A pile station network multi-energy collaborative control method and device, electronic equipment and storage medium

Abstract

The application discloses a pile station network multi-energy collaborative control method and device, electronic equipment and a storage medium. The pile station network multi-energy collaborative control method comprises collecting multi-source operation data; inputting the multi-source operation data into a combined prediction model to generate a scheduling target; constructing a multi-objective optimization model to obtain a candidate scheduling scheme; solving the multi-objective optimization model by using an improved multi-objective particle swarm optimization algorithm to obtain an optimal scheduling scheme; generating and issuing corresponding control instructions according to the optimal scheduling scheme; collecting actual operation data after the control instructions are executed, calculating a prediction deviation and an optimization target deviation, and dynamically correcting related parameters of the combined prediction model and related parameters of the multi-objective optimization model through an adaptive learning mechanism. The application is used to improve the operation performance of the pile station network.

Description

Technical Field

[0001] This application belongs to the technical field of distributed renewable energy smart grids, specifically relating to a multi-energy coordinated control method, device, electronic equipment, and storage medium for a pile-station network. Background Technology

[0002] With the rapid popularization of electric vehicles and the large-scale access of distributed renewable energy, integrated "charging pile-storage-source" power stations, which integrate charging piles, energy storage systems and photovoltaic / wind power distributed renewable energy generation units, are becoming important new power system nodes. While serving the electrification of transportation, such charging pile-storage networks also have the potential to participate in grid regulation and achieve local energy optimization.

[0003] In existing technologies, such charging station networks face multiple challenges in actual operation. First, the charging load exhibits strong randomness and volatility, which, combined with the intermittent output of renewable energy, leads to severe fluctuations in the overall power of the station, impacting the distribution network and affecting power quality and equipment lifespan. Second, the energy management strategies for these stations often employ control methods based on simple rules or single-objective optimization, making it difficult to achieve global optimization among multiple conflicting objectives such as meeting user charging needs, reducing energy costs, smoothing grid-connected power, and improving green energy consumption. Third, most control models rely on fixed parameters and static predictions, failing to adapt to real-time changes in source-load characteristics, resulting in long-term performance degradation. Finally, existing solutions lack effective adaptive adjustment mechanisms, leading to poor control performance when prediction deviations or sudden changes in external conditions occur. Summary of the Invention

[0004] To address one or more of the aforementioned problems, the purpose of this application is to provide a multi-energy collaborative control method, device, electronic equipment, and storage medium for pile station networks, which enables pile station networks to avoid source load fluctuation impacts, achieve global optimal equilibrium among multiple objectives, improve long-term operation control performance, and enhance the high anti-interference capability of regulation.

[0005] The technical solution of this application is as follows.

[0006] The first aspect is a multi-energy collaborative control method for pile station networks, including the following steps: Collect multi-source operational data from the pile station network; Multi-source operational data is input into a combined prediction model to predict load and distributed renewable energy output at multiple time scales, and then a scheduling target is generated. Construct a multi-objective optimization model to obtain alternative scheduling schemes; The optimal solution set of the multi-objective optimization model is obtained by using a multi-objective optimization algorithm, and the optimal scheduling scheme is selected according to the preset decision preferences. Generate and issue corresponding control commands based on the optimal scheduling scheme; After collecting and executing control commands, the actual operation data of the pile station network is collected, the prediction deviation between the scheduling target and the actual operation data is calculated, and the optimization target deviation between the optimal scheduling scheme and the actual operation data is calculated. The relevant parameters of the combined prediction model and the relevant parameters of the multi-objective optimization model are dynamically corrected through an adaptive learning mechanism.

[0007] Preferably, the specific process for generating scheduling targets includes: The collected multi-source operational data is time-aligned and cleaned, and reconstructed into continuous time-series samples to form the packet prediction input dataset. Based on the predicted input dataset, a basic model for time series prediction within 24 hours of the next day and a fine-grained model for short-term prediction within 15 minutes to several hours are trained respectively.

[0008] Furthermore, the basic model, by mining the long-term correlation between load and weather, calendar, and electricity price, outputs charging load forecast curves and distributed renewable energy output forecast curves for each time period within the next 24 hours. The fine-grained model models rapid fluctuations caused by concentrated arrivals and sudden changes in cloud cover within a short timescale. In actual operation, it receives the latest time series data in a sliding time window and outputs short-term prediction results for the next 15 minutes to 4 hours in real time.

[0009] Preferably, the multi-objective optimization model includes an operating cost objective function, a grid interaction smoothing objective function, a renewable energy consumption objective function, and constraints.

[0010] Preferably, the Pareto optimal solution set is obtained by solving the multi-objective optimization model using an improved multi-objective particle swarm optimization algorithm; The improved multi-objective particle swarm optimization algorithm obtains the Pareto optimal solution set by solving a multi-objective optimization model through the following steps: The particle swarm is initialized within the decision space, and the position of each particle is encoded into a complete pile station network scheduling scheme. In each iteration, the target values ​​of operating cost, grid interaction smoothness and distributed renewable energy absorption rate are calculated based on the current position of the particle in the multi-objective optimization model. The Pareto dominance relation is used to sort the particle swarm in a non-dominated manner to form several layers of non-dominated solution sets. A decision evaluation model is constructed based on preset decision preferences such as operating cost, grid interaction smoothness, and distributed renewable energy absorption rate. The weight of each objective in the decision evaluation model is automatically calculated by the entropy weight method. The candidate schemes in the Pareto optimal solution set are comprehensively scored by the approximation ideal solution ranking method. The scheme with the highest comprehensive score is determined as the optimal scheduling scheme for this scheduling cycle.

