A multi-objective partition coordination scheduling method for mobile energy storage vehicles for rural distribution networks

Abstract

The application discloses an interconnected energy storage vehicle optimization scheduling method for rural power distribution network power supply risk control, which comprises the following steps: collecting rural power distribution network node parameters, branch parameters, real-time load data and distributed new energy output data, and analyzing the uncertainty and time sequence correlation of the distributed new energy output; dividing the rural power distribution network into a plurality of scheduling partitions and configuring mobile energy storage vehicles; constructing a multi-objective distributed robust optimization model, establishing a probability distribution fuzzy set constraint considering the uncertainty and time sequence correlation of the distributed new energy output based on JS divergence; performing convex optimization processing on the non-convex constraint in the multi-objective distributed robust optimization model, and iteratively solving by adopting a CCG algorithm to obtain the corresponding resident area position, charging power, discharging power and driving path of each mobile energy storage vehicle in each scheduling period. The method can improve the power supply guarantee capacity of key areas during the busy season and the centralized scheduling efficiency during the slack season while taking into account the network loss, new energy curtailment rate and operation cost.

Description

Technical Field

[0001] This invention relates to the field of rural power distribution network dispatching technology, specifically to a multi-objective regional collaborative dispatching method for mobile energy storage vehicles for rural power distribution networks. Background Technology

[0002] As the proportion of distributed renewable energy sources such as household photovoltaic systems and small-scale wind power in rural power distribution networks continues to increase, rural power grids are gradually showing a trend of increased fluctuations on both the source and load sides and more complex operation modes. On the one hand, rural power distribution networks are generally characterized by dispersed distribution areas, long power supply radii, and uneven load distribution; on the other hand, agricultural irrigation and agricultural product processing loads have obvious seasonality and time-specificity, making rural power distribution networks prone to problems such as local load concentration, increased node voltage fluctuations, and difficulties in local consumption of renewable energy during the busy farming season.

[0003] In existing technologies, energy storage systems are widely used in distribution networks for peak shaving and valley filling, reducing network losses, improving voltage quality, and promoting the consumption of new energy sources. Compared to stationary energy storage power stations, mobile energy storage vehicles combine energy storage and mobility characteristics, allowing for flexible transfer between different nodes and responding to the power supply and regulation needs of different areas at different times. Therefore, they are more suitable for distribution networks with dispersed distribution areas and varied operating scenarios. Existing research has explored the participation of mobile energy storage vehicles in the optimal scheduling of distribution networks. This research primarily focuses on minimizing distribution network losses, wind and solar curtailment, and operating costs, while also considering the uncertainty and time-series correlation of new energy output to establish corresponding multi-objective distributed bar optimization scheduling models.

[0004] However, existing research on the optimization and scheduling of mobile energy storage vehicles is mostly geared towards general new power distribution systems, focusing on optimizing the operation paths and charging / discharging plans of these vehicles. It fails to adequately consider factors such as village distribution, field substation division, road accessibility, differences in agricultural load timing, and the different scheduling priorities during busy and off-seasons in rural power distribution scenarios. Particularly in rural power distribution network applications, if a unified scheduling approach for ordinary power distribution systems is still adopted, problems such as untimely regional response, insufficient local support capacity, unreasonable mobile paths, and high overall operating costs can easily arise, making it difficult to meet the actual needs of efficient renewable energy consumption and reliable power supply in rural areas. Summary of the Invention

[0005] To address the problems of power source-load imbalance, voltage fluctuations, insufficient renewable energy absorption, and inadequate power supply guarantee during busy farming seasons, as well as high operating costs, caused by the dispersed nature of rural power distribution networks, long power supply radii, dispersed access to distributed renewable energy sources such as household photovoltaic and small-scale wind power in fields, and significant load fluctuations in agricultural irrigation and agricultural product processing, this invention provides a multi-objective, zoned, collaborative scheduling method for mobile energy storage vehicles in rural power distribution networks. This method adapts to the operational characteristics of rural power distribution networks—dispersed distribution areas, fragmented renewable energy access, and significant seasonal load fluctuations—while simultaneously considering network losses, renewable energy curtailment rates, and operating costs, thereby improving power supply guarantee capacity in key areas during busy farming seasons and centralized scheduling efficiency during off-seasons.

[0006] The technical solution adopted in this invention is as follows: An optimized scheduling method for interconnected energy storage vehicles for power supply risk control in rural power distribution networks includes the following steps: Step 1: Collect rural power distribution network node parameters, branch parameters, real-time load data, and distributed renewable energy output data, and analyze the uncertainty and time-series correlation of distributed renewable energy output; Step 2: Based on village distribution, field transformer distribution, road accessibility, and load time complementarity, divide the rural power distribution network into several dispatch zones, configure mobile energy storage vehicles, and establish corresponding dispatch constraints. Step 3: Construct a multi-objective sub-Bruker optimization model with the objectives of minimizing rural power distribution network loss, minimizing renewable energy curtailment rate, and minimizing operating costs, and establish a probability distribution fuzzy set constraint based on JS divergence that considers the uncertainty of distributed renewable energy output and time series correlation. Step 4: Perform convex optimization on the non-convex constraints in the multi-objective sub-Bruker optimization model, and use the column and constraint generation CCG algorithm for iterative solution to obtain the location of each mobile energy storage vehicle in each scheduling period, the charging power, the discharging power and the driving path, and generate a collaborative scheduling scheme for each scheduling zone of the rural power distribution network.

[0007] It also includes step 5: implementing differentiated scheduling based on the different operational needs of the busy and slack farming seasons, and making rolling corrections to the established scheduling plan based on the real-time load of the rural power distribution network, the real-time output of new energy sources, the charge status of mobile energy storage vehicles, and the traffic conditions of field roads.

[0008] In step 1, the collected data includes rural power distribution network node parameters, branch parameters, load data, renewable energy output data, and road condition data. Node parameters include node number, substation area, rated voltage, load type, capacity of mobile energy storage vehicle that can be connected, and critical load level; Branch parameters include the start and end points of the line, resistance, reactance, capacity limit, and switch status; Load data includes the active and reactive power of residential basic load, irrigation pump load, and agricultural product drying or cold chain processing load during each scheduling period; New energy output data includes the installed capacity, predicted output, real-time output, and historical prediction errors of household photovoltaic and small-scale wind power in fields; Road condition data includes road distances between stations, travel time, and accessibility status.

