Charging pile intelligent management system based on big data

Abstract

The invention discloses a charging pile intelligent management system based on big data. The system comprises a data acquisition module, a data preprocessing module, a charging pile load prediction module, a charging pile controllable load adjustment module, a model hyper-parameter optimization module and a charging pile intelligent management module. The invention relates to the technical field of charging pile data processing, in particular to a charging pile intelligent management system based on big data, according to the scheme, a charging pile load prediction module and a charging pile controllable load adjustment module are innovatively combined, the overall power grid load condition is considered, and the level of charging pile intelligent management is improved; a charging pile load prediction activation function, a momentum updating strategy and a mechanism for dynamically adjusting learning factors are designed, so that the accuracy of a prediction result is improved; and through nonlinear sine chaotic mapping, adaptive individual genetic probability and superior product retention strategy, the global optimal solution search capability is enhanced, an optimal parameter combination is obtained, and the accuracy of a model result is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of charging pile data processing, and in particular to a charging pile intelligent management system based on big data. Background Art

[0002] The intelligent management system of charging piles based on big data is an effective means to cope with the rapid popularization of electric vehicles and fluctuations in power grid load. By integrating big data, machine learning and intelligent scheduling technology, it is possible to accurately predict and intelligently adjust the load of charging piles, thereby achieving multiple goals of power grid load balance, maximizing charging efficiency and minimizing charging costs, and providing support for the long-term sustainable development of the power grid; however, there are technical problems in the traditional charging pile management system that do not consider the load conditions of the power grid in the region, low management efficiency and inflexible charging power adjustment; the activation function in the existing prediction model for charging pile load has limitations under specific conditions, and the overfitting problem may occur during the training process, resulting in inaccurate output results of the prediction model; there are technical problems in the traditional prediction model that the parameters are improperly set, and the parameter optimization algorithm has a weak ability to obtain the global optimal solution, resulting in inaccurate final prediction results. Summary of the invention

[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a charging pile intelligent management system based on big data. In view of the technical problems that the traditional charging pile management system does not consider the regional power grid load situation, has low management efficiency and inflexible charging power adjustment, the present invention innovatively proposes a solution of combining the charging pile load prediction module with the charging pile controllable load regulation module. The system can consider the power grid load situation in real time and intelligently adjust the charging power of each charging pile, ensure the optimization of power distribution under different load conditions, effectively balance the regional power grid load, avoid power grid overload during peak hours, and reduce power grid pressure. It not only improves the management efficiency and operation performance of the charging piles, but also ensures the efficient operation of the charging piles, thereby improving the level of intelligent management of the entire charging pile and the user experience; in view of the existing load prediction model suitable for charging piles, the system can effectively adjust the charging power of each charging pile and optimize the power distribution under different load conditions, effectively balance the regional power grid load, avoid power grid overload during peak hours, and reduce power grid pressure. The activation function has limitations under specific conditions, and overfitting problems may occur during the training process, which leads to inaccurate output results of the prediction model. This solution innovatively designs the charging pile load prediction activation function, momentum update strategy and dynamic adjustment of learning factors, which improves the expressiveness and flexibility of the prediction model, enhances its generalization ability, and significantly improves the accuracy of the model, thereby achieving accurate prediction of the load of each charging pile; in view of the technical problems of improper parameter settings in traditional models and weak ability of parameter optimization algorithms to obtain global optimal solutions, resulting in inaccurate final results, this solution enhances the global optimal solution search and improves the robustness of the algorithm through nonlinear sinusoidal chaos mapping, adaptive individual genetic probability and superior product retention strategy, thereby obtaining the optimal parameter values ​​and solving the problem of inaccurate output results of the final adjustment model.

[0004] The technical solution adopted by the present invention is as follows: A charging pile intelligent management system based on big data provided by the present invention includes a data acquisition module, a data preprocessing module, a charging pile load prediction module, a charging pile controllable load adjustment module, a model hyperparameter optimization module and a charging pile intelligent management module;

[0005] The data acquisition module obtains the original data of intelligent management of charging piles by collecting data from the national power grid system;

[0006] The data preprocessing module performs data cleaning, standardization processing and feature selection on the charging pile intelligent management raw data to obtain the charging pile intelligent management preliminary data;

[0007] The charging pile load prediction module establishes a charging pile load prediction model, performs model training on the charging pile load prediction model, realizes real-time prediction of a single charging pile load, and obtains a real-time prediction result of a single charging pile load;

[0008] The controllable load regulation module of the charging pile determines the load regulation target and establishes a controllable load regulation model using a deep Q network algorithm, and performs model training on the controllable load regulation model, thereby obtaining a controllable load regulation strategy for a single charging pile, and obtaining a controllable load regulation strategy for a single charging pile;

[0009] The model hyperparameter optimization module obtains the optimal parameter combination of the controllable load regulation model through a seed genetic algorithm;

[0010] The charging pile intelligent management module realizes intelligent management and scheduling of charging piles in the area based on the output result of the charging pile controllable load adjustment module.