[0011] Preferably, control commands are generated and issued, and the specific process includes: The optimal scheduling scheme is discretized on the time axis according to the preset scheduling cycle. Based on the target energy storage power, total power of charging piles, and grid connection point exchange power of each time period after discretization, the corresponding control command sequence is constructed to obtain a set of scheduling reference trajectories indexed by timestamps. Based on the set of scheduling reference trajectories, the energy storage power components are decomposed, and the target energy storage active power, power direction, and duration of each scheduling period are encapsulated into the charging and discharging control commands of the energy storage system. Based on the scheduling reference trajectory set, the total power component of the charging pile is refined at the user level. The target total charging power for each time period is matched with the real-time access vehicle list, the remaining charging amount of each vehicle, the preset departure time, and priority information. The power setting value that the corresponding charging pile should output in the current time period is calculated through the load allocation algorithm, and the power dynamic adjustment instruction of the charging pile is generated. Based on the grid-connected power components in the scheduling reference trajectory set, the target grid-connected active power for each time period, the reactive power when needed, or the power factor reference value are encapsulated into control commands for the grid connection point.

[0012] Preferably, the specific process for calculating the prediction bias and optimizing the target bias includes: Collect actual operational data at each sampling time within the corresponding period; The actual operating data is aligned hourly with the target values ​​corresponding to the full scheduling target and the optimal scheduling scheme to form a set of prediction errors and a sequence of deviations for each optimization target; The prediction error set and the deviation sequence of each optimization objective are input into the adaptive learning module, and the relevant parameters of the combined prediction model and the relevant parameters of the multi-objective optimization model are dynamically corrected according to the set learning rules.

[0013] Secondly, a multi-energy collaborative control device for a pile station network includes: The data acquisition module is used to collect multi-source operational data of the pile station network; The scheduling target generation module is used to input multi-source operation data into the combined prediction model, perform multi-time-scale load and distributed renewable energy output prediction, and generate scheduling targets. The multi-objective optimization module is used to construct a multi-objective optimization model to obtain alternative scheduling schemes; The optimal scheduling scheme module is used to solve the optimal solution set of the multi-objective optimization model using a multi-objective optimization algorithm, and select the optimal scheduling scheme according to the preset decision preferences. The control command module is used to generate and issue corresponding control commands based on the optimal scheduling scheme. The acquisition and correction module is used to acquire the actual operation data of the pile station network after the execution of control commands, calculate the prediction deviation between the scheduling target and the actual operation data, and the optimization target deviation between the optimal scheduling scheme and the actual operation data. It dynamically corrects the relevant parameters of the combined prediction model and the relevant parameters of the multi-objective optimization model through an adaptive learning mechanism.

[0014] Thirdly, an electronic device includes a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the multi-energy collaborative control method for pile station networks as described above.

[0015] Fourthly, a storage medium is a non-transitory computer-readable storage medium storing computer instructions for enabling a computer to implement the multi-energy collaborative control method for pile station networks as described above.

[0016] Compared with the prior art, the advantages of this application are as follows: It can effectively address the dual uncertainties of charging load and distributed renewable energy output, improve the robustness and feasibility of scheduling plans, and reduce the risk of control failure due to prediction deviations. It comprehensively considers multi-objective optimization of economy, grid friendliness, and green energy consumption, providing a flexible decision-making space and achieving synergistic improvement of multiple benefits. It can continuously update the combined prediction model and the single-objective optimization model based on actual operating data, has self-evolution capabilities, and significantly improves adaptability and performance retention in long-term operation. It transforms abstract optimization results into executable instructions for energy storage systems, charging piles, and grid connection points, providing a systematic solution for the intelligent and lean operation of the charging pile network. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a multi-energy collaborative control method for a pile station network according to this application.

[0018] Figure 2 This is a schematic diagram of the structural framework of a multi-energy collaborative control device for a pile station network according to this application.

[0019] Figure 3 This is a schematic diagram of the structural framework of an electronic device according to this application. Detailed Implementation

[0020] Referring to the illustrations, the principles of this application are illustrated by way of example implementation in a suitable operating environment. The following description is based on the specific embodiments of this application as illustrated, and should not be construed as limiting other specific embodiments not detailed herein.

[0021] like Figure 1As shown in the figure, this application discloses a multi-energy coordinated control method for a charging station network. The charging station network mainly includes multiple charging piles, energy storage systems, distributed renewable energy sources, and heat dissipation systems, and simultaneously connects to information transmitted from the upstream power grid. The method of this embodiment includes the following steps.

[0022] S1. Collect multi-source operation data of the pile station network.

