Two-stage optimization method for distribution network considering electric vehicles and distributed generation integration

Abstract

This invention proposes a two-stage optimization method for distribution networks integrating electric vehicles (EVs) and distributed generation. In the first stage, an orderly scheduling of EVs is achieved by establishing a user participation dispatch probability model and a dynamic electricity price model considering load. The optimized EV load is then superimposed onto the distribution network reconfiguration optimization model in the second stage. Finally, network reconfiguration and distributed generation output are used to reduce network losses and node voltage deviations across the entire distribution network. The proposed method reduces peak-to-valley load differences. Considering user participation rates, it not only reduces user charging costs but also improves node voltage levels and reduces network losses.

Description

Technical Field

[0001] This invention relates to the field of smart grid and electric vehicle (EV) collaborative optimization technology, specifically to a two-stage optimization method for distribution networks that considers the access of electric vehicles and distributed power sources. Background Technology

[0002] With the energy structure transforming towards low-carbon and intelligent operation, the large-scale integration of distributed generation (DG) and electric vehicles (EVs) into the distribution network has become an inevitable trend. However, the penetration of high proportions of renewable energy and dynamic loads poses serious challenges to the planning and operation of traditional distribution networks. Existing technologies have significant shortcomings in multi-timescale coordination, uncertainty management, and multi-system integrated optimization, necessitating a new optimization framework to address the following technical bottlenecks.

[0003] First, the superposition of electric vehicle (EV) charging load and distribution network base load leads to two major problems: one is the exacerbation of the peak-valley load difference. Large-scale, disorderly charging of EVs superimposed on the base load will cause peak-on-peak phenomena, resulting in a larger peak-valley difference. Studies show that when the EV penetration rate is 30%, the peak load can increase by 25%. Second, there is insufficient consideration of the stochasticity of EV charging. Previous studies have treated the grid connection and disconnection time and initial state of charge of EVs as stochastic and assumed that all EVs participate in optimal scheduling, without modeling whether users choose to participate in optimal scheduling. Third, most distribution network optimization for EV access is achieved by optimizing energy storage configuration, reactive power compensation, and distributed generation output to achieve the optimal state of the distribution network. However, in addition to these methods, distribution network reconfiguration has a more significant effect on reducing network losses and improving the voltage level of each node.

[0004] Currently, for the optimized scheduling of electric vehicles (EVs), the industry mainly guides EV charging behavior through time-of-use (TOU) pricing and demand response (DR), such as smoothing grid fluctuations through price signals. However, these methods have significant limitations. For demand response, the problem is insufficient user participation; there is no effective strategy to incentivize user participation. As for TOU pricing, traditional methods divide it into fixed time periods, such as peak, off-peak, and valley periods, which cannot adapt to the fluctuations in distribution network load, resulting in significant peak-to-valley differences. Therefore, a pricing strategy correlated with load fluctuations is needed to reduce peak-to-valley differences to a greater extent, while simultaneously lowering charging costs for users.

[0005] Currently, most optimization methods consider comprehensive optimization of various systems, such as the integrated optimization of electric vehicle (EV) scheduling, energy storage output, and distributed power output. The drawbacks are that the comprehensive optimization model is complex, the solution speed is slow, and it lacks specificity. Especially when considering EV optimization, comprehensive optimization may result in EV optimization failing to meet user needs, as EV optimization must take into account user scheduling preferences. Therefore, performing targeted optimization of EVs first, followed by distribution network optimization, would be a more reasonable approach.

[0006] Existing technologies still have shortcomings in terms of optimization methods for electric vehicles (EVs) accessing the power distribution network, and the strategies for optimizing the scheduling of EVs need further improvement. Furthermore, the handling of uncertainties in EVs can be further refined. Summary of the Invention

[0007] To address the aforementioned issues of increased peak-to-valley load differences and excessive node voltage deviations caused by the integration of electric vehicles (EVs) into the distribution network, this invention fully considers the uncertainties of EVs and proposes a two-stage optimization method for the distribution network that takes into account both EVs and distributed generation. In the first stage, this method first addresses uncertainties by establishing a probabilistic model of EV grid entry / exit times and EV participation rates. Then, it establishes a dynamic electricity price model that considers load to incentivize EV charging and discharging, reducing peak-to-valley load differences and user charging costs. Next, the optimized EV load is superimposed onto the second-stage distribution network reconfiguration optimization model. By adjusting the output of distributed generation and the switching states of sectionalizing and tie switches, network losses and node voltage deviations are reduced, ultimately achieving a win-win situation for both the power grid and users.