[0009] As shown in Figure 1, the target rural power distribution network includes multiple villages, transformer substations, household photovoltaic systems, small-scale wind power in fields, irrigation loads, agricultural product processing loads, and mobile energy storage vehicles, among other scheduling objects. Further, as shown in Figure 2, step 1 inputs the rural power grid's basic parameters, branch parameters, real-time load data, renewable energy output data, and road traffic status into the regional collaborative scheduling center, providing a data foundation for subsequent regional division, optimization solutions, and dynamic corrections.

[0010] Analysis of the uncertainty and time-series correlation of distributed renewable energy output based on historical data and real-time monitoring data; Set nodes i The actual output of distributed renewable energy in time period t is Predicted output is The prediction error of distributed renewable energy output is expressed as: (1); Calculate the prediction error variance based on historical samples: (2); In equation (2), H is the number of historical samples. Let h be the prediction error of the h-th historical sample in time period t. This represents the mean of the prediction error; The time-series correlation of distributed renewable energy output between adjacent scheduling periods is expressed as follows: (3); In equation (3), Let be the correlation coefficient of new energy output of node i in adjacent time periods t and t+1; Covariance calculation is used to measure the joint fluctuation of new energy output in adjacent time periods; The actual output of distributed new energy at node i in time period t+1; , denoted as the standard deviation of the distributed renewable energy output prediction error for node i in time periods t and t+1, respectively.

[0011] In step 2, the rural power distribution network is divided into several dispatch zones based on the geographical adjacency of villages, the distribution of field transformer areas, road accessibility, and load time complementarity. A corresponding number of mobile energy storage vehicles are configured according to the load level and new energy access scale of each dispatch zone. Let the partition correlation degree between any two transformer nodes a and b be . ,but: (4); In equation (4), This indicates the geographical adjacency of villages or areas; if they are adjacent, the value is 1, otherwise it is 0. This indicates the travel time between node a and node b. This represents the correlation coefficient between the load curves of two transformer substations. Indicates the degree of electrical connection correlation; These are the weighting coefficients; The load timing complementarity is characterized by the following formula (5): (5); In equation (5), The correlation coefficient between the load curves of transformer substation node a and transformer substation node b; These represent the load power of transformer nodes a and b during time period t, respectively. These are the corresponding average loads; For scheduling period index, This is the set of scheduling time periods.

[0012] In step 2, the number of mobile energy storage vehicles configured for the z-th scheduling partition is: (6); In equation (6), The number of mobile energy storage vehicles configured for the z-th scheduling partition; The maximum load of the z-th scheduling partition, This refers to the distributed renewable energy installed capacity of this region. This represents the maximum discharge power of a single mobile energy storage vehicle. For the importance load level coefficient of the zone, To configure weighting coefficients; This indicates the rounding up operation, used to ensure that the number of mobile energy storage vehicles is an integer.

[0013] In step 2, the scheduling constraints include at least node voltage deviation constraints, branch transmission power constraints, mobile energy storage vehicle travel path constraints, charging and discharging spatiotemporal coupling constraints, and energy storage state of charge constraints. 1) The node voltage deviation constraint is: (7); In equation (7), Let be the voltage amplitude of node i during time period t; These are the lower and upper voltage limits, respectively.

[0014] 2) Branch transmission power constraints are: (8); In equation (8) Let be the active power and reactive power of branch (i,j), respectively. This is the maximum capacity of the branch line; 3) Let This indicates whether the m-th mobile energy storage vehicle is stationed at node i in the distribution area during time period t. Let represent whether the m-th mobile energy storage vehicle travels from node i to node j during time period t. Then the location uniqueness constraint is: (9); In equation (9) N is the set of nodes in the transformer area, and A is the set of passable roads.

[0015] 4) Let Let be the charging state variables and discharging state variables of the m-th mobile energy storage vehicle, respectively. Then, the mutual exclusion constraints for charging / discharging and driving are as follows: (10); (11); (12); In the above formula These are charging power and discharging power, respectively. These are the maximum charging power and maximum discharging power, respectively, to ensure that each mobile energy storage vehicle can only be located in one dispatch zone or distribution node at any given time, and does not charge or discharge during the movement.

[0016] 5) The energy storage state of charge constraint is: (13); (14); In the above formula Let m be the state of charge of the m-th mobile energy storage vehicle during time period t. For rated capacity, These are charging efficiency and discharging efficiency, respectively. Energy consumption during driving; This represents the time interval between adjacent scheduling periods. These represent the minimum and maximum states of charge allowed for the m-th mobile energy storage vehicle, respectively.

[0017] In step 3, the objective function of the multi-objective split-bar optimization model is expressed as: (15); In equation (15): For a multi-objective function vector, Let the objective function be the network loss of the rural power distribution network. Let the objective function be the curtailment rate of renewable energy. The objective function is the comprehensive operating cost.

[0018] The target for network loss in rural power distribution networks is as follows: (16); In equation (16) For branch resistance, The square of the branch current amplitude; Indicates a branch Belongs to the rural power distribution network branch set ; For scheduling period index, This is the set of scheduling time periods.

[0019] The target for the curtailment rate of renewable energy is: (17); In equation (17) Contribute to new energy forecasting Actual absorption capacity of new energy The operating cost target is: (18); In formula (18) Electricity prices for main online purchases, Purchase power from the main network. The penalty coefficient for curtailment of renewable energy. This refers to the power that has been abandoned. Let m be the distance traveled by the m-th mobile energy storage vehicle during time period t. This is the driving cost coefficient. This is the charging and discharging loss cost coefficient; Number the mobile energy storage vehicle. A collection of mobile energy storage vehicles.