[0011] Furthermore, the data acquisition module specifically acquires the original data of intelligent management of charging piles from the national grid system by acquisition; the original data of intelligent management of charging piles includes historical charging pile data, real-time charging pile data, historical charging pile controllable adjustment data and real-time charging pile controllable adjustment data.

[0012] Furthermore, the data preprocessing module specifically performs data cleaning, standardization and feature selection on the original data of the intelligent management of the charging pile to obtain preliminary data of the intelligent management of the charging pile; the data cleaning is to process missing values, abnormal values ​​and duplicate values; the standardization is to standardize the data based on the maximum and minimum normalization method; the feature selection is to use the correlation analysis method to screen out the relevant features with the charging pile load prediction and the controllable load regulation of the charging pile.

[0013] Furthermore, the charging pile load prediction module includes establishing a charging pile load prediction model, training the charging pile load prediction model, and real-time prediction of a single charging pile load; specifically includes the following steps:

[0014] Establishing a charging pile load prediction model includes the following steps:

[0015] Design the charging pile load prediction activation function, specifically design the prediction model input activation function and the prediction model output activation function. The formula used is as follows:

[0016] ;

[0017] ;

[0018] In the formula, represents the prediction model input activation function, represents the output activation function of the prediction model, x represents the input variable of the activation function, represents the parameter of the prediction model input activation function controlling the nonlinear shape, It represents the parameter of the output activation function of the prediction model to control the nonlinear shape, and p represents the dynamic adjustment parameter;

[0019] Initialize the internal structure MT of the prediction model, which includes the following steps:

[0020] Input data processing, the formula used is as follows:

[0021] ;

[0022] In the formula, Represents the output value after input data processing, Represents the weight matrix corresponding to the input data processing, represents the bias parameter for input data processing, It indicates input data. Indicates the hidden state at the last moment;

[0023] The historical information is forgotten and the formula used is as follows:

[0024] ;

[0025] In the formula, Represents the output after the historical information is forgotten. Represents the weight matrix corresponding to the historical information forgetting process, The bias parameter representing the forgetting process of historical information;

[0026] Output data calculation, the formula used is as follows:

[0027] ;

[0028] In the formula, Represents the result of the output data calculation, represents the weight matrix corresponding to the output data calculation, Represents the bias parameter for output data calculation;

[0029] The status update candidate value calculation uses the following formula:

[0030] ;

[0031] In the formula, Indicates the new state update candidate value, represents the weight matrix corresponding to the calculation of the state update candidate value, Represents the bias parameter for calculating the candidate value for state update;

[0032] The current status is obtained using the following formula:

[0033] ;

[0034] In the formula, c k Indicates the current state, c k-1 Indicates the state at the previous moment. represents element-wise multiplication;

[0035] Get the current hidden state. The formula used is as follows:

[0036] ;

[0037] In the formula, h k Indicates the hidden state at the current moment;

[0038] The final output result is obtained using the following formula:

[0039] ;

[0040] In the formula, represents the charging pile load prediction result at the kth moment, represents the weight matrix corresponding to the output, Represents the bias parameter of the output layer;

[0041] The charging pile load prediction model training is specifically to use the historical charging pile data in the preliminary data of the charging pile intelligent management as input data to perform model parameter update training on the charging pile load prediction model to obtain the charging pile load prediction model after the model parameters are updated; comprising the following steps:

[0042] Design the prediction model loss function, the formula used is as follows:

[0043] ;

[0044] Where N represents the total number of time steps of the output result. represents the actual result of the charging pile load at the kth moment, represents the charging pile load prediction result at the k+2th moment, represents the charging pile load prediction result at the k+1th moment, represents the prediction model loss function, represents the weight parameter of the smoothness loss, represents the weight parameter of the weighted loss, represents the weighting coefficient at the kth moment;