[0008] The technical solution adopted in this invention is as follows:

[0009] A two-stage optimization method for distribution networks considering the integration of electric vehicles and distributed power sources is proposed. In the first stage, orderly scheduling of electric vehicles is achieved by establishing a user participation scheduling probability model and a dynamic electricity price model that considers load. The optimized electric vehicle load is then superimposed on the distribution network reconfiguration optimization model in the second stage. Finally, network reconfiguration and distributed power output are used to reduce network losses and node voltage deviations in the entire distribution network.

[0010] A two-stage optimization method for distribution networks considering the integration of electric vehicles and distributed power sources includes the following steps:

[0011] Step 1: Considering the uncertainties of electric vehicles, establish a probability model for user participation in scheduling;

[0012] Step 2: Establish a dynamic electricity pricing model that takes load into account to incentivize electric vehicle charging and discharging, thereby reducing the peak-valley load difference and user charging costs;

[0013] Step 3: Construct an electric vehicle optimal scheduling model. Based on the uncertainty data from Step 1 and the dynamic electricity price obtained in Step 2, obtain the final electric vehicle scheduling scheme.

[0014] Step 4: Establish a distribution network reconfiguration optimization model. Overlay the optimized load data of electric vehicles obtained in Step 3 onto the distribution network for reconfiguration optimization to reduce network losses and node voltage deviations in the distribution network.

[0015] Step 5: Compare the load peak-to-valley difference, network loss, and node voltage deviation before and after electric vehicle optimization and distribution network reconfiguration optimization; and compare the optimization results under each scenario to verify the effectiveness of this optimization method.

[0016] Phase 1:

[0017] In step 1, a user participation scheduling probability model is established, and the participation of electric vehicles in scheduling, grid connection / disconnection time, initial SOC, and distributed power output are generated using the Monte Carlo method. Specifically, this includes:

[0018] 1) The time for electric vehicles to enter the grid is described using a normal distribution.

[0019]

[0020] In equation (1): t i Let μ be the network access time of the i-th electric vehicle. s and δ s These are the mean and variance of the network access time, respectively, which are taken as 17.6 and 3.4 here.

[0021] 2) The initial SOC adopts a log-normal distribution probability model, specifically including:

[0022] Daily mileage can be described using a log-normal distribution:

[0023]

[0024] In equation (2): s i Let μ be the daily mileage of the i-th electric vehicle. D ,δ D These are the mean and variance, which are taken as 3.2 and 0.88 respectively.

[0025] The initial state of charge (SOC) and charging duration of an electric vehicle are obtained based on the daily mileage.

[0026]

[0027] t d =t i +T i (5);

[0028] In the above formula: S initial,i Let S be the initial charge of the i-th electric vehicle. max For the maximum driving range, T i Let E be the charging time of the i-th electric vehicle, E be the battery capacity, η be the charging / discharging efficiency, and P be the charging time of the i-th electric vehicle. i Let t be the charging and discharging power of the i-th electric vehicle. d This refers to the time spent offline.

[0029] 3) Photovoltaic power output is described using the beta probability density function:

[0030]

[0031] P pv =Arη pv (8);

[0032] In the above formula: H is the per-unit value of irradiance, r is the actual irradiance, and r max For maximum irradiance, ω and θ are the photovoltaic shape and size parameters, respectively, A is the area of ​​the photovoltaic panel, and η is the maximum irradiance. pv Let Γ be the photovoltaic conversion efficiency, and Γ be the Gamma function.

[0033] 4) Wind power is described using the Weibull probability density function:

[0034]

[0035] In the above formula: v is the wind speed, c is the wind power scale parameter, and v o To cut off the wind speed; v i For the cut-in wind speed; v r Rated wind speed; P w For wind power, P r f(v) represents the rated power of the wind turbine; f(v) represents the wind speed probability density function. The attenuation factor determines the attenuation rate of the probability density function; k1k2 are parameters related to the rated power of the wind turbine, the cut-in wind speed, and the rated wind speed, used to describe the relationship between wind power and wind speed.