[0023] In this embodiment, the multi-source operation data includes load data of multiple charging piles, state of charge and charging / discharging power of the energy storage system, real-time output data of distributed renewable energy, dispatch instructions from the upper-level power grid and real-time electricity price signals, etc.

[0024] In this embodiment, the system operation data includes the coolant inlet temperature, outlet temperature, and flow rate of the heat dissipation system, the junction temperature of key power devices in each subsystem, the state of charge, charging and discharging power, and battery cluster temperature of the energy storage system, as well as the charging power and demand power of the charging pile and the output data of distributed renewable energy.

[0025] For step S1, the multi-source operation data acquisition method can specifically be as follows: Periodically sample each DC / AC charging pile using the metering module and pile control unit of the charging pile within the station. Obtain information such as the three-phase voltage U(i,t), three-phase current I(i,t), instantaneous active power Pch(i,t), reactive power Qch(i,t), current output power level, target state of charge of the connected vehicle, and scheduled departure time for each charging pile at the current sampling time (i represents one of the three phases). Then, integrate the information into a load data set for the charging pile. For example, at a certain time t1, DC fast charging pile No. 1 collects data showing U = 750V, I = 180A, Pch = For a load of 135kW and a user-preset departure time of t1+40min, this set of data serves as the load data set for the corresponding network element. Through communication interfaces with the energy storage system's Battery Management System (BMS) and bidirectional converter controller, the system acquires operating parameters for each sampling cycle, including State of Charge (SOC(t), current charge / discharge power Pess(t), maximum allowable charge / discharge power range, battery cluster temperature, and alarm status. This forms the energy storage system's operating data set. For example, at the same time t1, SOC(t1) is 65%, Pess(t1) = -300kW indicates charging operation. Through communication interfaces with distributed renewable energy sources such as photovoltaic inverters and wind power grid-connected devices, the output of each distributed renewable energy source is sampled to obtain its active power Pres(j,t), grid connection voltage, power factor, and commissioning / limiting status (j represents the corresponding distributed renewable energy type), forming a real-time output data set for distributed renewable energy. Through integration with the dispatch automation system and electricity information collection system of the upper-level power grid, active / reactive power boundary commands, demand response, peak shaving and valley filling commands, time-of-use pricing, and real-time pricing are received and compiled into a dispatch and pricing data set for the current and future periods. After collection, the multi-source operation data is synchronized and cached by the station-level energy management system according to a unified timestamp, outliers are removed, and short-term missing values ​​are preprocessed using interpolation and / or delayed completion. Finally, within each dispatch cycle, a multi-source operation dataset is constructed, containing load data from multiple charging piles, the state of charge and charging / discharging power of the energy storage system, real-time output data of distributed renewable energy, and dispatch commands and real-time pricing signals from the upper-level power grid.

[0026] S2. Input multi-source operation data into the combined prediction model to predict load and distributed renewable energy output at multiple time scales, and generate scheduling targets.

[0027] In this embodiment, the specific process of generating the scheduling target includes: S21. The collected load data of multiple charging piles, the state of charge and charging / discharging power of the energy storage system, the real-time output data of distributed renewable energy, and the dispatch instructions and real-time electricity price signals of the upper-level power grid are time-aligned and cleaned, and reconstructed into continuous time-series samples by day and time period to form a predictive input dataset containing historical data from multiple days and the current real-time status. S22. Based on the prediction input dataset, train a basic model for time series prediction within 24 hours of the next day and a fine-grained model for short-term prediction within 15 minutes to several hours. The basic model mines the long-term correlation between load and weather, calendar, and electricity price, and outputs the charging load forecast curve and the distributed renewable energy output forecast curve for each time period within the next 24 hours. The charging load forecast curve and the distributed renewable energy output forecast curve are used as the initial input of the optimized scheduling scheme, so that the scheduling of the charging pile network has a global grasp of the trend of the whole day from the beginning. The fine-grained model models rapid fluctuations caused by factors such as concentrated arrivals and sudden changes in cloud cover within a short time scale. In actual operation, it receives the latest time series data in the form of a sliding time window and outputs short-term prediction results for the next 15 minutes to 4 hours in real time. When generating the scheduling target, the charging load forecast curve, the distributed renewable energy output forecast curve, and the constraints of real-time electricity price signals and scheduling instructions are combined to obtain the target power trajectory covering the entire 24 hours of the next day. During the operation phase of the energy storage network the following day, the short-term forecast results are compared point by point with the corresponding charging load forecast curve and distributed renewable energy output forecast curve according to a unified time scale to construct a deviation sequence. If the deviation exceeds a preset threshold at a certain moment or in a certain continuous period, the period is marked as a rolling optimization period that needs to be corrected. The corresponding values ​​on the charging load forecast curve and distributed renewable energy output forecast curve of the original period are replaced with the short-term forecast results, triggering the redistribution of charging and discharging power of the energy storage system and the power and time offset adjustment of the flexible charging load in the period. Thus, without changing the overall scheduling target for the next day, the scheduling target can be dynamically updated according to the actual operation.

[0028] S3. Construct a multi-objective optimization model to obtain alternative scheduling schemes.

