Multi-functional slot intelligent charging pile remote control and dispatching system

Abstract

The application discloses a multifunctional slot intelligent charging pile remote control and scheduling system, and particularly relates to the field of charging piles, which is used to solve the problems of low charging resource utilization and large power grid load impact, and provides a precise basis for power distribution based on federal learning to collect and predict user charging demand and vehicle charging capacity in real time; the predicted result is combined with real-time load data of the power grid through a multi-objective optimization algorithm to generate a power distribution baseline value that takes into account user demand and power grid peak clipping and valley filling, so that the charging resource and the stability of the power grid are effectively balanced; a dynamic game model of the user and the power grid is constructed by using a reinforcement learning algorithm, the output power of the charging pile is adjusted in real time, and the user behavior is guided through an integral incentive strategy, so that the conflict between demand and load is dynamically coped with, and the adaptability of the system is improved; finally, the digital twin technology is used to perform full-link simulation verification on the charging pile group, the scheduling strategy is continuously optimized based on actual operation data comparison, and the continuous improvement of the system performance is ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of charging piles, more particularly, to a multifunctional slot intelligent charging pile remote control and scheduling system. BACKGROUND

[0002] In the actual operation of the intelligent charging pile system, guiding users to charge during the off-peak period of the power grid through time-of-use pricing strategy is an important means to achieve peak load shifting. However, the actual charging behavior of users is often affected by factors such as travel changes and temporary vehicle needs, resulting in deviations between the scheduled charging time and the actual situation. At the same time, the battery status (such as the remaining power and the charging rate limit) of different vehicles changes dynamically during the charging process, which makes it difficult for the system to adapt to real-time needs with a fixed power allocation strategy. When the actual charging needs of users are misaligned with the preset time period, or the vehicle battery cannot be charged at the expected power, the system cannot dynamically adjust the charging task priority, nor can it balance the contradiction between the power grid load regulation target and the individual needs of users. This rigid scheduling logic not only wastes charging resources during the off-peak period, but also may exacerbate power grid fluctuations due to inaccurate power distribution during peak periods, ultimately weakening the actual effect of the peak load shifting strategy.

[0003] To solve the above problems, a technical solution is provided. SUMMARY

[0004] To overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a multifunctional slot intelligent charging pile remote control and scheduling system. First, based on federated learning technology, user charging needs and vehicle charging capacity are collected and predicted in real time, ensuring data privacy while providing accurate basis for power allocation; through a multi-objective optimization algorithm, the prediction results are combined with real-time load data of the power grid to generate a power allocation baseline value that takes into account both user needs and power grid peak load shifting, effectively balancing charging resources and power grid stability; a dynamic game model of users and the power grid is constructed using reinforcement learning algorithm to adjust the output power of the charging pile in real time and guide user behavior through an integral incentive strategy, dynamically responding to demand and load conflicts, and improving system adaptability; finally, digital twin technology is used to simulate and verify the entire link of the charging pile group, and the scheduling strategy is continuously optimized based on actual operation data to ensure continuous improvement of system performance, thereby solving the problems raised in the background art.

[0005] To achieve the above-mentioned purposes, the present application provides the following technical solutions:

[0006] Federal learning module: used to collect vehicle arrival time deviation and battery state parameters in real time, establish user behavior prediction model and battery charging capacity model based on federal learning;

[0007] Optimization scheduling module: combines the output results of the user behavior prediction model with the real-time load data of the power grid, and generates power distribution baseline values through a multi-objective optimization algorithm;

[0008] Dynamic game module: based on reinforcement learning algorithm, constructs a dynamic game model between users and the power grid, integrates user behavior deviation information and power grid load relief information, adjusts the output power of the charging pile in real time according to the power distribution baseline value, and drives the user charging behavior to shift to low load period according to the integral incentive strategy;

[0009] Simulation verification module: through digital twin technology, the charging pile group is simulated and verified for full link, the simulation data and actual operation data are compared, and the optimization prompt signal is output.

[0010] Further, the federal learning module collects and preprocesses the vehicle arrival time deviation and battery state information of the charging pile in real time; based on federal learning technology, local models are trained on each charging pile node, and a global model is generated by aggregation through a central server.

[0011] Further, the federal learning module uses support vector regression to construct a battery charging capacity model to predict the charging capacity of the vehicle; finally, the trained model is deployed to each charging pile node to generate real-time charging demand prediction values and charging capacity prediction values.