[0036] In step 1, a user participation scheduling probability model is established, and its probability density function is shown below:

[0037]

[0038] In equation (12): P s,i Let T be the probability of the i-th electric vehicle participating. arrive,i ,T depart,i S represents the grid entry and exit times of the i-th electric vehicle, respectively. initial,iLet be the initial SOC of the i-th electric vehicle, and β1 and β2 be the time sensitivity coefficient and the SOC sensitivity coefficient, respectively. This reflects factors that influence the probability of participation, which are related to the user's charging duration and initial SOC.

[0039] Equation (12) shows that the higher the initial SOC, the greater the probability of participation; the longer the charging time, the greater the probability of participation. Bernoulli random numbers u are generated using the probability density function and the Monte Carlo method. i,t ;

[0040]

[0041] In equation (13): 1 represents electric vehicles participating in optimized scheduling; 0 represents electric vehicles not participating in optimized scheduling and choosing disordered charging mode.

[0042] In step 2, the dynamic electricity price model considering the load includes a functional relationship between the charging and discharging electricity price of electric vehicles and the total load, so that the electricity price is lower when the total load is at its peak and higher when the load is at its trough, thereby incentivizing users to choose charging and discharging times.

[0043] The dynamic electricity price is based on a certain correlation with load fluctuations, and its calculation formula is as follows:

[0044]

[0045] In the above formula: C t For dynamic electricity pricing; P t P represents the total load at time t; i,t ,P base,t The net load and base load of the electric vehicle at time t are respectively; P average The average load of the entire distribution network is denoted as α; α and δ are the electricity price correlation coefficient and reference electricity price, respectively; and T is the dispatching cycle.

[0046] The formula above describes the relationship between the electricity price for charging and discharging electric vehicles and the total load including the electric vehicle load. This allows for lower electricity prices during peak load periods to incentivize charging, and higher prices during off-peak periods to incentivize discharging, thereby generating revenue and reducing costs. This charging price is then fed back to the user through the charging station, informing them of the optimal charging times.

[0047] In step 3, an electric vehicle (EV) optimal scheduling model is established, including the objective function and constraints. Using the data from step 1 and the dynamic electricity price obtained in step 2, a solver is used to obtain the final EV scheduling scheme. Specifically, the EV optimal scheduling model first collects the SOC, grid connection time, grid disconnection time, battery capacity, and decision-making information of EVs connected to charging piles. These will serve as parameters for the subsequent model, including the objective function and constraints.

[0048] The first stage is to achieve optimal scheduling of electric vehicles through dynamic electricity pricing. Therefore, the objective function is to minimize the charging and discharging costs for users. The decision variables are the charging and discharging power of each electric vehicle at each time, and a load fluctuation penalty term is added.

[0049] The objective function is as follows:

[0050]

[0051] In equation (5): M is the objective function, T is the scheduling period, N is the total number of electric vehicles, and C t For dynamic time-of-use pricing, P i,t Let P be the electric vehicle charging / discharging power of the i-th electric vehicle at time t, where a value greater than zero indicates charging and a value less than zero indicates discharging; μ is the penalty coefficient, and P is the electric vehicle charging / discharging power. base,t To exclude the basic electrical load at time t during the charging and discharging of electric vehicles;

[0052] In the objective function, The total charging cost for the user, This is a penalty measure, intended to mitigate load fluctuations.

[0053] The constraints include charging and discharging power constraints, battery SOC constraints, total power limit constraints of charging piles, and user expectation constraints, as detailed below:

[0054] ①. Electric vehicle charging and discharging power constraints:

[0055]

[0056] In equation (6): P max,discharge P is the maximum discharge power; max,charge The maximum charging power is set; if the user chooses to participate in the scheduling, the charging and discharging power is subject to upper and lower limits; if the user does not participate in the scheduling, the charging power is set to the maximum power, and no discharging is performed.

[0057] ②. Battery SOC constraint:

[0058]

[0059] In the above formula: SOC i,tLet SOC be the state of charge of the i-th electric vehicle at time t. i,t-1 Let be the state of charge of the i-th electric vehicle at time t-1. and Let be the charging power and discharging power of the i-th electric vehicle, respectively; η be the charging / discharging efficiency; Δt be the time interval; E be the charging / discharging power. i Let be the battery capacity of the i-th electric vehicle; SOC max and SOC min These represent the maximum and minimum states of charge of an electric vehicle, respectively.

[0060] ③. Total power limit constraint for charging piles:

[0061]

[0062] In equation (20): P grid_max This represents the maximum allowable charging power of the charging station.