[0029] In this embodiment, the multi-objective optimization model includes an operating cost objective function, a grid interaction smoothing objective function, a renewable energy consumption objective function, and constraints. Among them, the constraints include the grid transformer capacity and line power, the energy storage system charging and discharging power and state of charge, the minimum and maximum power of charging piles and the user's preset charging completion time, and the upper and lower limits of distributed renewable energy output.

[0030] In this embodiment, the data processing method of the multi-objective optimization model, under the condition of setting discrete time steps in the scheduling cycle, uses grid-connected power, energy storage system charging and discharging power, and total load power of charging piles at each discrete moment as decision variables. Combined with the charging load prediction curve and the distributed renewable energy output prediction curve, it constructs optimization objectives that characterize economic efficiency, grid friendliness, and the level of distributed renewable energy absorption. The operating cost objective function takes into account the cost of purchasing electricity from the grid, revenue from selling electricity to the grid, energy storage loss costs, and renewable energy generation subsidies. Specifically, it sums the products of grid-side power purchase and corresponding electricity prices for each time period within the entire scheduling cycle, subtracts the revenue from power fed back to the grid and the electricity price sold to the grid, and adds the energy storage loss cost calculated based on the energy storage charging and discharging power and cycle count. Simultaneously, it multiplies the locally absorbed distributed renewable energy power in each time period by the subsidy unit price as a negative cost term, thus obtaining the comprehensive operating cost objective function. Grid interaction smoothing... The objective function characterizes the fluctuation of power exchange between the power station network and the power grid. It accumulates the squares of the difference in active power between two adjacent scheduling periods over the entire scheduling cycle to obtain the sum of squares of the power change rate. Minimizing this value is used as the smoothing objective, making the optimized grid-connected power curve as continuous and gentle as possible, thereby reducing the impact on the upstream power grid. The renewable energy consumption objective function measures the proportion of distributed renewable energy absorbed by local loads and energy storage systems. It statistically analyzes the actual total output of distributed renewable energy in each period and the amount of electricity directly used by local charging loads and absorbed by energy storage systems to calculate the proportion of locally consumed electricity to total output. This proportion is then weighted and averaged or summed over the scheduling cycle. Maximizing this indicator is used as the target for local consumption of distributed renewable energy. When a scheme involves a large amount of distributed renewable energy being abandoned and fed to the grid or directly limited, this target value is relatively low, thus being penalized in comprehensive decision-making.

[0031] In this embodiment, constraints are used to ensure the physical feasibility and operational safety of the solution. The constraints on the grid transformer capacity and line power are as follows: at any given time, the apparent power formed by the combination of active and reactive power at the grid connection point must not exceed the transformer's rated capacity, and the power flow along each section of the line must not exceed the line's thermal stability limit and voltage allowable deviation range. If the calculated apparent power or line power exceeds the upper limit at any given time, the current scheduling scheme is determined to be an infeasible solution. The constraints on the energy storage system's charging and discharging power and state of charge (SOC) are as follows: upper and lower limits are set for the energy storage charging and discharging power; an energy balance equation is established for the SOC changes in adjacent time periods, and the SOC is limited to a preset safety range. When the predicted SOC value exceeds the upper or lower limits or the charging and discharging power exceeds the equipment's rated capacity for a certain time period, the solution is also determined to be infeasible. The constraints on the minimum and maximum power of charging piles and the user-preset charging completion time are as follows: The charging power of each connected vehicle is set with upper and lower limits within its technically permissible range. Based on the user-set target SOC and expected departure time, an energy accumulation constraint is established from the connection time to the departure time. It is required that the charging amount obtained by integration under the optimized scheme is not less than the user's demand. If charging cannot be completed within the predetermined time even under all power boundary conditions, the scheduling scheme is considered to fail to meet the user's service quality requirements. The constraints on the upper and lower limits of distributed renewable energy output are as follows: The output of photovoltaic, wind power, and other equipment in each time period is set with upper and lower limits determined by rated capacity, weather conditions, and operation and maintenance strategies. When the distributed energy utilization power given during the optimization process exceeds the predicted upper limit or falls below the technical lower limit, the scheme must be excluded.

[0032] In this embodiment, by jointly constructing the above-mentioned multiple types of objective functions and multiple categories of constraints, the multi-objective optimization model can simultaneously take into account the three objectives of lowest cost, smooth grid interaction, and high absorption of distributed renewable energy when solving the problem, and ensure that all alternative scheduling schemes meet the grid safety operation boundary, equipment operation capacity, and user charging needs.

[0033] S4. The improved multi-objective particle swarm optimization algorithm is used to solve the multi-objective optimization model, obtain the Pareto optimal solution set, and select the optimal scheduling scheme according to the preset decision preference.