[0045] The prediction model reverse parameter update is specifically to update the parameters of the internal structure MT parameters of the prediction model. The internal structure MT parameters of the prediction model include: , , , , , , and ; including the following steps:

[0046] Calculate the current gradient, specifically the gradient of the loss function with respect to the parameters. The formula used is as follows:

[0047] ;

[0048] In the formula, represents the gradient vector at the current moment, Indicates the parameters The partial derivative of Represents the model parameters at the previous moment;

[0049] To update momentum, the formula used is as follows:

[0050] ;

[0051] ;

[0052] ;

[0053] In the formula, represents the first-order momentum update value at the current moment, represents the second-order momentum update value at the current moment, represents the infinite norm momentum update value at the current moment, represents the infinite norm momentum update value at the previous moment, represents the decay rate of the first-order momentum, represents the decay rate of the second-order momentum, represents the first-order momentum update value at the previous moment, Represents the second-order momentum update value at the previous moment;

[0054] The mixed momentum is calculated using the following formula:

[0055] ;

[0056] In the formula, Represents the hybrid momentum value at the current moment, which is used to accelerate convergence and stabilize parameter updates. represents the cumulative decay of all first-order momentum at the current time t;

[0057] The dynamic adjustment learning factor is calculated using the following formula:

[0058] ;

[0059] In the formula, represents the dynamic adjustment of learning factor, t represents the number of time steps, T represents the total number of time steps, represents the cumulative decay of all second-order momentum at the current time t;

[0060] Parameter optimization update, the formula used is as follows:

[0061] ;

[0062] In the formula, Indicates the current parameter value. Indicates the parameter value at the previous moment. Represents the learning rate, which is used to control the step size of parameter updates;

[0063] Iteration judgment, specifically monitoring the change of the loss function and the current number of iterations. If the loss function is lower than the set threshold, the iterative training is stopped, otherwise the iterative training is continued. If the maximum number of iterations is reached, the iterative training is stopped, otherwise the iterative training is continued, and finally the charging pile load prediction model with updated model parameters is obtained;

[0064] The real-time prediction of the load of a single charging pile is specifically to use the real-time charging pile data in the preliminary data of the intelligent management of the charging pile as the input data of the charging pile load prediction model after the model parameters are updated, to obtain the real-time prediction result of the load of a single charging pile, and the real-time prediction result of the load of a single charging pile includes the prediction time and load power.

[0065] Furthermore, the controllable load regulation module of the charging pile includes determining the load regulation target, establishing a controllable load regulation model, training the controllable load regulation model and obtaining a controllable load regulation strategy for a single charging pile; specifically, the following steps are included:

[0066] Determine the controllable load adjustment target of charging piles, specifically to achieve grid load balance, charging pile load scheduling, charging demand maximization and overall energy efficiency maximization;

[0067] A controllable load regulation model is established, specifically using a deep Q network algorithm to establish the model, including the following steps:

[0068] Define the state using the following formula:

[0069] ;

[0070] In the formula, Indicates the current state. Indicates the current total load of the power grid. represents the predicted load of the i-th charging pile, represents the power demanded by the user of the i-th charging pile, represents the load forecast value of the power grid at time t, represents the health status of the power grid at time t, represents the environmental data at time t, represents the charging price of the i-th charging pile;

[0071] Define the action space, the formula used is as follows:

[0072] ;

[0073] In the formula, Indicates the charging power limit. Indicates charging power delay, Indicates that the charging power is increased. It means lowering the electricity price of charging piles. represents the increase of charging pile electricity price, and A represents the action space;

[0074] Design the reward function, the formula used is as follows:

[0075] ;

[0076] In the formula, represents the current reward function value, Indicates that the charging demand meets the reward. represents the overall energy efficiency reward, , , Represents the weight parameters controlling each target;

[0077] Interaction and storage experience, depending on the current state , using the ε-greedy strategy to select actions , perform the action , record the next state and rewards ; Experience Deposit into the experience replay pool;

[0078] To calculate the target value, the formula used is as follows:

[0079] ;

[0080] In the formula, represents the target value, represents the discount factor, represents the target network, Indicates possible action at the next moment. represents the target network parameters;

[0081] Training the controllable load regulation model, specifically using the historical charging pile controllable regulation data in the preliminary data of the charging pile intelligent management to train the controllable load regulation model, to obtain a trained controllable load regulation model;

[0082] A controllable load regulation strategy for a single charging pile is obtained, specifically, the real-time controllable regulation data of the charging pile in the preliminary data of intelligent management of the charging pile and the real-time prediction result of the load of the single charging pile are used as input data of the trained controllable load regulation model to obtain the controllable load regulation strategy for a single charging pile.