[0063] ④. User expectation constraint: To ensure that electric vehicles can meet their travel requirements when off-grid, a minimum battery level requirement is set for off-grid operation.

[0064] SOC i,depart ≥SOC target (twenty one);

[0065] In equation (21): SOC i,depart Let SOC be the state of charge of the i-th electric vehicle when it leaves the grid. target This represents the lowest off-grid state of charge.

[0066] ⑤. Electricity price restrictions: Electricity price range restrictions are set to prevent electricity prices from becoming too high or too low due to load fluctuations;

[0067] C min ≤C t ≤C max (twenty two);

[0068] In equation (22): C min C max These are the upper and lower limits for electricity prices, respectively.

[0069] In summary, the electric vehicle optimal scheduling model based on dynamic electricity pricing is as follows:

[0070] Objective: Minimize M.

[0071] St: Equations (17) to (22)

[0072] Where: Objective is the objective function, and Minimize M is the minimum value of the objective function.

[0073] Phase Two:

[0074] In step 4, the optimized load data of electric vehicles obtained in step 3 is superimposed onto the distribution network for distribution network reconfiguration optimization. The objective function is network loss and voltage deviation, and the decision variable is the state of the switches; specifically as follows:

[0075] Distribution network optimization primarily achieves its objectives by optimizing the status of sectionalizing switches and tie switches, as well as the output of distributed generation sources. The objective function is network loss, with voltage deviation added as a penalty term.

[0076] The objective function is as follows:

[0077]

[0078] In equation (23): P loss For network loss; x k,t Let I represent the switching state of branch k at time t. It is a binary variable; a value of 1 indicates the branch is closed, and a value of 0 indicates the branch is open. k,t R is the current in branch k at time t; k Let V be the resistance of branch k; λ be the penalty coefficient; V i,t V is the voltage at node i at time t; 0 Reference voltage; N l N b These represent the number of branches and nodes, respectively; T represents the number of time periods.

[0079] The constraints include node voltage constraints, branch current constraints, power balance constraints, radial topology constraints, and distributed generation output constraints.

[0080] a: Power balance constraint:

[0081]

[0082] In equation (24): P mi,t Q mi,t Let P be the active and reactive power of the branch with node i as the last node at time t; in,t Q in,t R represents the active and reactive power of the branch with node i as the first node at time t; mi ,X mi Let I be the branch resistance and reactance with node i as the last node; mi,t Let be the branch current at time t; Q represents the total active power at the node, including the load from electric vehicles; i,t For node reactive load; Let m, i, and n be the active power output of node i at time t; m, i, and n are the node numbers.

[0083] b: Voltage constraint:

[0084]

[0085] In equation (25): U i,t U j,t Let X be the voltages at nodes i and j, respectively; k is the branch; X is the voltage at node i and node j, respectively. k Let k be the reactance of branch k; k(i,j) represents the two nodes i and j connected by branch k.

[0086] c: Current constraint:

[0087]

[0088] d: Topological constraints:

[0089]

[0090] In equation (27): x k,t N represents the switching state of branch k at time t; b N is the number of nodes. L This represents the total number of branch roads.

[0091] e: Distributed power generation output constraints:

[0092]

[0093] In equation (28): These represent the maximum and minimum active power outputs of DG, respectively.

[0094] By optimizing the objective function model of the distribution network shown in equation (23) and the constraint model shown in equations (24)-(28), and after superimposing the electric vehicle load, the solution is obtained to find the branch switch states x that minimize network loss and voltage deviation. k,t The specific value is obtained by solving the switch state x. k,t The final reconfigured distribution network topology can be obtained.

[0095] This invention presents a two-stage optimization method for distribution networks considering the integration of electric vehicles and distributed power sources. The technical effects are as follows: 1) This invention enables the power grid to achieve better peak shaving and valley filling. By establishing a load-related electricity price model, a highly sensitive response between electricity price and total load is established, allowing electric vehicles (EVs) to actively track the total load curve during charging and discharging. The electricity price model incentivizes EVs to discharge during peak load periods and charge during off-peak periods, thus smoothing load fluctuations, achieving peak shaving and valley filling, and simultaneously reducing charging costs for users.

[0096] 2) This invention considers the probability of users participating in electric vehicle scheduling, that is, not all users participate in the optimal scheduling of electric vehicles, which reflects the randomness of user charging, making the model more realistic and more valuable for practical application.