[0012] Further, the optimization scheduling module obtains the charging demand prediction values and charging capacity prediction values, and collects power grid load data and power grid load target curve in real time; the power grid real-time load is smoothed and the power grid load deviation is calculated; a multi-objective optimization model is constructed, taking the power distribution baseline value of each charging pile at each time point as the decision variable, while realizing the optimization of power grid peak clipping and valley filling and user charging demand satisfaction; set the constraint condition to ensure that the power distribution baseline value is within the charging pile capacity range, meets the charging demand and controls the power grid load deviation; design the objective function, evaluate the power grid peak clipping and valley filling effect in the form of exponential decay penalty, and measure the user charging demand satisfaction in the form of excess index; use the improved particle swarm optimization algorithm to obtain the optimal solution; finally, the power distribution baseline value of each charging pile at each time point is generated.

[0013] Further, the user behavior deviation information includes a user behavior deviation index, which is obtained by calculating the relative deviation of the actual charging power and the predicted charging power within a time window and nonlinear mapping, with a value range of 0 to 1.

[0014] Furthermore, the grid load relief information includes a grid load relief index, which is obtained by evaluating the average ratio of the difference between the real-time load and the load target in the historical window and converting it through a Sigmoid function, and has a value range of 0 to 1.

[0015] Furthermore, the dynamic game module calculates the average value of the user behavior deviation index of all vehicles at the selected time point; the geometric mean method is used to integrate the average value and the inverse value of the grid load relief index to obtain the user-grid response synergy index.

[0016] Furthermore, the dynamic game module adopts a multi-agent reinforcement learning framework, treating each charging pile as an agent and the power grid as part of the environment; the state space includes the grid load at the current time point, the power allocation baseline value, the actual output power of each charging pile, and the user-grid response synergy index; the action space is the power increase or decrease relative to the power allocation baseline value, and the amplitude is limited by the adjustable range of a single pile; the reward function is designed to be the inverse of the weighted sum of the user-grid response synergy index and the grid load deviation; training adopts the deep Q network algorithm, and the experience replay mechanism is used to improve learning stability.

[0017] Furthermore, if the actual output power continues to deviate from the power allocation baseline value and the user-grid response synergy index exceeds the preset threshold, the dynamic game module guides users to adjust their charging behavior through point incentives.

[0018] Furthermore, the simulation verification module uses a dynamic time warping algorithm to align the simulation data with the actual data, and generates an optimization prompt signal if the power deviation or load deviation exceeds the corresponding preset threshold.

[0019] The technical effects and advantages of the multifunctional slot intelligent charging pile remote control and scheduling system of the present invention are as follows:

[0020] The multifunctional slot intelligent charging pile remote control and scheduling system of the present application significantly improves the utilization rate of charging resources, optimizes power grid load management, enhances user experience and realizes intelligent remote management. First, based on federated learning technology, user charging demand and vehicle charging capacity are collected and predicted in real time, ensuring data privacy while providing accurate basis for power allocation; the predicted results are combined with real-time load data of the power grid through a multi-objective optimization algorithm to generate a power allocation baseline value that takes into account user demand and power grid peak load shifting, effectively balancing charging resources and power grid stability; a dynamic game model of users and the power grid is constructed using reinforcement learning algorithm to adjust the output power of the charging pile in real time and guide user behavior through an integral incentive strategy, dynamically responding to demand and load conflicts and improving system adaptability; finally, digital twin technology is used to simulate and verify the entire link of the charging pile group, and the scheduling strategy is continuously optimized based on actual operation data to ensure continuous improvement of system performance. In summary, the present application effectively solves the problems of waste of traditional charging pile resources, lagging management and power grid impact, promotes the intelligentization and sustainable development of charging infrastructure, and has significant practical value and popularization potential. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 FIG. 1 is a structural schematic diagram of the multifunctional slot intelligent charging pile remote control and scheduling system of the present application.

[0022] Figure 2 FIG. 4 is a running process schematic diagram of the federated learning module of the multifunctional slot intelligent charging pile remote control and scheduling system of the present application. DETAILED DESCRIPTION

[0023] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0024] Embodiment 1: Figure 1 The multifunctional slot intelligent charging pile remote control and scheduling system of the present application is given, which includes:

[0025] Federated learning module: used for real-time collection of vehicle arrival time deviation and battery state parameters, establishment of user behavior prediction model and battery charging capacity model based on federated learning;

[0026] Optimization scheduling module: combining the output results of the user behavior prediction model with real-time load data of the power grid, generating a power allocation baseline value through a multi-objective optimization algorithm;

[0027] Dynamic game module: based on reinforcement learning algorithm, construct the dynamic game model between users and power grid, integrate user behavior deviation information and power grid load relief information, adjust the output power of charging pile in real time according to the power allocation baseline value, and drive the user charging behavior to deviate to low load period according to the integral incentive strategy;

[0028] Simulation verification module: through digital twin technology, the whole link simulation verification of charging pile group is carried out, the simulation data and the actual operation data are compared, and the optimization prompt signal is output.