[0034] In this embodiment, the improved multi-objective particle swarm optimization algorithm obtains the Pareto optimal solution set by solving the multi-objective optimization model, including: S41. Initialize the particle swarm in the decision space and encode the position of each particle into a complete pile station network scheduling scheme. In this embodiment, the charging station network scheduling scheme includes at least the charging and discharging power sequence of the energy storage system, the total power allocation sequence of the charging piles, and the exchange power sequence of the grid connection point for each scheduling period, so that the position of any particle corresponds to a set of feasible decision variables of the multi-objective optimization model. S42. In each iteration, based on the position of the current particle, the multi-objective optimization model is substituted to calculate the target values ​​such as operating cost, grid interaction smoothness, and distributed renewable energy absorption rate. The Pareto dominance relation is used to perform non-dominated sorting of the particle swarm to form several layers of non-dominated solution sets. In this embodiment, to maintain the diversity of the solution set distribution, particles in the same non-dominated layer are measured according to their crowding in the target space, and particles with larger crowding distances are preferentially retained to avoid the optimal solution being over-concentrated in local regions. During the velocity and position update process, a nonlinear dynamic reduction strategy is adopted for the inertia weight of the standard particle swarm optimization algorithm. That is, a larger inertia weight is given in the early stage of iteration to enhance the global search capability, and the inertia weight is gradually reduced in the later stage of iteration to improve the local optimization accuracy. At the same time, a guiding term based on Pareto dominance and crowding is introduced into the update formula to make particles move towards the "guiding particles" that are more representative in the current non-dominated solution set, thereby accelerating the convergence to the uniformly distributed Pareto front. When the number of iterations reaches a preset upper limit or the change in the global non-dominated solution set is less than a threshold for several consecutive generations, the current non-dominated solution set is output as the Pareto optimal solution set. S43. Based on preset decision preferences such as operating costs, grid interaction smoothness, and distributed renewable energy absorption rate, a decision evaluation model is constructed. The weights of each objective in the decision evaluation model are automatically calculated by the entropy weight method or the weights are set manually. The Pareto optimal solution set is comprehensively scored by combining the approximation ideal solution ranking method. The scheme with the highest comprehensive score is determined as the optimal scheduling scheme for this scheduling cycle.

[0035] In this embodiment, a nonlinear dynamic decreasing strategy is adopted for the inertia weight of the particle swarm optimization algorithm, and a particle position update mechanism based on congestion and Pareto dominance is introduced. The particle swarm is initialized, and each particle position represents a complete scheduling scheme including energy storage power, charging pile power, and grid-connected power. The target value of the scheme corresponding to each particle is calculated iteratively, and the individual optimal and global optimal solution sets are updated according to the Pareto dominance relationship. After reaching the maximum number of iterations, the non-dominated solution set is output as the Pareto front. A decision method based on entropy weight method and approximation ideal solution sorting method is adopted to select the optimal scheduling scheme to be executed from the Pareto front. During the iteration process, the inertia weight is set as a function that decreases nonlinearly with the iteration number. A large inertia weight is maintained in the early stage of iteration to enhance the global search capability, and the inertia weight is gradually reduced in the later stage of iteration to improve the local optimization accuracy, thereby taking into account both global exploration and local exploitation in the solution space.

[0036] S5. Based on the optimal scheduling scheme, generate and issue charging and discharging control commands for the energy storage system, dynamic power adjustment commands for charging piles, and power control commands for grid connection points to achieve multi-energy coordinated operation of charging piles, storage, power sources, and grids.

[0037] In this embodiment, control commands are generated and issued according to the optimal scheduling scheme. The specific process includes: S51. Discretize the optimal scheduling scheme on the time axis according to the preset scheduling cycle. Based on the target energy storage power, total power of charging piles, and grid connection point exchange power of each time period after discretization, construct the corresponding control command sequence to obtain a set of scheduling reference trajectories indexed by timestamps. S52. Based on the scheduling reference trajectory set, the energy storage power component is decomposed, and the target energy storage active power P_ess for each scheduling period is calculated. (t), power direction (charging / discharging), duration, encapsulated as a charge and discharge control command for an energy storage system; In this embodiment, if multiple battery clusters are configured within the station, the total power is further allocated among the different battery clusters according to the current SOC, health status, and rated capacity of each cluster, based on a set weight, forming a refined instruction that includes a "total power, cluster power mapping relationship." For example, at a certain time t2, the optimal solution gives P_ess. If (t2) = +500kW (representing discharge), it can be divided into two sets of instructions for discharging 300kW and 200kW respectively in two clusters of batteries. These instructions are then sent through the communication link between the station-level energy management system and the converter controller, enabling the energy storage system to participate in peak shaving and valley filling and ancillary services as planned. S53. Based on the scheduling reference trajectory set, refine the total power component of the charging pile at the user level, and determine the target total charging power P_ch,tot for each time period. (t) is matched with the real-time access vehicle list, the remaining charging capacity of each vehicle, the preset departure time, and priority information, and the power setpoint P_ch,i that the corresponding charging pile should output in the current time period is calculated using a load allocation algorithm. (t) generates a dynamic power adjustment command for the charging pile. When the target total power is lower than the full power charging demand of the vehicle, higher power is allocated to vehicles that are about to leave the station, and lower power is allocated to vehicles that leave the station later. For example, at the same time t3, the total power demand is 800kW. The system can allocate it to four charging piles with power of 250kW, 250kW, 200kW and 100kW respectively, so as to achieve flexible adjustment of the charging load under the premise of meeting the preset completion time constraints of all users. S54. Based on the grid-connected power components in the scheduling reference trajectory set, construct the power control command for the grid connection point, and set the target grid-connected active power P_grid for each time period. (t) When needed, the reactive power or power factor reference value is encapsulated as a control command for the grid connection point. This command is then used by the grid connection point controller and / or distribution network automation terminal to coordinate and control the energy storage and reactive power compensation devices within the station, ensuring that the measured grid-connected power tracks P_grid as closely as possible. (t); In this embodiment, when the grid frequency and / or voltage are detected to deviate from the normal range, the active / reactive power output of the energy storage system can be adjusted in a short time without significantly disrupting the original dispatching scheme, thereby completing a frequency regulation and / or voltage support. In this embodiment, the control commands for the energy storage system, charging piles, and grid connection points are uniformly time-aligned and over-limit checked by the station-level energy management system before being issued to ensure that there are no commands that exceed the rated capacity of the equipment or violate the grid operation boundary. During the issuance process, reliable communication is achieved through industrial Ethernet and / or fieldbus, and execution feedback is collected in real time to monitor command execution deviations.

