requirements throughout the entire period. The energy storage system effectively suppresses the fifth and seventh characteristic harmonics generated by the photovoltaic inverter and wind turbine converter by adjusting the charge and discharge power in real time, verifying the optimization effect of the second-order cone programming model on harmonic suppression.

[0236] In terms of three-phase voltage imbalance, Figure 9 As shown in subfigure b, the imbalance in Scheme 1 exhibits significant diurnal fluctuations, reaching a maximum of 4.2% during the day, primarily due to interphase power differences caused by single-phase DG connection. Scheme 2, however, uses an active three-phase power allocation strategy within the energy storage system to control the imbalance within a range of 0.8%-2.1%, a 67% reduction in average value compared to Scheme 1. This is due to the upper-layer particle swarm optimization model's optimal selection of energy storage installation locations (nodes 6 and 13), enabling the DESS to compensate for the phase imbalance caused by DG connection locally, demonstrating the synergistic effect of spatial configuration optimization on improving power quality.

[0237] In terms of system frequency deviation, such as Figure 9As shown in sub-figure c in [1], the frequency fluctuation range of Scheme 1 reaches ±0.15 Hz, with significant peak-to-valley differences, especially during the period of sudden changes in wind and solar output from 10:00 to 14:00, when a transient deviation of 0.18 Hz occurs. Scheme 2, however, compresses the frequency deviation to ±0.03 Hz through the fast frequency response (FRT) function of the energy storage system, an 80% improvement over Scheme 1. This verifies the effectiveness of the SOC dynamic constraint (19) established by the lower-level scheduling model. By adjusting the dP / dt of the charge and discharge power in real time, the DESS mitigates the impact of DG output fluctuations on system inertia.

[0238] This comprehensive approach to the strategy proposed in this invention can significantly improve the reliability and adaptability of the system, enhance the reliability and flexibility of the low-voltage distribution system, ensure the effective use of local power generation resources and energy storage capacity in the low-voltage distribution network, effectively solve the loss problem caused by the access of new energy to the low-voltage distribution network, and ultimately support the realization of economically stable operation.

Claims

1. A two-layer optimization method for distributed energy storage in low-voltage distribution networks taking into account distributed power generation access, characterized in that The following steps are involved: Step 1: Based on the randomness of distributed power output, construct the network equations of distributed power and distributed energy storage; Step 2: Considering the distributed generation access and distributed energy storage configuration strategies, a two-layer optimization model is established; Step 3: Use the improved particle swarm optimization algorithm combined with the second-order cone convex relaxation linear programming technology to solve the two-level optimization model established in step 2.

2. The method for optimizing distributed energy storage in a low-voltage distribution network taking into account distributed power supply access according to claim 1, characterized in that: In the two-layer optimization model, the upper layer optimizes the capacity configuration of distributed power generation and distributed energy storage using an improved particle swarm optimization algorithm, while the lower layer uses CPLEX to accurately solve the daily scheduling plan through second-order cone programming to optimize the charging and discharging strategy of the energy storage unit to meet the changes in power load and fluctuations in grid demand. The results of the lower-level daily scheduling strategy are fed back to the upper-level model to achieve interaction between the upper and lower-level models.