[0029] In the intelligent charging pile remote control and dispatching method, the federal learning module aims to use federal learning technology, based on the real-time collected vehicle arrival time deviation and battery state parameters of charging piles, to construct user behavior prediction model and battery charging capacity model, and generate accurate charging demand prediction value and charging capacity prediction value. These prediction values will be directly used for calculating the power allocation baseline value in the optimization dispatching module combined with the power grid load data, so as to realize the dynamic response of charging piles to user demand and vehicle state, and provide support for power grid load balancing.

[0030] As shown in Figure 2 , the federal learning module includes the following contents:

[0031] S1-1, data acquisition and preprocessing:

[0032] Each charging pile records relevant data in real time and performs preliminary processing. Each charging pile is referred to as a node, and the number of nodes is numbered from 1 to the total number. For each vehicle, record the difference between its actual arrival time and expected arrival time, called vehicle arrival time deviation, in minutes. At the same time, collect the battery state information of each vehicle, including battery remaining capacity (expressed in percentage), battery total capacity (expressed in kilowatt hours) and maximum charging power allowed by the battery (expressed in kilowatts). In the preprocessing stage, first, remove abnormal data: if the arrival time deviation of a vehicle exceeds 120 minutes, or the battery remaining capacity is less than 0% or greater than 100%, the data of that vehicle is considered abnormal and removed. Next, normalize the remaining data: for each vehicle's arrival time deviation, battery remaining capacity, battery total capacity and maximum charging power, calculate the processing value of each parameter, that is, subtract the minimum value of the parameter in the history of the charging pile from the original value of the parameter, and divide by the difference between the maximum and minimum values of the parameter in the history of the charging pile, to obtain a value between 0 and 1. After processing, each charging pile obtains a set of normalized data, including the normalized values of each vehicle's arrival time deviation, battery remaining capacity, battery total capacity and maximum charging power.

[0033] S1-2, federal learning framework construction:

[0034] In this step, a federated learning framework is constructed based on the local data of each charging pile. First, each charging pile trains a local model using the locally collected and normalized data. The local model adopts a multi-layer perceptron structure, which specifically includes: an input layer that receives 4 normalized features (i.e., the normalized values of arrival time deviation, battery remaining capacity, battery total capacity, and maximum charging power), two hidden layers each containing 32 neurons and using ReLU activation function, and an output layer adjusting the output content according to subsequent requirements. The model parameters are updated by gradient descent method, with a fixed learning rate of 0.001 and processing 32 data each time. After training, each charging pile calculates the parameter update amount of the local model and uploads it to the central server. To protect data privacy, noise is added to the parameter update amount before uploading. The method is as follows: first, calculate the Euclidean norm of the parameter update amount, then multiply it by 0.01, then multiply it by a random number drawn from the standard normal distribution, and add the result to the original parameter update amount. After receiving all the noisy parameter update amounts uploaded by the charging piles, the central server aggregates them by averaging, specifically: add the current model parameters of each charging pile to the noisy parameter update amount, and then take the average of all charging piles to obtain the global model parameters. The aggregated global model parameters are distributed back to each charging pile. Finally, each charging pile obtains the global model parameters for subsequent training.

[0035] S1-3, User behavior prediction model establishment:

[0036] A user behavior prediction model is constructed to predict the amount of electricity each vehicle needs to supplement in the future time period, referred to as the charging demand prediction value, with units of kilowatt-hours. The input of the model is the normalized arrival time deviation, battery remaining capacity, battery total capacity, and maximum charging power of each vehicle, and the output is the charging demand prediction value of each vehicle. The model adopts a long short-term memory network structure, which specifically includes: an input layer that receives 4 normalized parameters, a bidirectional long short-term memory network layer containing 64 hidden units, and an output layer that outputs the charging demand prediction value through a single neuron. During training, the loss function is defined as the sum of the negative exponential functions of the absolute values of the differences between the predicted values and the actual charging demands. The specific calculation method is as follows: for each vehicle, calculate the absolute value of the difference between the predicted value and the actual value, take its negative exponential, and then add all the values of all vehicles. The model is initialized using the global model parameters obtained by aggregation in federated learning, and the early stopping strategy is used during training, i.e., if the validation set loss does not decrease for 5 consecutive training periods, training is stopped. After training, each charging pile obtains the user behavior prediction model, which outputs the charging demand prediction value of each vehicle.