[0097] 3) This invention adopts a two-stage optimization method. After optimizing the electric vehicle in the first stage, the power distribution network is restructured and optimized. Under the premise of minimizing user charging costs, the entire power distribution network can maintain a stable state, thereby reducing power distribution network losses and improving node voltage levels.

[0098] 4) This invention considers the optimized scheduling of electric vehicles in V2G mode, which can not only serve as a load but also as an energy storage unit, thus realizing the rational utilization of resources. Attached Figure Description

[0099] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0100] Figure 1 This is a flowchart of the two-stage optimization method of the present invention.

[0101] Figure 2 This is a comparison chart of EV charging and discharging power before and after optimization.

[0102] Figure 3 It is an optimization of the load fluctuation diagram before and after.

[0103] Figure 4 It is a dynamic electricity price curve.

[0104] Figure 5 This is a comparison diagram of voltage at distribution network nodes.

[0105] Figure 6 This is a comparison chart of voltage at each node at each time point. Detailed Implementation

[0106] A two-stage optimization method for distribution networks considering the integration of electric vehicles (EVs) and distributed generation (DG) power sources is proposed. This method introduces a user-participatory dispatch probabilistic model. In the first stage, an EV optimization dispatch model considering user participation rates is established to achieve orderly charging and discharging of EVs. In the second stage, a distribution network reconfiguration optimization model is established, effectively addressing the problems of excessive load fluctuations and increased distribution network losses caused by the large-scale integration of EVs. Specifically, in the first stage, orderly dispatch of EVs is achieved by establishing a user-participatory dispatch probabilistic model and a dynamic electricity price model considering load. The optimized EV load is then superimposed on the distribution network reconfiguration optimization model in the second stage. Finally, network reconfiguration and DG power output are used to reduce network losses and node voltage deviations in the entire distribution network.

[0107] Numerical examples demonstrate that the method of this invention can achieve three core benefits within a 24-hour optimization cycle: 1) significantly reduced network losses in the distribution network and improved node voltage levels; 2) significantly reduced user charging costs; and 3) reduced load peak-to-valley difference, effectively realizing the "peak shaving and valley filling" effect of the distribution network load. This method provides an innovative solution for the coordinated control of electric vehicles and the distribution network in new power systems, and has significant engineering application value.

[0108] In practical applications, DER data can be acquired by installing intelligent acquisition modules at the generator, while data for electric vehicles is acquired by charging piles. The grid connection and disconnection times of electric vehicles are chosen by the users at the charging piles. After uploading this data to the dispatch center and optimizing the solution, a dispatch scheme is obtained. In this simulation example, to simulate the uncertainties of distributed power generation and electric vehicle travel, the Monte Carlo method is used to randomly generate 24 hours of DER output and the initial SOC of electric vehicles, as well as grid connection and disconnection times, using a probability density function.

[0109] The two-stage optimization method for distribution networks considering the integration of electric vehicles and distributed power sources mainly includes the following steps:

[0110] Phase 1:

[0111] 1) Electric vehicle parameters: The optimization scheduling cycle is 24 hours, with 1 hour as a time interval, and 200 electric vehicles are optimized and scheduled. The data of the electric vehicles are shown in Table 1 below.

[0112] Table 1 Relevant parameters of electric vehicles

[0113] Charge / discharge efficiency η Maximum power of charging pile Maximum charging power Maximum discharge power Expected SOC 0.95 1500KW 7KW 5KW 0.8

[0114] 2) Model parameter settings are shown in Table 2:

[0115] Table 2 Optimization Model Parameters

[0116]

[0117] User participation is determined using the Monte Carlo method based on the participation rate probability density function. 168 vehicles are selected to participate, while 32 vehicles do not participate, resulting in unordered scheduling.

[0118] 3) Using the parameters of the given model, the selected electric vehicles are optimally scheduled using a dynamic electricity price model. The charging and discharging loads of the electric vehicles before and after optimization are as follows: Figure 2 As shown, its dynamic electricity price curve is as follows: Figure 4 As shown in the figure, its fluctuations are roughly consistent with the load curve. The optimization results are shown in Table 3. It can be seen from the table that after optimizing the scheduling of electric vehicles, user charging costs will be significantly reduced, the load peak-valley difference will be narrowed, and the stable operation of the distribution network will be promoted.

[0119] Table 3 Comparison of optimization results before and after.