[0020] In step 3, a fuzzy set of probability distributions considering the uncertainty and time-series correlation of distributed renewable energy output is constructed based on JS divergence: set up Let be the random variable representing the output of distributed renewable energy in time period t. This represents the actual probability distribution. As a reference probability distribution, the fuzzy set of the probability distribution for a single time period is: (19); In equation (19): This represents the actual probability distribution Pt and the reference probability distribution for time period t. JS divergence between t; The JS divergence radius for a single time period; JS divergence is defined as: (20); (twenty one); In the above formula: , They represent the actual probability distributions respectively. and reference probability distribution Relative to mixed distribution KL divergence; Let KL divergence function be used. It is a mixed distribution formed by the equal weighted average of the actual probability distribution and the reference probability distribution.

[0021] To characterize the temporal correlation between adjacent scheduling periods, a joint probability distribution fuzzy set is constructed: (twenty two); In equation (22) adjacent time periods The joint probability distribution of new energy power output For reference to the joint probability distribution, Let be the JS divergence radius of the joint distribution.

[0022] In step 4, the non-convex constraints in the multi-objective sub-Bruker optimization model are subjected to convex optimization processing, including second-order conical convex relaxation of the branch power flow equations, and equivalent transformation of the mobility and charging / discharging coupling relationship of the mobile energy storage vehicle by introducing binary variables of the mobile energy storage vehicle's operating state, charging state, discharging state, and auxiliary variables; specifically as follows: Let the active power, reactive power, squared current amplitude, and squared node voltage amplitude of branch (i,j) in time period t be respectively... The branch power flow non-convex constraint is: (twenty three); Relax it to a second-order cone constraint: (twenty four); And equivalently represented as: (25); In equation (25): This represents the L2 norm operation, with the left side being the L2 norm of a vector composed of active power, reactive power, the square of voltage magnitude, and the square of current magnitude.

[0023] To investigate the coupling relationship between the charging / discharging and parking positions of mobile energy storage vehicles, auxiliary variables are introduced. And satisfy: (26); (27); (28); (29); In the above formula: Let be the charging auxiliary power variable of the m-th mobile energy storage vehicle when it is stationed at node i during time period t. Let be the discharge auxiliary power variable of the m-th mobile energy storage vehicle when it is stationed at node i during time period t. , These represent the total charging power and total discharging power of the m-th mobile energy storage vehicle during time period t, respectively. Let be a binary variable indicating whether the m-th mobile energy storage vehicle resides at node i during time period t; , These represent the maximum charging power and the maximum discharging power, respectively; N is the set of nodes in the distribution area.

[0024] Specifically, the charging and discharging auxiliary power variable is restricted to the corresponding dwell node through equations (26) and (27), when At time 0, the charging and discharging auxiliary power of this node is forced to 0; when At that time, the auxiliary power is constrained by the maximum charging power or the maximum discharging power. Then, by summing the auxiliary power of each node using equations (28) and (29), the total charging power and total discharging power of the vehicle are obtained, thereby converting the nonlinear coupling relationship between the mobile energy storage vehicle's parking position and the charging and discharging power into a solvable mixed integer constraint form.

[0025] In step 4, the column and constraint generation CCG algorithm is used to iteratively solve each single-objective sub-bar optimization model: The Bruker bar optimization model is expressed as follows: (30); In equation (30) Let X be the scheduling decision variable and X be the feasible region. Random variables contributing to new energy sources Probability distribution of power output from new energy sources.

[0026] A fuzzy set of probability distributions constructed based on JS divergence; Let fk(x,ξ) represent the expected value of the k-th objective function under probability distribution P; For the k-th optimization objective; The CCG main question is represented as: (31); (32); In the above formula Indicates simultaneous scheduling decision variables and auxiliary variables To optimize the variables, minimize them; In the main problem, auxiliary variables are used to characterize the upper bound of the target value in the worst-case scenario. In the first Extreme scenarios generated in the next iteration The next One objective function value; s is the extreme scenario number; The extreme scenario is generated in the s-th iteration; r is the current iteration number.

[0027] The subproblem is used to find the worst-case scenario under the current scheduling decision: (33); In equation (33) Add a new extreme scenario to the main problem for the (r+1)th iteration; Indicates an uncertain set The goal is to find new energy power output scenarios that maximize the value of the k-th objective function.

[0028] It is an uncertain set jointly determined by the JS divergence fuzzy set and the temporal correlation constraint; In order to make the current scheduling decision and scene The k-th objective function value; This represents the scheduling decision obtained in the r-th iteration.

[0029] When the upper and lower bounds are satisfied: (34) Stop iteration, where UB is the upper bound and LB is the lower bound. For convergence accuracy.

[0030] After solving each individual objective model separately, the optimal solution for each individual objective is obtained. Where x1*, x2*, and x3* represent network loss targets, respectively. curtailment rate target and operating cost targets The optimal scheduling decision obtained when optimizing alone.

[0031] And construct a payment matrix: (35); In equation (35) Payoff is the payment matrix; , , These represent the values ​​of the three single-objective optimal solutions in relation to the network loss objective; , , These represent the values ​​of the optimal solutions for the three single-objective goals on the renewable energy curtailment rate target; , , These represent the values ​​of the three single-objective optimal solutions in relation to the operating cost objective.

[0032] Determining the optimal value of the k-th objective based on the payoff matrix. and worst value Then normalize each single objective function (16) to (18) contained in equation (15): (36); In equation (36) The normalized target value of the k-th objective function; Let x be the original objective value of the k-th objective function under scheduling decision x.

[0033] The optimal solution for the multi-objective compromise is: (37); In equation (37) To make a compromise optimal scheduling decision for multiple objectives; The weight coefficients for the k-th objective satisfy: (38).

[0034] The solution results in step 4 include: the location of each mobile energy storage vehicle in its designated substation, charging power, discharging power, and travel path during each dispatching period. Based on this, a collaborative dispatching scheme for each dispatching zone of the rural power distribution network is generated. (39); In equation (39) This is a set of optimal solutions for the coordinated dispatch of mobile energy storage vehicles in rural power distribution networks. Let m be the optimal location for the m-th mobile energy storage vehicle during time period t. These are the decision variables for the optimal driving path. For optimal charging power, To achieve the optimal discharge power, The optimal state of charge; Number the mobile energy storage vehicle. For mobile energy storage vehicle collection, This is the set of nodes in the transformer area.