[0037] S1-4, Battery charging capacity model establishment:

[0038] A battery charging capability model is constructed to predict the maximum acceptable charging power for each vehicle in its current state. This is called the predicted charging capability value, expressed in kilowatts. The model inputs are each vehicle's normalized remaining battery charge, total battery capacity, and maximum charging power, and the output is the predicted charging capability value for each vehicle. The model employs support vector regression with a radial basis kernel function, kernel parameter set to 0.1, and regularization parameter set to 1.0. The loss function is defined as the sum of the squared reciprocals of the difference between the predicted and actual charging capabilities. The specific calculation method is: for each vehicle, the square of the difference between the predicted and actual values ​​is calculated, a constant of 0.01 is added to prevent the denominator from being zero, the reciprocal of this difference is taken, and the resulting value is summed across all vehicles. During training, the relevant parameters are initialized using the global model parameters aggregated in federated learning. The kernel and regularization parameters are optimized using a grid search method. After training, each charging station obtains a battery charging capability model and outputs a predicted charging capability value for each vehicle.

[0039] S1-5, model deployment and real-time prediction:

[0040] The user behavior prediction model and battery charging capacity model are deployed at each charging station for real-time prediction. The real-time inputs are the collected and normalized arrival time deviation, remaining battery charge, total battery capacity, and maximum charging power. The model outputs a predicted charging demand and charging capacity for each vehicle, respectively. Subsequently, a summary is performed for each charging station: the total charging demand is calculated as the sum of the predicted charging demand values ​​for all vehicles at that station; and the total charging capacity is calculated as the sum of the predicted charging capacity values ​​for all vehicles at that station, subject to the maximum output power limit of the station. Finally, each charging station generates the total charging demand and total charging capacity, which serve as input data for subsequent steps.

[0041] The federated learning module uses federated learning technology to build user behavior prediction models and battery charging capacity models using vehicle arrival time deviations and battery status parameters collected by charging stations. This generates charging demand and charging capacity predictions for each vehicle, and aggregates them into total charging demand and charging capacity for each charging station. This data will be used in subsequent steps to calculate power allocation baselines in conjunction with grid load data, ensuring efficient charging station scheduling based on dynamic user demand and vehicle status.

[0042] The optimization scheduling module includes the following:

[0043] S2-1, data preparation:

[0044] In the data preparation stage, the input data is first collected and preprocessed. The input data includes two categories: one is the charging demand prediction value and the charging capacity prediction value of each charging pile obtained, which respectively reflects the user's expected demand for power and the maximum power output capacity of the charging pile in the future time period; the other is the real-time collected power grid load data and power grid load target curve, which respectively represent the total load condition of the power grid at the current time and the expected future load level of the power grid. The preprocessing process first smooths the real-time power grid load, adopts the exponential smoothing method, and realizes it by weighted summation of the real-time load at the current time and the smoothed load at the previous time. The weight design is based on the smoothing factor and its complement, in order to reduce the interference of short-term fluctuations. The initial smoothed load takes the initial real-time load as the benchmark. Subsequently, the power grid load deviation is calculated, which is the difference between the smoothed real-time load and the target load.

[0045] S2-2, multi-objective optimization model construction:

[0046] In the construction of the multi-objective optimization model, the goal is to simultaneously maximize the peak load shifting effect of the power grid and optimize the satisfaction degree of user charging demand. The decision variable is defined as the power allocation baseline value of each charging pile at each time point. The model needs to meet several constraint conditions: first, the power allocation baseline value of each charging pile at any time point must be between zero and its charging capacity prediction value; second, the total power provided by each charging pile in the prediction time period, calculated by accumulating its power allocation baseline value at each time point, must not be lower than its charging demand prediction value; third, the total load of the power grid at any time point, composed of the smoothed real-time load and the sum of the power allocation baseline values of all charging piles at that time point, should control the deviation from the target load within the allowable range of power grid dispatching demand.