3. The method for optimizing distributed energy storage in a low-voltage distribution network taking into account distributed power supply access according to claim 1, characterized in that: In step 1, the network equations of distributed power sources and distributed energy storage are constructed, specifically including: 1) Photovoltaic power generation model: The active output of photovoltaic power generation is affected by the local light intensity; the light intensity r is assumed to obey the Beta distribution, and its probability density function is: In formula (1), f(pv) is the probability density function of photovoltaic power output power; Γ(α+β) is the gamma function of the sum of shape parameters used to describe the Beta distribution characteristics; Γ(α) is the value of the gamma function at parameter α; Γ(β) is the value of the gamma function at parameter β; I, I pv are the actual value and rated value of light intensity respectively; ɑ and β are the shape parameters of Bate; Γ is the gamma function; Among them, the photovoltaic output power P pv The relationship between the actual value of the light intensity I is: In formula (2): P r is the rated power of the photovoltaic; 2) Wind power generation model: The change in wind turbine output is related to wind speed, which is considered to be closer to the Weibull distribution; its probability density function is: In formula (3), f(v) is the probability density function of wind power output; c is the scale parameter, k is the shape parameter, and v is the average wind speed in the region; The actual output power P of the fan WT The relationship between it and wind speed v is expressed as: In formula (4): P W is the fan output power; P WT is the rated output power of the fan; v r , v in , v co are the wind turbine rated wind speed, cut-in wind speed and cut-out wind speed respectively; 3) Energy storage system model: The change in the state of charge of the energy storage system at different time points is closely related to the state at the current moment, which can be expressed by the following formula: In formula (5): SOC t+1 The state of charge at the next moment; SOC t is the state of charge at the current moment; P charge P is the charging power of the energy storage system at the current moment; discharge is the discharge power of the energy storage system at the current moment, a positive value indicates discharge, and a negative value indicates charging; C is the total capacity of the energy storage system; Δt is the time step; SOC reflects the current charge level of the energy storage system, affecting the battery charging and discharging strategy and available power. In SOC calculations, the difference between charging power and discharging power determines the SOC trend. That is, if the charging power is greater than the discharging power, the SOC increases; otherwise, it decreases. The total capacity C reflects the maximum energy that the energy storage system can store.

4. The method for optimizing distributed energy storage in a low-voltage distribution network taking into account distributed power supply access according to claim 1, characterized in that: In step 2, a two-layer optimization model is established, which specifically includes: A two-layer optimization model considering DG access is established to achieve the optimal configuration of distributed energy storage; The upper model of the two-layer optimization model aims to minimize the comprehensive investment cost of the distributed energy storage system and find the optimal capacity configuration of various energy storage devices; The lower model is used to maximize the reduction of low-voltage distribution network losses under the energy storage capacity constraints determined by the upper model; Subsequently, the optimization results of the lower-level model are fed back to the upper level of the model to guide investment decisions; The difference between the upper and lower objective functions is represented by the parameter TOL, and the iteration will stop when TOL is less than the set value; by continuously iterating the two-layer optimization model, the optimal capacity configuration plan of distributed energy storage can be found.

5. A two-layer optimization method for distributed energy storage in a low-voltage distribution network taking into account access to distributed power sources according to claim 4, characterized in that: Upper model objective function: minC total =C DG +C DESS +C BUY (6); In formula (6): C total is the total investment cost; C DG New energy investment and operating costs; C DESS is the investment cost of distributed energy storage; C BUY The cost of purchasing electricity; ①. New energy investment and operating costs: In formula (7): x i Indicates that DG is installed at node i; P DG，i C is the installed capacity of the DG at node i; in,i is the investment cost per unit capacity of DG installed at node i; ②. Distributed energy storage investment cost: In formula (8): y i To install the node collection; c E ,c P are the operating costs per unit capacity and per unit power of the DESS installed at node i, respectively; E DESS ,P DESS are the rated capacity and rated power of the DESS installed at node i, respectively; ③. Electricity purchase cost: In formula (9): C b The price of electricity purchased from the large power grid; P load is the load active power; P WT 、P PV is the active output power of wind power generation and photovoltaic power generation; P DESS is the active power output of DESS; △t is the length of each time period.

6. A two-tier optimization method for distributed energy storage in a low-voltage distribution network taking into account access to distributed power sources according to claim 5, characterized in that: Upper model constraints: ①.DESS power and capacity constraints: In formula (10): P DESS 、P DESS，min The upper and lower limits of the energy storage rated power; E DESS,max 、E DESS，min The maximum and minimum values of the energy storage rated capacity; n DESS is the maximum installed quantity of energy storage; P DESS is the active power value of DESS; E DESS is the capacity value of DESS; n max The upper limit of the number of DESSs allowed to be deployed in the system; ②. Distributed power output constraints: In formula (11): P wind(t) is the wind power output at time t; P wind,max(t) Indicates the maximum available output of the wind power system at that moment; In formula (12): P solar(t) is the photovoltaic output at time t; P solar,max(t) The maximum output of photovoltaic power generation at that moment; In formula (13): P total,max(t) The maximum output limit of renewable energy is set according to the demand and dispatch capacity of the power grid.