[0035] The station node where the m-th mobile energy storage vehicle resides during time period t is: (40); The set of travel paths for the m-th mobile energy storage vehicle is: (41); In equation (41) and These are the starting and ending station nodes of the driving route, respectively. For scheduling periods; Let be the optimal path decision variable for whether the m-th mobile energy storage vehicle travels from node i to node j during time period t.

[0036] In step 5, the scheduling rules include: during the busy farming season, priority is given to ensuring the power supply needs of areas with concentrated irrigation loads and areas with concentrated agricultural product processing loads, and a zone-priority scheduling method is adopted; during the off-season, a centralized scheduling method is adopted based on the level of surplus new energy and the cost of purchasing electricity from the main grid. Let the proportion of the busy agricultural load in the z-th scheduling partition during time period t be: (42); In equation (42) Let z be the proportion of the busy farming load in the z-th scheduling partition during time period t; For irrigation load, To reduce the processing load on agricultural products, Total load for the zone; When the following conditions are met: (43) The z-th scheduling partition is determined to enter a key support status during the busy farming season, where, Determine the threshold for the busy farming season; Assume the surplus power of new energy in rural power grids is: (44); In equation (44) The surplus new energy power that can be absorbed by the rural power distribution network for charging mobile energy storage vehicles during the time period t; Provide power to the distributed renewable energy source at node i during time period t; Let represent the load power of node i during time period t. i is the node number of the distribution area, and N is the set of distribution area nodes in the rural power distribution network.

[0037] When the following conditions are met: (45) A centralized dispatching method is adopted to prioritize the absorption of surplus power from new energy sources by mobile energy storage vehicles. The threshold for surplus power of new energy sources.

[0038] In step 5, the rolling correction adopts a finite-time rolling optimization method. After obtaining the real-time load of the rural power distribution network, the real-time output of new energy sources, the state of charge of mobile energy storage vehicles, and the traffic status of field roads at the current time t_0, the following optimization problem is solved: (46); In equation (46) From the current moment to The sequence of scheduling schemes within the rolling optimization window; H represents the current start time of scroll optimization; H represents the length of the scroll optimization window. The objective function includes network losses, curtailment rate, and operating costs. The set of mobile energy storage vehicle scheduling decisions to be optimized; This is the updated load forecast value; This is the updated forecast for new energy sources; This indicates the current state of charge of the mobile energy storage vehicle. This is the current road traffic status matrix.

[0039] Each rolling optimization only executes the scheduling decision for the current time period: (47); In equation (47) The scheduling decision actually executed in the current time period; Let m be the optimal dwell node for the m-th mobile energy storage vehicle at the current moment; The decision variable for the optimal driving path at the current moment; The optimal charging power at the current moment; This represents the optimal discharge power at the current moment.

[0040] After acquiring real-time data again in the next scheduling period, the station location, charging and discharging power and driving route of the mobile energy storage vehicle will be updated. In step 5, the multi-objective compromise model (37) and the rolling correction model (46) after convex optimization can be uniformly transformed into mixed integer quadratic constrained programming models (48) to (51), and solved using the Gurobi solver: The standard form of the mixed-integer quadratic constrained programming model is expressed as follows: (48); (49); (50); (51); In the above formula x' is a comprehensive decision variable in the standard form of the solver, which includes the location variables of the mobile energy storage vehicle, path variables, charging and discharging power variables, node voltage variables, and branch power flow variables. It is physically consistent with the scheduling decision variable x in equation (30). However, x' also includes second-order cone relaxation, auxiliary power variables, and auxiliary variables introduced after linearization, which are used to distinguish it from the abstract scheduling decision variable in equation (30).

[0041] Let c be the transpose of the linear cost coefficient vector. For comprehensive decision variables transpose, For secondary constraint numbering, It is the transpose symbol for a vector or matrix. It is the field of real numbers. For binary decision variables, For continuous decision variables; Q represents the model parameters obtained by transforming the rural power grid flow constraints, mobile energy storage vehicle operation constraints, and JS divergence fuzzy set constraints.

[0042] This invention provides an optimized scheduling method for interconnected energy storage vehicles for power supply risk control in rural power distribution networks, with the following technical advantages: 1) Step 1 collects data on power grid structure, agricultural load, renewable energy output, road traffic and mobile energy storage vehicle status simultaneously to form integrated dispatching basic data of source, load, grid, vehicle and road. This can more accurately identify the fluctuation of renewable energy output in rural power grid and the correlation between adjacent time periods, and reduce dispatching deviations caused by relying on a single load or a single renewable energy forecast.

[0043] 2) Step 2 combines the geographical adjacency of villages, the distribution of field transformer areas, road accessibility, and the complementarity of load timing for scheduling zones, so that mobile energy storage vehicles can prioritize serving areas with concentrated loads and road accessibility, reduce ineffective cross-zone travel distances, and improve the response speed of key transformer areas and the zoned power supply guarantee capability during the busy farming season.

[0044] 3) Step 3 takes network loss, renewable energy curtailment rate and operating costs as multiple objectives, and constructs a fuzzy set of distributed bar using JS divergence, so that the scheduling scheme can still maintain strong adaptability under renewable energy prediction errors and time-series fluctuations, and avoids the increase in renewable energy curtailment or operating costs caused by optimizing only a single prediction scenario.

[0045] 4) Step 4 uses second-order cone relaxation and auxiliary variable transformation to handle power flow constraints and vehicle parking and charging / discharging coupling relationships. This transforms the originally difficult-to-solve non-convex scheduling problem into a mixed integer optimization problem that can be stably solved by commercial solvers. Furthermore, extreme scenarios are gradually added using the CCG algorithm to improve the computability and convergence efficiency of robust scheduling results.

[0046] 5) Step 5 implements zoned priority scheduling or centralized scheduling based on the differences between the busy and slack farming seasons, and updates the load, new energy sources, charge status and road access status on a rolling basis during each scheduling period, so that the scheduling plan can be dynamically corrected as irrigation load increases suddenly, new energy sources drop sharply or roads are restricted, thereby improving the power supply reliability of rural power distribution networks and the utilization efficiency of mobile energy storage vehicles. Attached Figure Description

[0047] The present invention will be further described below with reference to the accompanying drawings and examples; Figure 1 This is a schematic diagram of a rural power grid zone collaborative scheduling scenario according to the present invention.