[0047] S2-3, optimization objective function:

[0048] To achieve the above optimization objectives, two objective functions are constructed. The first objective function is used to evaluate the peak load shifting effect of the power grid, adopting an exponential decay penalty form: for each time point, the absolute value of the difference between the smoothed real-time load and the sum of the power allocation baseline values of all charging piles relative to the target load is calculated, divided by a decay factor related to the average level of the power grid load, and then the exponential value with the natural constant as the base is taken. The sum of the exponential values of all time points is accumulated, and the goal is to minimize this sum. The second objective function is used to measure the satisfaction degree of user charging demand, adopting a surplus exponential form: for each charging pile, the ratio of the total power provided in the prediction time period to the charging demand prediction value is calculated, plus one, and then the natural logarithm is taken. The sum of the logarithmic values of all charging piles is accumulated, and the goal is to maximize this sum.

[0049] S2-4, solution and output:

[0050] In the solving stage, an improved particle swarm optimization algorithm is used to enhance the global search ability by introducing a dynamic inertia factor that decreases linearly with the number of iterations. The optimization process includes the following steps: First, a set of initial solutions that meet all the constraints is randomly generated, i.e., the power allocation baseline value of each charging pile at each time point. Then, multiple rounds of iterative optimization are performed, and the solution is updated and verified after each iteration to see if it meets the constraints. After optimization, a set of non-inferior solutions (Pareto front) is obtained, from which the solution with the optimal comprehensive index is selected as the final result. The comprehensive index is the weighted sum of the two objective functions, and the weights are determined according to the priority of grid scheduling requirements and user demand. The final output of the optimal solution is the power allocation baseline value of each charging pile at each time point, which is used for subsequent steps.

[0051] Through the above multi-objective optimization method, the optimization scheduling module combines the charging demand prediction value and charging capacity prediction value provided by the previous step with real-time grid load data to generate the power allocation baseline value of each charging pile at each time point. This result optimizes the peak load shifting effect of the grid while meeting the user's charging demand, providing accurate data support for subsequent real-time adjustment.

[0052] In the intelligent charging pile remote control and scheduling method, the previous step generates the power allocation baseline value of each charging pile at each time point through a multi-objective optimization algorithm. The power allocation baseline value optimizes the peak load shifting effect of the grid while meeting the user's charging demand. The dynamic game module builds a dynamic game model between users and the grid using reinforcement learning algorithms to adjust the output power of the charging pile in real time. By calculating the user behavior deviation index and the grid load relief index, the collaborative effect of the two is integrated to dynamically adjust the integral incentive strategy and guide the user behavior deviation to achieve dynamic balance between the user's actual charging demand and the grid load target.

[0053] The dynamic game module includes the following:

[0054] S3-1, the user behavior deviation index is used to measure the deviation degree of the actual charging behavior of each vehicle at a specific time point from the predicted behavior. First, the actual charging power of each vehicle at the time point is monitored in real time, and the predicted charging power is obtained from the user behavior prediction model. Then, the relative deviation of the actual charging power and the predicted charging power is calculated, and the specific method is: the absolute value of the difference between the two is divided by the sum of the predicted charging power and a small constant to avoid zero denominator. Next, in order to reduce the interference of short-term fluctuations, a time window is introduced, and the cumulative average of the deviation in the window is calculated. Finally, the cumulative deviation is converted into the user behavior deviation index through nonlinear mapping, and the mapping method is: 1 minus the negative value of the base of the natural constant, with the product of the deviation and a control parameter as the exponent. The value of the index is limited between 0 and 1, and the larger the value, the more significant the user behavior deviation. For example, the logic for obtaining the user behavior deviation index can be:

[0055] The user behavior deviation information includes the user behavior deviation index. The user behavior deviation index is used to measure the deviation degree of the actual charging behavior of the i-th vehicle at time t from the predicted behavior. The calculation process is as follows: first, the actual charging power of the i-th vehicle at time t is monitored in real time , and the predicted charging power is obtained from the user behavior prediction model of the federated learning module . Then, the relative deviation of the actual and predicted charging power is calculated, and the formula is:

[0056]

[0057] wherein, is a small constant to prevent the denominator from being zero. Next, in order to reduce the interference of short-term fluctuations, a time window is introduced, and the cumulative average of the deviation in the window is calculated:

[0058]

[0059] Finally, the cumulative deviation is converted into the user behavior deviation index through nonlinear mapping:

[0060]

[0061] wherein, is a parameter to control the sensitivity of the deviation, and the index value range is [0, 1], and the larger the value, the more significant the user behavior deviation. This method dynamically captures the subtle changes in user behavior through time window smoothing and exponential transformation.