[0120] Cost (RMB) Peak-to-valley difference (kW) Average charging power (kW) Before optimization 1713.6 2674 119 After optimization 739.8 1774.95 83.21

[0121] Electric vehicle charging and discharging power before and after optimization, such as Figure 2 As shown, by Figure 2 Based on the electricity price curve, it can be seen that after optimization, electric vehicle charging is concentrated in the 0-7 period when the electricity price is lower, while electric vehicle discharging is concentrated in the 16-20 period when the electricity price is higher. This achieves the goal of electric vehicles discharging during peak load and charging during off-peak load, thereby reducing the load peak-valley difference.

[0122] Load curve as Figure 3 As shown, by Figure 3 It is known that when electric vehicles are charged haphazardly, peak loads are amplified, increasing load fluctuations. Optimized scheduling reduces peak loads and increases valley loads, achieving load shifting. Therefore, establishing a dynamic electricity price mechanism related to the total load for optimized scheduling of electric vehicles can effectively reduce the peak-valley load difference. The electricity price curve is shown below. Figure 4 As shown, by Figure 4 It can be seen that the dynamic electricity price curve can effectively track the fluctuation trend of the load curve, thus enabling better and more precise scheduling of electric vehicles.

[0123] Phase Two:

[0124] 2) Distribution Network Optimization. This simulation uses the IEEE 33 distribution network, with distributed photovoltaic access nodes 33, 10, and 2, and electric vehicle charging station access nodes 18 and 31. The net load of electric vehicles obtained in the first stage is superimposed onto the distribution network, and then the distribution network is reconfigured and optimized. The optimization results are shown in Table 4 below. As can be seen from Table 4, the network loss is effectively reduced after the distribution network reconfiguration.

[0125] Table 4 Network Losses Before and After Distribution Network Restructuring

[0126] Network loss KW Before Refactoring 6733.91 After refactoring 4231.84

[0127] For example, the voltage of the nodes in the distribution network at time 18. Figure 5 As shown, the branch that is disconnected at that moment is:

[0128] (7,8),(9,10),(14,15)(26,27),(18,33). From Figure 5 It can be seen that the node voltage level has increased significantly after the distribution network reconfiguration. Furthermore, the orderly scheduling of electric vehicles will further improve the node voltage level. Therefore, combining the orderly scheduling of electric vehicles with the distribution network reconfiguration can greatly improve the node voltage level and reduce network losses.

[0129] 5) Result Evaluation. Analyze the optimized load fluctuation curve, user charging costs, and node voltage deviation, and compare them with those before optimization to determine whether the expected levels have been achieved.

[0130] 6) Conclusion. The results of the two-stage optimization method for distribution networks considering electric vehicles and distributed power source integration show that the method proposed in this invention can reduce the load peak-valley difference. While considering user participation rate, it not only ensures a reduction in user charging costs but also improves node voltage levels and reduces network losses.

Claims

1. A two-stage optimization method for distribution networks considering the integration of electric vehicles and distributed power sources, characterized in that: Includes the following steps: Step 1: Considering the uncertainties of electric vehicles, establish a probability model for user participation in scheduling; Step 2: Establish a dynamic electricity pricing model that takes load into account to incentivize electric vehicle charging and discharging, thereby reducing the peak-valley load difference and user charging costs; Step 3: Construct an electric vehicle optimal scheduling model. Based on the uncertainty data from Step 1 and the dynamic electricity price obtained in Step 2, obtain the final electric vehicle scheduling scheme. Step 4: Establish a distribution network reconfiguration optimization model. Overlay the optimized load data of electric vehicles obtained in Step 3 onto the distribution network for reconfiguration optimization to reduce network losses and node voltage deviations in the distribution network. In step 1, a user participation scheduling probability model is established, and its probability density function is shown below: （12）； In equation (12): Let be the probability of the i-th electric vehicle participating. These represent the network entry and exit times of the i-th electric vehicle, respectively. Let i be the initial SOC of the i-th electric vehicle. These are the time sensitivity coefficient and the SOC sensitivity coefficient, respectively. This reflects factors that influence the probability of participation, which are related to the user's charging duration and initial SOC. It can be seen from equation (12) that the larger the initial SOC, the greater the probability of participation; the longer the charging time, the greater the probability of participation. Bernoulli random numbers are generated using the probability density function and the Monte Carlo method. ; （13）； In equation (13): 1 represents electric vehicles participating in optimized scheduling; 0 represents electric vehicles not participating in optimized scheduling and choosing disordered charging mode.