7. A two-tier optimization method for distributed energy storage in a low-voltage distribution network taking into account access to distributed power sources according to claim 6, characterized in that: The objective function of the lower model is: In formula (14): f is the network loss cost; ψ b is the system branch set; I ij,t is the effective value of the current in branch ij at time t; R ij is the resistance value of branch ij.

8. A two-tier optimization method for distributed energy storage in a low-voltage distribution network taking into account access to distributed power sources according to claim 7, characterized in that: Lower model constraints: ①. System power flow balance constraints: In the above formula: P ij,t , Q ij,t are the active power and reactive power from node i to node j at time t; P i,t , Q i,t are the active load and reactive load at node i at time t respectively; ②.Node voltage constraints: In i,max ≤V i,t ≤V i,max (17); In formula (17): V i,max and V i,min are the maximum and minimum voltage amplitudes at node i respectively; V i,t is the voltage of node i at time t; ③.Node current constraints: In formula (18): I ij，t is the effective value of the current flowing through ij at time t; I ij，max is the maximum current amplitude at branch ij; ④. Energy storage charging and discharging and SOC constraints: In formula (19): SOC min , SOC max are the lower and upper limits of the DESS state of charge, respectively; SOC(t) is the state of charge of the DESS at time t; P charge(t) , P discharge(t) are the charging power and discharging power of energy storage at time t respectively; P charge,max , P discharge,max are the maximum charging power and maximum discharging power of distributed energy storage DESS respectively.

9. A two-tier optimization method for distributed energy storage in a low-voltage distribution network taking into account access to distributed power sources according to claim 8, characterized in that: In step 3, the particle swarm algorithm PSO is improved as follows: The improved particle swarm algorithm PSO specifically includes: ①. Dynamic inertia weight: The inertia weight w adopts a linear decreasing strategy, namely: In formula (20): w max 、w min are the maximum and minimum inertia weight values, t n is the current iteration number, T max is the maximum number of iterations; ②. Adaptive learning factor: The individual learning factor c1 and the social learning factor c2 are dynamically adjusted according to the fitness value of the particle, namely: In formula (21): f best 、f worst are the optimal fitness value and the worst fitness value in the particle swarm, respectively, f i is the fitness value of the current particle; c 1,min 、c 1,max are the minimum and maximum values of the individual learning factor c1; c 2,min 、c 2,max are the minimum and maximum values of the social learning factor c2, respectively; ③. Mutation operation: In each iteration, the position of the particle is randomly perturbed with a certain probability, that is: x i =x i +n·randn(0,1) (22); In formula (22): x i is the current position of the i-th particle; η is the disturbance amplitude, and randn(0,1) is a random number from the standard normal distribution.

10. A two-layer optimization method for distributed energy storage in a low-voltage distribution network taking into account access to distributed power sources according to claim 9, characterized in that: Solving the two-level optimization model based on second-order cone programming: By introducing intermediate variables, the key parameters in the two-level optimization model are defined; the replacement formula is: In formula (23): V i is the quadratic relationship of node i voltage after introducing the intermediate variable; l ij is the quadratic relationship of branch ij current after the introduction of intermediate variables; U is the node voltage value; I is the effective value of branch current; In the two-level optimization model, constraint transformation is the key to improving the efficiency of model solution; By replacing variables, the model can be effectively simplified to solve the problem. There are nonlinear constraints in the original two-level optimization model, which are convexly relaxed by formula (23). In formula (24): ij is the quadratic relationship of branch ij current after introducing the intermediate variable; V i is the quadratic relationship of node i voltage after the introduction of intermediate variables; P ij , Q ij are the active power and reactive power of branch ij respectively; The two-level optimization model (Equations (14) to (19)) is transformed into the mixed integer second-order cone programming model (Equations (23) to (25), and the original nonlinear constraints are transformed into a second-order cone form that is easier to solve through relaxation technology; i ij,t ≤i ij,max (26); Where: i ij,t is the current value of branch ij at time t; i ij,max is the maximum current flowing through branch ij.

Images