[0038] In this embodiment, the optimal scheduling scheme can be further decomposed into a discrete control command sequence with 15-minute intervals; charging and discharging power commands issued by the station-level energy management system to each energy storage converter, including power value, direction, duration, etc.; power adjustment commands issued by the charging pile central controller to each charging pile, to dynamically allocate power while meeting the user's minimum charging needs; and adjusting the overall power exchange between the charging pile network and the power grid through the grid connection point controller, so that it tracks the target value of the scheduling.

[0039] Step S6: Collect the actual operation data of the pile station network after executing the corresponding control commands of the optimal scheduling scheme, calculate the prediction deviation between the scheduling target and the actual operation data, and the optimization target deviation between the optimal scheduling scheme and the actual operation data. Dynamically correct the relevant parameters of the combined prediction model and the relevant parameters of the multi-objective optimization model through an adaptive learning mechanism to achieve continuous optimization of the control strategy.

[0040] In this embodiment, the specific process of calculating the predicted deviation between the scheduling target and the actual operating data, and the optimized target deviation between the optimal scheduling scheme and the actual operating data, includes: S61. The station-level energy management system collects the actual charging load, actual distributed renewable energy output, actual energy storage system charging and discharging power, actual grid connection point switching power, and corresponding electricity price information at each sampling time within the cycle to form actual operating data. S62. Align the actual operating data with the charging load forecast curve, the distributed renewable energy output forecast curve, and the target value of the optimal dispatch scheme hourly, calculate the load forecast error and the distributed renewable energy forecast error at each time point, and summarize the forecast errors at each time point to form a forecast error set; at the same time, compare the actual values ​​of the optimization targets such as operating cost, grid interaction power fluctuation, and local absorption rate of distributed renewable energy with the target values ​​obtained by solving the multi-objective optimization model based on the actual operating data to obtain the deviation sequence of each optimization target; S63. Input the prediction error set and the deviation sequence of each optimization objective into the adaptive learning module, and dynamically correct the relevant parameters of the combined prediction model and the relevant parameters of the multi-objective optimization model through the set learning rules.

[0041] In this embodiment, a combined prediction model consisting of sub-models such as long short-term memory networks and temporal convolutional networks can be used. The weight coefficients of each sub-model are adjusted according to the magnitude of the prediction error of each sub-model in the most recent scheduling cycles. The weight of the sub-model with smaller error is gradually increased, while the weight of the sub-model with larger error is gradually decreased. If necessary, the internal parameters of the sub-model can also be updated by gradient descent or recursive least squares to make the new model parameters closer to the current operation characteristics of the pile station network.

[0042] In this embodiment, the weights of objectives such as "operating cost," "grid interaction smoothing," and "renewable energy absorption rate" in the multi-objective optimization model, as well as the slack variables of some constraints, can be adaptively adjusted. When the actual deviation of a certain objective is observed to be significantly greater than that of other objectives over a long period of time, the weight of that objective can be appropriately increased or the relevant constraints can be tightened to guide the particle swarm to search for scheduling schemes with higher smoothness in subsequent iterations. Conversely, when the distributed renewable energy absorption rate is lower than expected for a long period of time, the weight of the distributed renewable energy absorption target can be moderately increased and the operating cost can be allowed to rise slightly, thereby rebalancing among different objectives.

[0043] By updating the combined prediction model and the multi-objective optimization model, the prediction results and optimization preferences used in subsequent scheduling cycles can be corrected based on historical operating deviation data, realizing the self-learning and self-evolution of the control strategy. This enables the multi-energy collaborative control of the pile station network of the present invention to continuously approach the true characteristics and comprehensive optimal operating state of the system during long-term operation.