2. The two-stage optimization method for distribution networks considering the access of electric vehicles and distributed power sources according to claim 1, characterized in that: It also includes step 5: comparing the load peak-to-valley difference, network loss, and node voltage deviation before and after electric vehicle optimization and distribution network reconfiguration optimization; and comparing the optimization results under each scenario to verify them.

3. The two-stage optimization method for distribution networks considering the access of electric vehicles and distributed power sources according to claim 1, characterized in that: In step 1, a user participation scheduling probability model is established. The participation of electric vehicles in scheduling, grid connection / disconnection times, initial SOC, and distributed power output are generated using the Monte Carlo method. Specifically, this includes: 1) The time for electric vehicles to enter the grid is described using a normal distribution. （1）； In formula (1): Let be the time when the i-th electric vehicle enters the network. These are the mean and variance of the network access time, respectively; 2) The initial SOC adopts a log-normal distribution probability model, specifically including: Daily mileage can be described using a log-normal distribution: （2）； In formula (2): Let be the daily mileage of the i-th electric vehicle. These are the mean and variance, respectively. The initial SOC and charging duration of the electric vehicle are obtained based on the daily mileage. （3）； （4）； （5）； In the above formula: Let be the initial charge of the i-th electric vehicle. The maximum driving mileage, Let E be the charging time for the i-th electric vehicle, and E be the battery capacity. For charging and discharging efficiency, Let the charging and discharging power of the i-th electric vehicle be... Offline time; 3) Photovoltaic power output is described using the beta probability density function: （6）； （7）； （8）； In the above formula: H is the per-unit value of irradiance, and r is the actual irradiance. For maximum irradiance, These represent the photovoltaic shape parameter and the dimensional parameter, respectively, where A is the area of ​​the photovoltaic panel. For photovoltaic conversion efficiency, It is the Gamma function; 4) Wind power is described using the Weibull probability density function: （9）； （10）； （11）； In the above formula: v is the wind speed, and c is the wind power scale parameter. To cut off the wind speed; To cut in wind speed; Rated wind speed; For wind power, This refers to the rated power of the fan. Let be the wind speed probability density function; This is the decay factor part, which determines the decay rate of the probability density function; These are parameters related to the rated power, cut-in wind speed, and rated wind speed of the wind turbine, used to describe the relationship between wind power and wind speed.

4. The two-stage optimization method for distribution networks considering the access of electric vehicles and distributed power sources according to claim 1, characterized in that: In step 2, the dynamic electricity pricing model considering load includes a functional relationship between the charging and discharging electricity price of electric vehicles and the total load, so that the electricity price is lower during peak load periods and higher during off-peak periods; specifically as follows: The establishment of dynamic electricity pricing is related to load fluctuations, and its calculation formula is as follows: （14）； （15）； In the above formula: Dynamic electricity pricing; The total load at time t; These represent the net load and base load of the electric vehicle at time t, respectively. The average load of the entire distribution network; These are the electricity price correlation coefficient and the reference electricity price, respectively. The scheduling period; The above formula describes the relationship between the electricity price for charging and discharging electric vehicles and the total load that includes the electric vehicle load. This allows for a lower electricity price to incentivize electric vehicle charging during peak load periods and a higher electricity price to incentivize electric vehicle discharging during off-peak load periods, thereby generating revenue and reducing costs.

5. The two-stage optimization method for distribution networks considering the access of electric vehicles and distributed power sources according to claim 1, characterized in that: In step 3, an electric vehicle optimization scheduling model is established. The electric vehicle optimization scheduling model first collects the SOC, on-grid time, off-grid time, battery capacity, and decision-making of electric vehicles connected to the charging pile. The first stage is to achieve optimal scheduling of electric vehicles through dynamic electricity pricing. Therefore, the objective function is to minimize the charging and discharging costs for users. The decision variables are the charging and discharging power of each electric vehicle at each time, and a load fluctuation penalty term is added. The objective function is as follows: （16）； In formula (5): Let T be the objective function, T be the scheduling period, and N be the total number of electric vehicles. For dynamic time-of-use electricity pricing, Let be the charging / discharging power of the i-th electric vehicle at time t, where a value greater than zero indicates charging and a value less than zero indicates discharging; μ is the penalty coefficient. To exclude the basic electrical load at time t during the charging and discharging of electric vehicles; In the objective function, The total charging cost for the user, This is a penalty measure, intended to mitigate load fluctuations.