[0062] ​​​​S3-2, grid load relief index is used to measure the relief effect of the current charging pile power adjustment on the grid load target. First, the total grid load at this time point is monitored in real time, and the grid load target is obtained from the previous step. Then, the difference between the actual load and the target load is calculated. Next, a historical time window is introduced, and the average value of the grid load relief trend in this window is calculated, which is: the difference between the target load and the actual load at each historical time point is divided by the target load, and then averaged. Finally, the average value is converted to a grid load relief index using a Sigmoid function, and the parameters of the Sigmoid function include a control curve steepness and a relief threshold, and the index value range is between 0 and 1, and the larger the value, the better the relief effect. For example, the logic for obtaining the grid load relief index can be:

[0063] Where the grid load relief information includes the user behavior offset index. The grid load relief index is used to measure the relief effect of the current charging pile power adjustment on the grid load target. The calculation process is as follows: first, the total grid load at time is monitored in real time , and the grid load target is obtained from the optimization scheduling module . Then, the difference between the actual load and the target load is calculated:

[0064]

[0065] Next, introduce the load relief factor , evaluate the relief trend through historical data:

[0066]

[0067] Where, is a time variable, representing the historical time point moving backward from the current time point , and is the historical time window. Finally, the relief effect is quantified using a Sigmoid function:

[0068]

[0069] Where, controls the steepness of the curve, is the relief threshold, and the index value range is [0, 1], and the larger the value, the better the relief effect. This method combines historical trends and nonlinear mapping to improve the response ability of the index to dynamic changes in the grid.

[0070] S3-3, first, calculate the average value of the user behavior deviation index of all vehicles at that time point. Then, use the geometric mean method to combine the average value with the inverse value of the grid load relief index, that is, 1 minus the grid load relief index, to obtain the user-grid response synergy index. For example, the calculation of the user-grid response synergy index is:

[0071] User-grid response coordination index It is used to comprehensively evaluate the synergistic effect of user behavior deviation and grid load relief. The calculation process is as follows: First, calculate the time of all vehicles Average user behavior deviation index:

[0072]

[0073] in, is the total number of charging vehicles at present. Then, the geometric mean is used to integrate user behavior deviation and grid load relief:

[0074]

[0075] The geometric mean emphasizes the balance between the two. The smaller the value, the better the synergy effect. This method highlights the complementary relationship between user behavior and grid goals through geometric form.

[0076] S3-4 adopts a multi-agent reinforcement learning framework, treating each charging pile as an intelligent agent and the power grid as part of the environment. The state space includes the grid load at the current time point, the power distribution baseline value, the actual output power of each charging pile, and the user-grid response synergy index. The action space is the action amount, which is the power increase or decrease relative to the power distribution baseline value, and the amplitude is limited by the adjustable range of a single pile. The reward function is designed to be the inverse of the weighted sum of the user-grid response synergy index and the grid load deviation, and the weight is used to balance the importance of the two. The training adopts the deep Q network algorithm, and the learning stability is improved through the experience replay mechanism. For example, the reinforcement learning dynamic game model can be constructed as:

[0077] Using the multi-agent reinforcement learning (MARL) framework, each charging pile As an intelligent agent, the power grid is part of the environment. The state space includes time Grid load , power allocation baseline value , the actual output power of each charging pile and user-grid response synergy index The reward function is designed as:

[0078]

[0079] in, and are weights, respectively emphasizing the synergy effect and the load deviation. The training adopts a deep Q network (DQN) algorithm to improve learning stability through an experience replay mechanism.

[0080] S3-5, if the actual output power continuously deviates from the power allocation baseline value and the user-grid response synergy index exceeds the preset threshold, it indicates that there is a significant conflict between the user behavior and the grid load target, and the user needs to adjust the charging behavior through integral incentive. The incentive intensity is dynamically adjusted according to the difference between the synergy index and the threshold, and the specific method is: the basic incentive intensity is multiplied by an amplification factor based on the difference. The incentive form is integral reward or punishment, encouraging users to charge during the grid valley period. Dynamically adjust the integral incentive:

[0081] Incentive intensity is calculated as:

[0082]

[0083] wherein, is the basic incentive intensity. The incentive form is integral reward or punishment, encouraging users to charge during the grid valley period. This method dynamically adjusts the incentive intensity through the synergy index, ensuring the pertinence of the guidance effect.

[0084] The dynamic game module constructs a dynamic game model between users and the grid through reinforcement learning algorithm, calculates the user behavior deviation index and the grid load relief index based on the power allocation baseline value of the previous step, and obtains the user-grid response synergy index through geometric average. According to the comparison between the index and the preset threshold, the strength of the integral incentive strategy is dynamically adjusted, and the charging pile output power is updated in real time, realizing the dynamic balance between user demand and grid load and improving the adaptability and efficiency of the intelligent charging pile scheduling method.