[0044] Compared with the prior art, the beneficial effects of this embodiment are as follows: It can effectively address the dual uncertainties of charging load and distributed renewable energy output, improve the robustness and feasibility of scheduling plans, and reduce the risk of control failure due to prediction deviations. It comprehensively considers multi-objective optimization of economy, grid friendliness, and green energy consumption, providing a flexible decision-making space and achieving synergistic improvement of multiple benefits. It can continuously update the combined prediction model and the single-objective optimization model based on actual operating data, has self-evolution capabilities, and significantly improves adaptability and performance retention in long-term operation. It transforms abstract optimization results into executable instructions for energy storage systems, charging piles, and grid connection points, providing a systematic solution for the intelligent and lean operation of the charging pile network.

[0045] like Figure 2 As shown in the figure, this application discloses a multi-energy collaborative control device for pile station network, which includes the following structure.

[0046] The data acquisition module is used to collect multi-source operational data from the pile station network.

[0047] The scheduling target generation module is used to input multi-source operation data into the combined prediction model, perform multi-time-scale load and distributed renewable energy output prediction, and generate scheduling targets.

[0048] The multi-objective optimization module is used to build a multi-objective optimization model to obtain alternative scheduling schemes.

[0049] The optimal scheduling scheme module is used to solve the multi-objective optimization model using an improved multi-objective particle swarm optimization algorithm, obtain the Pareto optimal solution set, and select the optimal scheduling scheme according to the preset decision preferences.

[0050] The control command module is used to generate and issue corresponding control commands based on the optimal scheduling scheme.

[0051] The acquisition and correction module is used to acquire the actual operation data of the pile station network after the execution of control commands, calculate the prediction deviation between the scheduling target and the actual operation data, and the optimization target deviation between the optimal scheduling scheme and the actual operation data. It dynamically corrects the relevant parameters of the combined prediction model and the relevant parameters of the multi-objective optimization model through an adaptive learning mechanism.

[0052] The apparatus described above is used to implement the corresponding multi-energy collaborative control method for pile station network in the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0053] like Figure 3 As shown, based on the same inventive concept, corresponding to any of the above embodiments, this application also discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the above-mentioned multi-energy collaborative control method for pile station networks.

[0054] Specifically, the device includes: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, memory 1020, input / output interface 1030, and communication interface 1040 are interconnected within the device via the bus 1050.

[0055] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), GPU (Graphics Processing Unit), or one or more integrated circuits, to implement relevant programs and achieve the technical solutions provided in the embodiments of this specification.

[0056] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1020 and called by the processor 1010. The input / output interface 1030 is used to connect input / output modules to realize information input and output. Input / output modules can be configured as components in the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices may include keyboards, mice, touch screens, microphones, various sensors, etc., and output devices may include displays, projectors, speakers, vibrators, indicator lights, etc.

[0057] The communication interface 1040 is used to connect a communication module (not shown in the figure) to enable communication between this device and other devices. The communication module can communicate via wired means (such as USB (Universal Serial Bus), network cable, etc.) or wireless means (such as mobile network, WIFI (Wireless Fidelity), Bluetooth, etc.).

[0058] Bus 1050 includes a pathway for transmitting information between various components of the device, such as processor 1010, memory 1020, input / output interface 1030, and communication interface 1040.

[0059] It should be noted that although the above-described device only shows the processor 1010, memory 1020, input / output interface 1030, communication interface 1040, and bus 1050, in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may only include the components necessary for implementing the embodiments of this specification, and not necessarily all the components shown in the figures.

[0060] The electronic devices described above are used to implement the corresponding multi-energy collaborative control method for pile station networks in any of the foregoing embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0061] Based on the same inventive concept, corresponding to any of the above embodiments, this application also discloses a non-transitory computer-readable storage medium that stores computer instructions for enabling a computer to implement the above-described pile station network multi-energy collaborative control method.

[0062] The computer-readable storage medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transfer medium, which can be used to store information accessible by a computing device. The computer instructions stored in the storage medium of the above embodiments are used to enable the computer to implement the multi-energy collaborative control method of the pile station network as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0063] The above description is merely a preferred embodiment and the technical principles employed in this application. This application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions that can be made by those skilled in the art will not depart from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of this application. The scope of this application is determined by the scope of the claims.

Claims

1. A multi-energy collaborative control method for a pile station network, characterized in that, Including the following steps: Collect multi-source operational data from the pile station network; The multi-source operation data is input into the combined prediction model to predict load and distributed renewable energy output at multiple time scales, and then a scheduling target is generated. Construct a multi-objective optimization model to obtain alternative scheduling schemes; The optimal solution set of the multi-objective optimization model is obtained by using a multi-objective optimization algorithm, and the optimal scheduling scheme is selected according to the preset decision preferences. Generate and issue corresponding control commands based on the optimal scheduling scheme; After the control command is executed, the actual operation data of the pile station network is collected, the prediction deviation between the scheduling target and the actual operation data and the optimization target deviation between the optimal scheduling scheme and the actual operation data are calculated, and the relevant parameters of the combined prediction model and the relevant parameters of the multi-objective optimization model are dynamically corrected through an adaptive learning mechanism.

2. The multi-energy collaborative control method for pile station network according to claim 1, characterized in that, The specific process of generating scheduling targets includes: The collected multi-source operational data is time-aligned and cleaned, and reconstructed into continuous time-series samples to form the packet prediction input dataset. Based on the predicted input dataset, a basic model for time series prediction within 24 hours of the next day and a fine-grained model for short-term prediction within 15 minutes to several hours are trained respectively.