[0083] Furthermore, the model hyperparameter optimization module specifically obtains the optimal parameter combination of the controllable load regulation model through a seed genetic algorithm; and includes the following steps:

[0084] Initialization parameters, specifically, by constructing the initial algorithm parameters; the initial algorithm parameters include population size N, initial crossover probability interval, initial mutation probability interval and maximum number of iterations ;

[0085] Initializing the population, specifically, initializing the population by generating chaotic sequence values ​​through nonlinear sinusoidal chaotic mapping, including the following steps:

[0086] Generate N chaotic sequence values ​​and initialize the random value z 0 ; The formula used is as follows:

[0087] ;

[0088] In the formula, represents the i-th chaotic sequence value, represents the i+1th chaotic sequence value, Represents adjustment parameters, which are used to adjust the mapping behavior. Indicates a parameter to avoid the denominator being 0;

[0089] The chaotic sequence value transforms the seed genetic individual, and the formula is expressed as follows:

[0090] ;

[0091] In the formula, represents the initial genetic individual gene of the i-th individual, and represent the lower and upper genetic limits of the population, respectively;

[0092] Calculate the fitness value, specifically calculate the individual fitness value f in the population i ; The performance of the controllable load regulation model based on the individual position is used as the individual fitness value;

[0093] The superior product retention strategy ranks the fitness values, selects the top 5% of genetic individuals with the highest fitness values, and inserts them into the next generation population. No individual gene crossover operation or individual genetic mutation operation is performed. During each iteration, the size of the population is guaranteed to be the same as the number of genetic individuals in the initial population.

[0094] The selection operation is to sort the calculated fitness values ​​and select 80% of the genetic individuals for individual gene crossover operation and individual genetic mutation operation according to the roulette wheel selection probability; the formula used is as follows:

[0095] ;

[0096] In the formula, represents the probability of roulette selection for the i-th individual, represents the fitness function of the individual;

[0097] Calculate the adaptive individual genetic probability, specifically calculate the individual gene crossover probability and individual mutation probability, the formula used is as follows:

[0098] ;

[0099] ;

[0100] In the formula, represents the latest value of the crossover probability, represents the latest value of the mutation probability, and represents the maximum and minimum crossover probability, and represents the maximum and minimum values ​​of mutation probability, represents the current value of the crossover probability, represents the current value of the mutation probability, represents the average value of the population fitness, The maximum value of the population fitness;

[0101] Individual gene crossover operation, specifically if , then two random parent individuals are randomly selected from the population for gene crossover to generate two offspring individuals; the formula used is as follows:

[0102] ;

[0103] ;

[0104] In the formula, and They represent the individual genes inherited by the two offspring after gene crossover. and Represent the gene values ​​of two parent individuals, represents a random number in the range [0,1], c represents the intersection of the gene, k represents the length of the gene, and j represents the index of the gene;

[0105] Individual genetic variation operations are performed using the following formula:

[0106] ;

[0107] In the formula, Indicates the mutated genetic individual genes in the t+1 generation, represents the genetic individual gene with the highest fitness value in the tth generation, represents a random number in the range [0,1], represents the Cauchy probability distribution;

[0108] Genetic determination, specifically, by constructing genetic termination conditions, performing genetic determination on the optimal individual with the highest fitness value, and obtaining the genetic individual with the optimal fitness value in the genetic process;

[0109] The genetic termination conditions include threshold termination and iteration termination;

[0110] The threshold termination is specifically to set the fitness threshold, when the individual fitness value f i When the fitness threshold is exceeded, inheritance is completed;

[0111] The iteration termination specifically refers to terminating the iteration and obtaining the optimal genetic individual gene when the maximum number of iterations is reached;

[0112] The optimal genetic individual gene is specifically a controllable load regulation model parameter combination.

[0113] Furthermore, the charging pile intelligent management module specifically dynamically adjusts the charging power and charging price of each charging pile according to the controllable load adjustment strategy of the single charging pile, so as to realize intelligent management and scheduling of charging piles in the area.