[0048] Figure 2 This is a diagram showing the deployment and collaborative scheduling relationship of the mobile energy storage vehicle of this invention.

[0049] Figure 3 This is a diagram showing the timing scheduling results of mobile energy storage vehicles.

[0050] Figure 4 A comparison chart showing the effects of different scheduling strategies. Detailed Implementation

[0051] The present invention will be further described in detail below with reference to examples and accompanying drawings, but the embodiments of the present invention are not limited thereto.

[0052] like Figure 1 As shown, this invention adopts a mobile energy storage vehicle scheduling strategy process of "partition modeling - collaborative scheduling - rolling correction," which follows an overall operational framework of "data acquisition - partitioning - optimization solution - dynamic correction." First, scheduling partitions are constructed based on rural power grid electrical connections, load distribution, new energy access, and road accessibility. Then, multi-objective collaborative scheduling of mobile energy storage vehicles is carried out. The specific steps are as follows: Step 1: Collect rural power grid node parameters, branch parameters, real-time load data, distributed new energy output data, road traffic status, and mobile energy storage vehicle operation parameters to construct a basic data model for rural power grid zoned collaborative scheduling.

[0053] Step 2: Based on the geographical adjacency of villages, the spatial distribution of transformer substations, electrical connections, road accessibility, load temporal complementarity, and the scale of new energy access, the target rural power grid is divided into scheduling zones, and the configuration scheme of mobile energy storage vehicles for each zone is determined.

[0054] Step 3: Based on the scheduling partitioning results, establish a multi-objective collaborative scheduling model for mobile energy storage vehicles. With the goals of reducing overall operating costs, reducing load loss, reducing renewable energy curtailment, and reducing ineffective vehicle driving, optimize the parking location, charging and discharging power, and driving path of mobile energy storage vehicles.

[0055] Step 4: Based on real-time load changes, fluctuations in renewable energy output, and road traffic conditions, the scheduling plan is continuously revised, and the access nodes, charging and discharging status, and cross-regional support paths of mobile energy storage vehicles are dynamically updated to output the final collaborative scheduling plan.

[0056] Figure 1 This is a schematic diagram of a collaborative scheduling scenario for a target rural power grid. The target rural power grid is divided into multiple collaborative scheduling zones. Each zone includes resources such as transformer substations, residential loads, irrigation loads, agricultural product processing loads, household photovoltaic power, or field wind power, and is equipped with corresponding mobile energy storage vehicles. The mobile energy storage vehicles can move between candidate service nodes, performing charging or discharging according to changes in renewable energy output and load demand, thereby achieving local support within the zone and flexible mutual assistance between zones.

[0057] Figure 2 This is a schematic diagram of the regional collaborative dispatch system. The system, with the regional collaborative dispatch center at its core, receives parameters from rural power grid nodes, branch parameters, real-time load data, renewable energy output data, road traffic status, and mobile energy storage vehicle parameters. It then generates the mobile energy storage vehicle's location, charging / discharging power, and travel route through a partitioning module, an optimization solution module, and a dynamic correction module. In implementation, the dispatch center partitions the system based on the spatial distribution, electrical connections, and road accessibility of transformer substations, load nodes, renewable energy nodes, and candidate energy storage access points. It establishes constraints such as node power balance, branch capacity, node voltage, state of charge, charging / discharging power, location-to-travel switching, travel time, and regional collaborative support. Further, it constructs a multi-objective optimization model to determine the final collaborative dispatch scheme.

[0058] Figure 3 This is a schematic diagram illustrating the dispatching results of mobile energy storage vehicles. In a typical 24-hour dispatching scenario during a busy farming season, the mobile energy storage vehicles switch their residing locations and adjust their charging and discharging states between different dispatching zones based on the availability of renewable energy and agricultural load demands. Figure 3It can be seen that mobile energy storage vehicle 1 resides in zone 1 from 1:00 to 5:00, moves to zone 2 from 6:00 to 11:00, returns to zone 1 from 12:00 to 16:00, moves to zone 2 again from 17:00 to 20:00, and returns to zone 1 from 21:00 to 24:00; mobile energy storage vehicle 2 resides in zone 2 from 1:00 to 6:00, moves to zone 1 from 7:00 to 12:00, returns to zone 2 from 13:00 to 18:00, moves to zone 1 from 19:00 to 22:00, and returns to zone 2 from 23:00 to 24:00. A negative power value indicates that the mobile energy storage vehicle is in a charging state, and a positive power value indicates that the mobile energy storage vehicle is in a discharging state. Between 11:00 and 14:00, household photovoltaic and small-scale wind power outputs are high, and there is a surplus of renewable energy in the target areas. Mobile energy storage vehicles prioritize absorbing surplus electricity near renewable energy access nodes. Between 12:00 and 14:00, the charging power of two mobile energy storage vehicles is approximately 6 kW to 16 kW. Between 18:00 and 21:00, irrigation load and agricultural product processing load overlap and reach peak levels. Mobile energy storage vehicles are transferred to areas with concentrated loads to provide discharge support, with a discharge power of approximately 4 kW to 20 kW. This helps maintain voltage levels in key areas and reduces peak power purchases from the main grid. When there is a sudden increase in load, a sharp drop in renewable energy output, or changes in road conditions, the dispatch center re-optimizes the subsequent dispatch plan based on the latest operational information, and makes rolling corrections to the mobile energy storage vehicle's location, charging and discharging power, and travel route.