3. The multi-energy collaborative control method for pile station network according to claim 2, characterized in that, The basic model mines the long-term correlation between load and weather, calendar, and electricity price, and outputs charging load forecast curves and distributed renewable energy output forecast curves for each time period within the next 24 hours. The fine-grained model models rapid fluctuations caused by concentrated arrivals and sudden changes in cloud cover within a short timescale. In actual operation, it receives the latest time series data in a sliding time window and outputs short-term prediction results for the next 15 minutes to 4 hours in real time.

4. The multi-energy collaborative control method for pile station network according to claim 1, characterized in that, The multi-objective optimization model includes the operating cost objective function, the grid interaction smoothing objective function, the renewable energy consumption objective function, and the constraints.

5. The multi-energy collaborative control method for pile station network according to claim 1, characterized in that, The Pareto optimal solution set is obtained by solving the multi-objective optimization model using an improved multi-objective particle swarm optimization algorithm. The improved multi-objective particle swarm optimization algorithm for solving the multi-objective optimization model and obtaining the Pareto optimal solution set includes: The particle swarm is initialized within the decision space, and the position of each particle is encoded into a complete pile station network scheduling scheme. In each iteration, the target values ​​of operating cost, grid interaction smoothness and distributed renewable energy absorption rate are calculated based on the current position of the particle in the multi-objective optimization model. The Pareto dominance relation is used to sort the particle swarm in a non-dominated manner to form several layers of non-dominated solution sets. A decision evaluation model is constructed based on preset decision preferences such as operating cost, grid interaction smoothness, and distributed renewable energy absorption rate. The weight of each objective in the decision evaluation model is automatically calculated by the entropy weight method. The candidate schemes in the Pareto optimal solution set are comprehensively scored by the approximation ideal solution ranking method. The scheme with the highest comprehensive score is determined as the optimal scheduling scheme for this scheduling cycle.

6. The multi-energy collaborative control method for pile station network according to claim 1, characterized in that, The process of generating and issuing control commands includes: The optimal scheduling scheme is discretized on the time axis according to the preset scheduling cycle. Based on the target energy storage power, total power of charging piles, and grid connection point exchange power of each time period after discretization, the corresponding control command sequence is constructed to obtain a set of scheduling reference trajectories indexed by timestamps. Based on the set of scheduling reference trajectories, the energy storage power components are decomposed, and the target energy storage active power, power direction, and duration of each scheduling period are encapsulated into charging and discharging control commands for the energy storage system. Based on the scheduling reference trajectory set, the total power component of the charging pile is refined at the user level. The target total charging power for each time period is matched with the real-time access vehicle list, the remaining charging amount of each vehicle, the preset departure time, and priority information. The power setting value that the corresponding charging pile should output in the current time period is calculated through the load allocation algorithm, and the power dynamic adjustment instruction of the charging pile is generated. Based on the grid-connected power components in the scheduling reference trajectory set, the target grid-connected active power for each time period, the reactive power when needed, or the power factor reference value are encapsulated into control commands for the grid connection point.

7. The multi-energy collaborative control method for pile station network according to claim 1, characterized in that, The specific process of calculating prediction bias and optimizing target bias includes: Collect actual operational data at each sampling time within the corresponding period; The actual operating data is aligned hourly with the target values ​​corresponding to the full scheduling target and the optimal scheduling scheme to form a prediction error set and a deviation sequence for each optimization target; The prediction error set and the deviation sequence of each optimization objective are input into the adaptive learning module, and the relevant parameters of the combined prediction model and the relevant parameters of the multi-objective optimization model are dynamically corrected according to the set learning rules.

8. A multi-energy collaborative control device for a pile station network, characterized in that, include: The data acquisition module is used to collect multi-source operational data of the pile station network; The scheduling target generation module is used to input the multi-source operation data into the combined prediction model, perform multi-time-scale load and distributed renewable energy output prediction, and then generate scheduling targets. The multi-objective optimization module is used to construct a multi-objective optimization model to obtain alternative scheduling schemes; The optimal scheduling scheme module is used to solve the optimal solution set of the multi-objective optimization model using a multi-objective optimization algorithm, and select the optimal scheduling scheme according to the preset decision preferences. The control command module is used to generate and issue corresponding control commands according to the optimal scheduling scheme. The acquisition and correction module is used to acquire the actual operating data of the pile station network after the control command is executed, calculate the prediction deviation between the scheduling target and the actual operating data, the optimization target deviation between the optimal scheduling scheme and the actual operating data, and dynamically correct the relevant parameters of the combined prediction model and the relevant parameters of the multi-objective optimization model through an adaptive learning mechanism.

9. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, it implements the multi-energy collaborative control method for pile station network according to any one of claims 1-7.

10. A storage medium, being a non-transitory computer-readable storage medium, characterized in that, The system stores computer instructions for causing the computer to implement the multi-energy collaborative control method for pile station networks according to any one of claims 1-7.

Images