[0114] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0115] (1) In view of the technical problems of traditional charging pile management systems, such as not considering the load conditions of the power grid in the region, low management efficiency and inflexible charging power adjustment, this solution innovatively proposes a solution that combines the charging pile load prediction module with the charging pile controllable load adjustment module. The system can consider the power grid load conditions in real time and intelligently adjust the charging power of each charging pile to ensure optimal power distribution under different load conditions, effectively balance the regional power grid load, avoid power grid overload during peak hours, and reduce power grid pressure. It not only improves the management efficiency and operating performance of the charging piles, but also ensures the efficient operation of the charging piles, thereby improving the level of intelligent management of the entire charging pile and user experience.

[0116] (2) In view of the limitations of the activation function in the existing charging pile load prediction model under specific conditions and the overfitting problem that may occur during the training process, which leads to inaccurate output results of the prediction model, this scheme innovatively designs the charging pile load prediction activation function, momentum update strategy and dynamic adjustment mechanism of learning factors, which improves the expressiveness and flexibility of the prediction model, enhances its generalization ability, and significantly improves the accuracy of the model, thereby achieving accurate prediction of the load of each charging pile.

[0117] (3) In order to solve the technical problems of improper parameter settings in traditional models and weak ability of parameter optimization algorithms to obtain global optimal solutions, which lead to inaccurate final results, this scheme uses nonlinear sinusoidal chaos mapping, adaptive individual genetic probability and superior product retention strategy to enhance the global optimal solution search and improve the robustness of the algorithm, thereby obtaining the optimal parameter values ​​and solving the problem of inaccurate output results of the final adjustment model. BRIEF DESCRIPTION OF THE DRAWINGS

[0118] Figure 1 A schematic diagram of a module of a charging pile intelligent management system based on big data provided by the present invention;

[0119] Figure 2 This is a flow chart of the charging pile load prediction module;

[0120] Figure 3 It is a flow chart of the controllable load adjustment module of the charging pile;

[0121] Figure 4 This is a flowchart of the model hyperparameter optimization module;

[0122] Figure 5 A schematic diagram of the process of establishing a charging pile load prediction model in the charging pile load prediction module;

[0123] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0124] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0125] In the description of the present invention, it is necessary to understand that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred system or element must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as a limitation on the present invention.

[0126] Example 1, see Figure 1 The present invention provides a charging pile intelligent management system based on big data, including a data acquisition module, a data preprocessing module, a charging pile load prediction module, a charging pile controllable load adjustment module, a model hyperparameter optimization module and a charging pile intelligent management module;

[0127] The data acquisition module collects historical charging pile data, real-time charging pile data, historical charging pile controllable adjustment data and real-time charging pile controllable adjustment data from the national power grid system, obtains charging pile intelligent management raw data, and sends the data to the data preprocessing module;

[0128] The data preprocessing module receives the data sent by the data acquisition module, performs data cleaning, standardization processing and feature selection on the charging pile intelligent management raw data, obtains the charging pile intelligent management preliminary data, and sends the data to the charging pile load prediction module and the charging pile controllable load adjustment module;

[0129] The charging pile load prediction module receives the data sent by the data preprocessing module, establishes a charging pile load prediction model, performs model training on the charging pile load prediction model, realizes real-time prediction of a single charging pile load, obtains a real-time prediction result of a single charging pile load, and sends the data to the charging pile controllable load adjustment module;

[0130] The charging pile controllable load regulation module receives data sent by the data preprocessing module, the charging pile load prediction module and the model hyperparameter optimization module, determines the load regulation target and establishes a controllable load regulation model using a deep Q network algorithm, performs model training on the controllable load regulation model, thereby obtaining a single charging pile controllable load regulation strategy, and sends the data to the charging pile intelligent management module;

[0131] The model hyperparameter optimization module obtains the optimal parameter combination of the controllable load regulation model through a seed genetic algorithm, and sends the data to the charging pile controllable load regulation module;

[0132] The charging pile intelligent management module receives data from the charging pile controllable load adjustment module, and based on the output result of the charging pile controllable load adjustment module, realizes intelligent management and scheduling of charging piles in the area.

[0133] By performing the above operations, this solution innovatively proposes a solution that combines the charging pile load prediction module with the charging pile controllable load regulation module to address the technical problems of traditional charging pile management systems that do not consider the grid load conditions in the region, have low management efficiency, and inflexible charging power adjustment. The system can consider the grid load conditions in real time and intelligently adjust the charging power of each charging pile, ensuring optimized power distribution under different load conditions, effectively balancing the regional grid load, avoiding grid overload during peak hours, and reducing grid pressure. It not only improves the management efficiency and operating performance of the charging piles, but also ensures the efficient operation of the charging piles, thereby improving the level of intelligent management of the entire charging pile and user experience.