[0059] Figure 4 The comparison results of the method of this invention with conventional scheduling methods and non-regional collaborative scheduling methods are shown in the figure. Taking the conventional scheduling method without mobile energy storage vehicles as the benchmark, under the same typical busy farming day example, the line network loss of the conventional scheduling method is 128.4, the line network loss of the non-regional collaborative scheduling method is 113.7, and the line network loss of the method of this invention is reduced to 91.2, a reduction of approximately 29.0% compared to the conventional scheduling method; the renewable energy curtailment rate of the conventional scheduling method is 15.6%, the renewable energy curtailment rate of the non-regional collaborative scheduling method is 11.3%, and the renewable energy curtailment rate of the method of this invention is reduced to 5.8%, a reduction of 9.8 percentage points compared to the conventional scheduling method; the operating cost of the conventional scheduling method is 8.95, the operating cost of the non-regional collaborative scheduling method is 7.88, and the operating cost of the method of this invention is reduced to 6.42, a reduction of approximately 28.3% compared to the conventional scheduling method. Figure 4As can be seen, the method of this invention, through scheduling zone division, cross-regional coordination of mobile energy storage vehicles, and charging and discharging optimization, can effectively reduce the power flow pressure and network loss of rural power distribution networks, reduce the curtailment of household photovoltaic and small-scale wind power in fields, and reduce the overall operating costs caused by disordered vehicle scheduling. Through the above method, mobile energy storage vehicles can be flexibly configured and collaboratively supported between different scheduling zones without significantly increasing the number of fixed energy storage devices. This method is suitable for rural power distribution network scenarios characterized by dispersed distribution areas, significant load fluctuations, uncertain renewable energy output, and complex road conditions.

Claims

1. A method for optimized scheduling of interconnected energy storage vehicles for power supply risk control in rural power distribution networks, characterized in that... Includes the following steps: Step 1: Collect rural power distribution network node parameters, branch parameters, real-time load data, and distributed renewable energy output data, and analyze the uncertainty and time-series correlation of distributed renewable energy output; Step 2: Based on village distribution, field transformer distribution, road accessibility, and load time complementarity, divide the rural power distribution network into several dispatch zones, configure mobile energy storage vehicles, and establish corresponding dispatch constraints. Step 3: Construct a multi-objective sub-Bruker optimization model with the objectives of minimizing rural power distribution network loss, minimizing renewable energy curtailment rate, and minimizing operating costs, and establish a probability distribution fuzzy set constraint based on JS divergence that considers the uncertainty of distributed renewable energy output and time series correlation. Step 4: Perform convex optimization on the non-convex constraints in the multi-objective sub-Bruker optimization model, and use the column and constraint generation CCG algorithm for iterative solution to obtain the location of each mobile energy storage vehicle in each scheduling period, the charging power, the discharging power and the driving path, and generate a collaborative scheduling scheme for each scheduling zone of the rural power distribution network.

2. The interconnected energy storage vehicle optimization scheduling method for power supply risk control in rural power distribution networks according to claim 1, characterized in that: In step 1, the uncertainty and time-series correlation of distributed new energy output are analyzed based on historical data and real-time monitoring data. Set nodes i The actual output of distributed renewable energy in time period t is Predicted output is The prediction error of distributed renewable energy output is expressed as: （1）； Calculate the prediction error variance based on historical samples: （2）； In equation (2), H is the number of historical samples. Let h be the prediction error of the h-th historical sample in time period t. This represents the mean of the prediction error; The time-series correlation of distributed renewable energy output between adjacent scheduling periods is expressed as follows: （3）； In equation (3), Let be the correlation coefficient of new energy output of node i in adjacent time periods t and t+1; Covariance calculation is used to measure the joint fluctuation of new energy output in adjacent time periods; The actual output of distributed new energy at node i in time period t+1; , denoted as the standard deviation of the distributed renewable energy output prediction error for node i in time periods t and t+1, respectively.

3. The interconnected energy storage vehicle optimization scheduling method for power supply risk control in rural power distribution networks according to claim 2, characterized in that: In step 2, the rural power distribution network is divided into several dispatch zones based on the geographical adjacency of villages, the distribution of field transformer areas, road accessibility, and load time complementarity. A corresponding number of mobile energy storage vehicles are configured according to the load level and new energy access scale of each dispatch zone. Let the partition correlation degree between any two transformer nodes a and b be . ,but: （4）； In equation (4), This indicates the geographical adjacency of villages or areas; if they are adjacent, the value is 1, otherwise it is 0. This indicates the travel time between node a and node b. This represents the correlation coefficient between the load curves of two transformer substations. Indicates the degree of electrical connection correlation; These are the weighting coefficients; The load timing complementarity is characterized by the following formula (5): （5）； In equation (5), The correlation coefficient between the load curves of transformer node a and transformer node b; These represent the load power of transformer nodes a and b during time period t, respectively. These are the corresponding average loads; For scheduling period index, This is the set of scheduling time periods.

4. The interconnected energy storage vehicle optimization scheduling method for power supply risk control in rural distribution networks according to claim 3, characterized in that: In step 2, the number of mobile energy storage vehicles configured for the z-th scheduling partition is: （6）； In equation (6), The number of mobile energy storage vehicles configured for the z-th scheduling partition; The maximum load of the z-th scheduling partition, This refers to the distributed renewable energy installed capacity of this region. This represents the maximum discharge power of a single mobile energy storage vehicle. For the importance load level coefficient of the zone, To configure weighting coefficients; This indicates the rounding up operation, used to ensure that the number of mobile energy storage vehicles is an integer.

5. The interconnected energy storage vehicle optimization scheduling method for power supply risk control in rural distribution networks according to claim 4, characterized in that: In step 2, the scheduling constraints include at least node voltage deviation constraints, branch transmission power constraints, mobile energy storage vehicle travel path constraints, charging and discharging spatiotemporal coupling constraints, and energy storage state of charge constraints. 1) The node voltage deviation constraint is: （7）； In equation (7), Let be the voltage amplitude of node i during time period t; These are the lower and upper voltage limits, respectively. 2) Branch transmission power constraints are: （8）； In equation (8) Let be the active power and reactive power of branch (i,j), respectively. This is the maximum capacity of the branch line; 3) Let This indicates whether the m-th mobile energy storage vehicle is stationed at node i in the distribution area during time period t. Let represent whether the m-th mobile energy storage vehicle travels from node i to node j during time period t. Then the location uniqueness constraint is: （9）； In equation (9) N is the set of nodes in the transformer area, and A is the set of accessible roads; 4) Let Let be the charging state variables and discharging state variables of the m-th mobile energy storage vehicle, respectively. Then, the mutual exclusion constraints for charging / discharging and driving are as follows: （10）； （11）； （12）； In the above formula These are charging power and discharging power, respectively. These are the maximum charging power and the maximum discharging power, respectively, to ensure that each mobile energy storage vehicle can only be located in one dispatch zone or distribution node at any given time, and does not charge or discharge during the movement. 5) The energy storage state of charge constraint is: （13）； （14）； In the above formula Let m be the state of charge of the m-th mobile energy storage vehicle during time period t. For rated capacity, These are charging efficiency and discharging efficiency, respectively. Energy consumption during driving; The time interval between adjacent scheduling periods; These represent the minimum and maximum states of charge allowed for the m-th mobile energy storage vehicle, respectively.