[0085] The dynamic game module constructs a dynamic game model between users and the grid through reinforcement learning algorithm, calculates the user behavior deviation index and the grid load relief index based on the power allocation baseline value of the previous step, and obtains the user-grid response synergy index through geometric average. According to the comparison between the index and the preset threshold, the strength of the integral incentive strategy is dynamically adjusted, and the charging pile output power is updated in real time, realizing the dynamic balance between user demand and grid load and improving the adaptability and efficiency of the intelligent charging pile scheduling method.

[0086] The simulation verification module includes the following contents:

[0087] S4-1, construct a digital twin model:

[0088] When building the digital twin model, first use the collected vehicle arrival time deviation, battery state parameters and real-time load data of the power grid to establish the digital twin model of the charging pile group through deep learning algorithm. Deep learning algorithm predicts the operation trend of charging pile group by analyzing historical data and real-time input. The input of the model includes the adjusted charging pile output power, battery state parameters, user charging demand prediction and power grid load data, and the output is the simulated charging pile operation state and power grid load change. In the processing process, multi-source data fusion technology is adopted to integrate charging pile sensor data, user behavior data and power grid monitoring data into a unified input data set to improve the accuracy of model simulation.

[0089] S4-2, simulation verification:

[0090] In the simulation verification stage, the digital twin model is used to simulate the whole link of the adjusted charging pile output power, and the operation state of the charging pile group and the load response of the power grid in the future time period are predicted. In order to ensure that the scheduling strategy can adapt to various situations, random disturbance is introduced in the simulation to simulate the unpredictability of user behavior and the dynamic fluctuation of power grid load. The simulation process generates simulated charging pile output power and simulated power grid load through model calculation, which serves as the basis data for subsequent comparison.

[0091] S4-3, data comparison:

[0092] In the data comparison link, first collect the actual operation data of the charging pile group in real time, including the actual output power of each charging pile and the actual load of the power grid. Then, through the dynamic time warping algorithm, the time series of simulation results and actual data are aligned, and the deviation between the two is calculated. Dynamic time warping algorithm finds the best matching path between simulation data and actual data by analyzing the dynamic changes of time series, quantifies the differences between the two in time dimension, and generates deviation results. This process avoids the comparison error caused by time misalignment, ensuring the accuracy of deviation calculation.

[0093] S4-4, optimization prompt:

[0094] In the optimization prompt stage, the power deviation threshold and the load deviation threshold are preset as the judgment criteria. When the power deviation or the grid load deviation of any charging pile exceeds the corresponding threshold, the optimization prompt signal is triggered. The generation of the optimization prompt is based on deviation analysis: if the charging pile power deviation exceeds the threshold, the relevant parameters of the user behavior prediction model or the battery charging capacity model are adjusted; if the grid load deviation exceeds the threshold, the constraint conditions of the multi-objective optimization algorithm or the calculation logic of the reinforcement learning reward function are optimized. In addition, the Bayesian optimization algorithm is used to automatically search for the optimal solution of parameter adjustment according to the deviation analysis results, and the best parameter combination is determined through repeated iterative calculation to realize the adaptive optimization of the scheduling strategy.

[0095] The simulation verification module performs full-link simulation verification on the charging pile group through digital twinning technology, generates simulation results using the adjusted charging pile output power, and compares them with actual operation data. The dynamic time warping algorithm is used to quantify the deviation of the charging pile output power and the grid load, and when the deviation exceeds the preset threshold, specific optimization prompts for the aforementioned steps are generated, and the Bayesian optimization algorithm is used to realize parameter adjustment. This process improves the accuracy of the intelligent charging pile scheduling strategy and the balancing ability of the grid load, forming a closed-loop control mechanism.

[0096] The above formulas are dimensionless numerical calculations, and the formulas are obtained by software simulation of a large amount of collected data to obtain a formula of the most recent real situation. The preset parameters in the formula are set by a person skilled in the art according to the actual situation.

[0097] It should be noted that the system of the present application can be deployed on the device itself to realize embedded application, or can be run on PC or other terminal with user interface, so as to meet various hardware environments and use requirements.

[0098] The above only describes some exemplary embodiments of the present application by way of illustration, and it is needless to say that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present application. Therefore, the above figures and descriptions are illustrative in nature and should not be understood as limiting the scope of protection of the claims of the present application.