6. The two-stage optimization method for distribution networks considering the access of electric vehicles and distributed power sources according to claim 5, characterized in that: The constraints include charge / discharge power constraints, battery SOC constraints, total power limit constraints for charging stations, and user expectation constraints; the details are as follows: ①. Electric vehicle charging and discharging power constraints: （17）； In formula (6): This represents the maximum discharge power. The maximum charging power is set; if the user chooses to participate in the scheduling, the charging and discharging power is subject to upper and lower limits; if the user does not participate in the scheduling, the charging power is set to the maximum power, and no discharging is performed. ②. Battery SOC constraint: （18）； （19）； In the above formula: Let t represent the state of charge of the i-th electric vehicle at time t. Let be the state of charge of the i-th electric vehicle at time t-1. and These are the charging power and discharging power of the i-th electric vehicle, respectively; For charge and discharge efficiency; For time intervals; Let be the battery capacity of the i-th electric vehicle; and These represent the maximum and minimum states of charge of an electric vehicle, respectively. ③. Total power limit constraint for charging piles: （20）； In equation (20): This refers to the maximum allowable charging power of the charging station. ④. User expectation constraints: To ensure that electric vehicles can meet their travel requirements when off-grid, a minimum battery level requirement is set for off-grid operation; （21）； In equation (21): Let represent the state of charge of the i-th electric vehicle when it leaves the grid. This represents the lowest off-grid state of charge. ⑤. Electricity price restrictions: Electricity price range restrictions are set to prevent electricity prices from becoming too high or too low due to load fluctuations; （22）； In equation (22): These are the upper and lower limits for electricity prices, respectively. In summary, the electric vehicle optimal scheduling model based on dynamic electricity pricing is as follows: Objective :Minimize M . St Equations (17) to (22) in: Objective Let M be the objective function, and Minimize M is the method to find the minimum value of the objective function.

7. The two-stage optimization method for distribution networks considering the access of electric vehicles and distributed power sources according to claim 6, characterized in that: Phase Two: Step 4 involves overlaying the optimized load data of electric vehicles obtained in Step 3 onto the distribution network for distribution network reconfiguration optimization; the objective functions are network loss and voltage deviation, and the decision variables are the switch states; details are as follows: Distribution network optimization achieves its objectives by optimizing the status of sectionalizing switches and tie switches, as well as the output of distributed generation sources. The objective function is network loss, with voltage deviation added as a penalty term to the objective function. The objective function is as follows: （23）； In equation (23): For network losses; Let t be the switching state of branch k at time t. It is a binary variable. If it is 1, it means the branch is closed. If it is 0, it means the branch is open. Let be the current in branch k at time t; Let K be the resistance of branch k. This is the penalty coefficient; Let be the voltage at node i at time t; Reference voltage; These represent the number of branches and nodes, respectively; T represents the number of time periods.

8. The two-stage optimization method for distribution networks considering the access of electric vehicles and distributed power sources according to claim 7, characterized in that: The constraints include node voltage constraints, branch current constraints, power balance constraints, radial topology constraints, and distributed generation output constraints. a: Power balance constraint: （24）； In equation (24): These are the active and reactive power of the branch with node i as the last node at time t, respectively. These represent the active and reactive power of the branch with node i as the first node at time t; These are the branch resistance and reactance with i as the last node, respectively; Let be the branch current at time t; The total active power of the node that incorporates the electric vehicle load; For node reactive load; Let i be the active power output of the DG at time t; Number the nodes; b: Voltage constraint: （25）； In equation (25): , i and j are the voltages of node i and node j, respectively; k is the branch; The reactance of branch k; This represents the two nodes i and j connected by branch k; c: Current constraint: （26）； d: Topological constraints: （27）； In equation (27): The switching state of branch k at time t; The number of nodes; The total number of branch roads; e: Distributed power generation output constraints: （28）； In equation (28): These are the maximum and minimum active power outputs of DG, respectively; By optimizing the objective function model of the distribution network shown in Equation (23) and the constraint condition model shown in Equations (24) to (28), and after superimposing the electric vehicle load, the solution is obtained to find the switch states of each branch that minimize network loss and voltage deviation. The specific value is determined by the switch state after solving. It is possible to obtain the final reconfigured distribution network topology.

Images