6. The interconnected energy storage vehicle optimization scheduling method for power supply risk control in rural power distribution networks according to claim 5, characterized in that: In step 3, the objective function of the multi-objective split-bar optimization model is expressed as: （15）； In equation (15): For a multi-objective function vector, Let the objective function be the network loss of the rural power distribution network. Let the objective function be the curtailment rate of renewable energy. The objective function is the overall operating cost; The target for network loss in rural power distribution networks is as follows: （16）； In equation (16) For branch resistance, The square of the branch current amplitude; Indicates a branch Belongs to the rural power distribution network branch collection ; For scheduling period index, For the set of scheduling periods; The target for the curtailment rate of renewable energy is: （17）； In equation (17) Contribute to new energy forecasting Actual absorption capacity of new energy The operating cost target is: （18）； In formula (18) Electricity prices for main online purchases, Purchase power from the main network. The penalty coefficient for curtailment of renewable energy. This refers to the power that has been abandoned. Let m be the distance traveled by the m-th mobile energy storage vehicle during time period t. This is the driving cost coefficient. This is the charging and discharging loss cost coefficient; Number the mobile energy storage vehicle. A collection of mobile energy storage vehicles.

7. The interconnected energy storage vehicle optimization scheduling method for power supply risk control in rural distribution networks according to claim 6, characterized in that: In step 3, a fuzzy set of probability distributions considering the uncertainty and time-series correlation of distributed renewable energy output is constructed based on JS divergence: set up Let be the random variable representing the output of distributed renewable energy in time period t. This represents the actual probability distribution. As a reference probability distribution, the fuzzy set of the probability distribution for a single time period is: （19）； In equation (19): This represents the actual probability distribution Pt and the reference probability distribution for time period t. JS divergence between t; The JS divergence radius for a single time period; JS divergence is defined as: （20）； （21）； In the above formula: , They represent the actual probability distributions respectively. and reference probability distribution Relative to mixed distribution KL divergence; Let KL divergence function be used. It is a mixed distribution formed by the equal weighted average of the actual probability distribution and the reference probability distribution; To characterize the temporal correlation between adjacent scheduling periods, a joint probability distribution fuzzy set is constructed: （22）； In equation (22) adjacent time periods The joint probability distribution of new energy power output For reference to the joint probability distribution, Let be the JS divergence radius of the joint distribution.

8. The interconnected energy storage vehicle optimization scheduling method for power supply risk control in rural distribution networks according to claim 7, characterized in that: In step 4, the non-convex constraints in the multi-objective sub-Bruker optimization model are subjected to convex optimization processing, including second-order conical convex relaxation of the branch power flow equations, and equivalent transformation of the mobility and charging / discharging coupling relationship of the mobile energy storage vehicle by introducing binary variables of the mobile energy storage vehicle's operating state, charging state, discharging state, and auxiliary variables; specifically as follows: Let the active power, reactive power, squared current amplitude, and squared node voltage amplitude of branch (i,j) in time period t be respectively... The branch power flow non-convex constraint is: （23）； Relax it to a second-order cone constraint: （24）； And equivalently represented as: （25）； In equation (25): Represents the L2 norm operation; To investigate the coupling relationship between the charging / discharging and parking positions of mobile energy storage vehicles, auxiliary variables are introduced. And satisfy: （26）； （27）； （28）； （29）； In the above formula: Let be the charging auxiliary power variable of the m-th mobile energy storage vehicle when it is stationed at node i during time period t. Let be the discharge auxiliary power variable of the m-th mobile energy storage vehicle when it is stationed at node i during time period t. , These represent the total charging power and total discharging power of the m-th mobile energy storage vehicle during time period t, respectively. Let be a binary variable indicating whether the m-th mobile energy storage vehicle resides at node i during time period t; , These represent the maximum charging power and the maximum discharging power, respectively; N is the set of nodes in the distribution area.