[0099] It has to be noted that, in the present document, the terms "first", "second", etc. merely serve the purpose of differentiation and do not require or imply any kind of ordering or sequence of the entities or actions associated therewith. Furthermore, the terms "comprising", "containing", etc. are to be interpreted as non- exclusive in the sense that a process, method, article, or apparatus that comprises a list of elements is not necessarily limited to those elements, but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a", "contains... a", etc. does not, without further restriction, exclude the presence of additional identical elements in the process, method, article, or apparatus.

[0100] The above description is only specific embodiments of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present application, which should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. Multifunctional slot intelligent charging pile remote control and dispatching system, characterized by: include: Federated Learning Module: This module collects vehicle arrival time deviations and battery status parameters in real time, and builds user behavior prediction models and battery charging capacity models based on federated learning. Optimization and scheduling module: This module combines the output of the user behavior prediction model with real-time grid load data to generate a power allocation baseline value through a multi-objective optimization algorithm. Dynamic Game Module: Based on a reinforcement learning algorithm, this module builds a dynamic game model between users and the grid. It integrates user behavior deviation information and grid load relief information, adjusts the output power of charging piles in real time according to the power allocation baseline value, and drives user charging behavior toward low-load periods based on a point incentive strategy. User behavior deviation information includes a user behavior deviation index. The user behavior deviation index is obtained by calculating the relative deviation between the actual charging power and the predicted charging power within the time window and performing nonlinear mapping. The value range is 0 to 1. The grid load relief information includes the grid load relief index, which is obtained by evaluating the average ratio of the difference between the real-time load and the load target in the historical window and converting it through the Sigmoid function. The value range is 0 to 1. The dynamic game module calculates the average value of the user behavior deviation index of all vehicles at the selected time point; the geometric mean method is used to combine the average value with the inverse value of the grid load relief index to obtain the user-grid response coordination index; The dynamic game module uses a multi-agent reinforcement learning framework, treating each charging station as an agent and the power grid as part of the environment. The state space includes the grid load at the current time point, the power allocation baseline value, the actual output power of each charging station, and the user-grid response coordination index. The action space is the power increase or decrease relative to the power allocation baseline, limited by the adjustable range of a single pile. The reward function is designed as the inverse of the weighted sum of the user-grid response synergy index and the grid load deviation. Training uses a deep Q-network algorithm, and an experience replay mechanism improves learning stability. If the actual output power continues to deviate from the power allocation baseline value and the user-grid response synergy index exceeds the preset threshold, the dynamic game module guides users to adjust their charging behavior through point incentives; Simulation verification module: Use digital twin technology to perform full-link simulation verification of the charging pile group, compare simulation data with actual operation data, and output optimization prompt signals.

2. The multifunctional slot intelligent charging pile remote control and dispatching system according to claim 1 is characterized by: The federated learning module collects vehicle arrival time deviations and battery status information at charging piles in real time and performs preprocessing. Based on federated learning technology, a local model is trained on each charging pile node and aggregated through a central server to generate a global model.

3. The multifunctional slot intelligent charging pile remote control and dispatching system according to claim 2 is characterized by: The federated learning module uses support vector regression to build a battery charging capacity model to predict the vehicle's charging capacity; finally, the trained model is deployed to each charging pile node to generate charging demand prediction values ​​and charging capacity prediction values ​​in real time.

4. The multifunctional slot intelligent charging pile remote control and dispatching system according to claim 3 is characterized by: The optimization scheduling module obtains the charging demand forecast value and the charging capacity forecast value, and collects the grid load data and the grid load target curve in real time; smoothes the real-time grid load and calculates the grid load deviation; constructs a multi-objective optimization model, using the power allocation baseline value of each charging pile at each time point as the decision variable, and simultaneously optimizes the grid peak shaving and valley filling and the satisfaction of user charging needs; sets constraints to ensure that the power allocation baseline value is within the capacity of the charging pile, meets the charging demand and controls the grid load deviation; designs the objective function, uses the exponential decay penalty form to evaluate the grid peak shaving and valley filling effect, and uses the margin index form to measure the satisfaction of user charging needs; uses the improved particle swarm optimization algorithm to obtain the optimal solution; and finally generates the power allocation baseline value of each charging pile at each time point.

5. The multifunctional slot intelligent charging pile remote control and dispatching system according to claim 1 is characterized by: The simulation verification module uses a dynamic time warping algorithm to align the simulation data with the actual data. If the power deviation or load deviation exceeds the corresponding preset threshold, an optimization prompt signal is generated.

Images