9. The interconnected energy storage vehicle optimization scheduling method for power supply risk control in rural distribution networks according to claim 8, characterized in that: In step 4, the column and constraint generation CCG algorithm is used to iteratively solve each single-objective sub-bar optimization model: The Bruker bar optimization model is expressed as follows: （30）； In equation (30) Let X be the scheduling decision variable and X be the feasible region. Random variables contributing to new energy sources Probability distribution of power output from new energy sources; It is a fuzzy set of probability distributions constructed based on JS divergence; Let fk(x,ξ) represent the expected value of the k-th objective function under probability distribution P; For the k-th optimization objective; The CCG main question is represented as: （31）； （32）； In the above formula Indicates simultaneous scheduling decision variables and auxiliary variables To optimize the variables, minimize them; In the main problem, auxiliary variables are used to characterize the upper bound of the target value in the worst-case scenario. In the first Extreme scenarios generated in the next iteration The next One objective function value; s is the extreme scenario number; The extreme scenario is generated in the s-th iteration; r is the current iteration number; The subproblem is used to find the worst-case scenario under the current scheduling decision: （33）； In equation (33) Add a new extreme scenario to the main problem for the (r+1)th iteration; Indicates an uncertain set Find the new energy output scenario that maximizes the value of the k-th objective function; It is an uncertain set jointly determined by the JS divergence fuzzy set and the temporal correlation constraint; In order to make the current scheduling decision and scene The k-th objective function value; The scheduling decision is obtained in the r-th iteration; When the upper and lower bounds are satisfied: (34) Stop iteration, where UB is the upper bound and LB is the lower bound. For convergence accuracy; After solving each individual objective model separately, the optimal solution for each individual objective is obtained. Where x1*, x2*, and x3* represent network loss targets, respectively. curtailment rate target and operating cost targets The optimal scheduling decision obtained when optimizing alone; And construct a payment matrix: （35）； In equation (35) Payoff is the payment matrix; , , These represent the values ​​of the three single-objective optimal solutions in relation to the network loss objective; , , These represent the values ​​of the optimal solutions for the three single-objective goals on the renewable energy curtailment rate target; , , These represent the values ​​of the three single-objective optimal solutions in relation to the operating cost objective; Determining the optimal value of the k-th objective based on the payoff matrix. and worst value Then normalize each single objective function (16) to (18) contained in equation (15): （36）； In equation (36) The normalized target value of the k-th objective function; Let x be the original objective value of the k-th objective function under scheduling decision x; The optimal solution for the multi-objective compromise is: （37）； In equation (37) To make a compromise optimal scheduling decision for multiple objectives; The weight coefficients for the k-th objective satisfy: （38）； The solution results include: the location of each mobile energy storage vehicle in its designated substation, charging power, discharging power, and travel path during each scheduling period. Based on this, a collaborative scheduling scheme for each scheduling zone of the rural power distribution network is generated. （39）； In equation (39) This is a set of optimal solutions for the coordinated dispatch of mobile energy storage vehicles in rural power distribution networks. Let m be the optimal location for the m-th mobile energy storage vehicle during time period t. These are the decision variables for the optimal driving path. For optimal charging power, To achieve the optimal discharge power, The optimal state of charge; Number the mobile energy storage vehicle. For mobile energy storage vehicle collection, For the set of nodes in the transformer area; The station node where the m-th mobile energy storage vehicle resides during time period t is: （40）； The set of travel paths for the m-th mobile energy storage vehicle is: （41）； In equation (41) and These are the starting and ending station nodes of the driving route, respectively. For scheduling periods; Let be the optimal path decision variable for whether the m-th mobile energy storage vehicle travels from node i to node j during time period t.

10. The interconnected energy storage vehicle optimized scheduling method for power supply risk control in rural distribution networks according to claim 9, characterized in that: It also includes step 5: Implement differentiated scheduling based on the different operational needs of the busy farming season and the off-season, and make rolling corrections to the established scheduling plan based on the real-time load of the rural power distribution network, the real-time output of new energy sources, the charge status of mobile energy storage vehicles, and the traffic status of field roads. The dispatching rules include: during the busy farming season, priority will be given to ensuring the power supply needs of areas with concentrated irrigation loads and areas with concentrated agricultural product processing loads, and a zone-priority dispatching method will be adopted; during the off-season, a centralized dispatching method will be adopted based on the level of surplus renewable energy and the cost of purchasing electricity from the main grid. Let the proportion of the busy agricultural load in the z-th scheduling partition during time period t be: （42）； In equation (42) Let z be the proportion of the busy farming load in the z-th scheduling partition during time period t; For irrigation load, To reduce the processing load of agricultural products, Total load for the zone; When the following conditions are met: (43) The z-th scheduling partition is determined to enter a key support status during the busy farming season, where, Determine the threshold for the busy farming season; Assume the surplus power of new energy in rural power grids is: （44）； In equation (44) The surplus new energy power that can be absorbed by the rural power distribution network for charging mobile energy storage vehicles during the time period t; Provide power to the distributed renewable energy source at node i during time period t; Let represent the load power of node i in time period t; i is the node number of the distribution area, and N is the set of distribution area nodes in the rural distribution network; When the following conditions are met: (45) A centralized dispatching method is adopted to prioritize the absorption of surplus power from new energy sources by mobile energy storage vehicles. The threshold for surplus power of new energy sources; The rolling correction adopts a finite-time rolling optimization method. After obtaining the real-time load of the rural power distribution network, the real-time output of new energy sources, the state of charge of mobile energy storage vehicles, and the traffic status of field roads at the current time t_0, the following optimization problem is solved: （46）； In equation (46) From the current moment to The sequence of scheduling schemes within the rolling optimization window; H represents the current start time of scroll optimization; H represents the length of the scroll optimization window. The objective function includes network losses, curtailment rate, and operating costs. The set of mobile energy storage vehicle scheduling decisions to be optimized; This is the updated load forecast value; This is the updated forecast for new energy sources; This indicates the current state of charge of the mobile energy storage vehicle. This is the current road traffic status matrix; Each rolling optimization only executes the scheduling decision for the current time period: （47）； In equation (47) The scheduling decision actually executed in the current time period; Let m be the optimal dwell node for the m-th mobile energy storage vehicle at the current moment; The decision variable for the optimal driving path at the current moment; The optimal charging power at the current moment; This represents the optimal discharge power at the current moment. After acquiring real-time data again in the next scheduling period, the station location, charging and discharging power and driving route of the mobile energy storage vehicle will be updated. The multi-objective compromise model (37) and the rolling correction model (46) after convex optimization can be uniformly transformed into mixed integer quadratic constrained programming models (48) to (51), and solved using the Gurobi solver: The standard form of the mixed-integer quadratic constrained programming model is expressed as follows: （48）； （49）； （50）； （51）； In the above formula x' is a comprehensive decision variable in the standard form of the solver, which includes the location variables of the mobile energy storage vehicle, path variables, charging and discharging power variables, node voltage variables and branch power flow variables. It is consistent with the scheduling decision variable x in equation (30) in physical meaning. However, x' also includes second-order cone relaxation, auxiliary power variables and auxiliary variables introduced after linearization, which are used to distinguish it from the abstract scheduling decision variable in equation (30). Let c be the transpose of the linear cost coefficient vector. For comprehensive decision variables transpose, For secondary constraint numbering, It is the transpose symbol for a vector or matrix. For the real number field; For binary decision variables, For continuous decision variables; Q represents the model parameters obtained by transforming the rural power grid flow constraints, mobile energy storage vehicle operation constraints, and JS divergence fuzzy set constraints.

Images