[0134] Example 2, see Figure 1 , this embodiment is based on the above embodiment, and the data acquisition module specifically obtains the original data of intelligent management of charging piles from the national grid system by acquisition; the original data of intelligent management of charging piles includes historical charging pile data, real-time charging pile data, historical charging pile controllable adjustment data and real-time charging pile controllable adjustment data; the historical charging pile data and the real-time charging pile data both include current data, voltage data, charging power, charging time data, charging pile status, charging mode, charging pile equipment health data, charging amount and charging pile environmental data; the historical charging pile data also includes charging pile load forecast data; the historical charging pile controllable adjustment data and the real-time charging pile controllable adjustment data both include the total load of the power grid, the predicted load of the power grid, the health status of the power grid, the charging pile electricity price, the required power of the charging pile and the environmental data, the historical charging pile controllable adjustment data also includes the charging pile controllable adjustment strategy, the charging pile controllable adjustment strategy includes charging power limitation, charging power delay, charging power increase, lowering the charging pile electricity price, and increasing the charging pile electricity price; the charging pile environmental data includes geographic location and weather data.

[0135] Example 3, see Figure 1 This embodiment is based on the above embodiment, and the data preprocessing module specifically performs data cleaning, standardization processing and feature selection on the original data of the intelligent management of the charging pile to obtain preliminary data of the intelligent management of the charging pile; the data cleaning is to process missing values, abnormal values ​​and duplicate values; the standardization processing is to standardize the data based on the maximum and minimum normalization method; the feature selection is to use the correlation analysis method to screen out the relevant features with the charging pile load prediction and the controllable load regulation of the charging pile.

[0136] Example 4, see Figure 1 , Figure 2 and Figure 5This embodiment is based on the above embodiment. The charging pile load prediction module includes establishing a charging pile load prediction model, training the charging pile load prediction model, and real-time prediction of a single charging pile load; specifically includes the following steps:

[0137] Establishing a charging pile load prediction model includes the following steps:

[0138] Design the charging pile load prediction activation function, specifically design the prediction model input activation function and the prediction model output activation function. The formula used is as follows:

[0139] ;

[0140] ;

[0141] In the formula, represents the prediction model input activation function, represents the output activation function of the prediction model, x represents the input variable of the activation function, represents the parameter of the prediction model input activation function controlling the nonlinear shape, It represents the parameter of the output activation function of the prediction model to control the nonlinear shape, and p represents the dynamic adjustment parameter;

[0142] Initialize the internal structure MT of the prediction model, which includes the following steps:

[0143] Input data processing, the formula used is as follows:

[0144] ;

[0145] In the formula, Represents the output value after input data processing, Represents the weight matrix corresponding to the input data processing, represents the bias parameter for input data processing, It indicates input data. Indicates the hidden state at the last moment;

[0146] The historical information is forgotten and the formula used is as follows:

[0147] ;

[0148] In the formula, Represents the output after the historical information is forgotten. Represents the weight matrix corresponding to the historical information forgetting process, The bias parameter representing the forgetting process of historical information;

[0149] Output data calculation, the formula used is as follows:

[0150] ;

[0151] In the formula, Represents the result of the output data calculation, represents the weight matrix corresponding to the output data calculation, Represents the bias parameter for output data calculation;

[0152] The status update candidate value calculation uses the following formula:

[0153] ;

[0154] In the formula, Indicates the new state update candidate value, represents the weight matrix corresponding to the calculation of the state update candidate value, Represents the bias parameter for calculating the candidate value for state update;

[0155] The current status is obtained using the following formula:

[0156] ;

[0157] In the formula, c k Indicates the current state, c k-1 Indicates the state at the previous moment. represents element-wise multiplication;

[0158] Get the current hidden state. The formula used is as follows:

[0159] ;

[0160] In the formula, h k Indicates the hidden state at the current moment;

[0161] The final output result is obtained using the following formula:

[0162] ;

[0163] In the formula, represents the charging pile load prediction result at the kth moment, represents the weight matrix corresponding to the output, Represents the bias parameter of the output layer;

[0164] The charging pile load prediction model training is specifically to use the historical charging pile data in the preliminary data of the charging pile intelligent management as input data to perform model parameter update training on the charging pile load prediction model to obtain the charging pile load prediction model after the model parameters are updated; comprising the following steps:

[0165] Design the prediction model loss function, the formula used is as follows:

[0166] ;

[0167] Where N represents the total number of time steps of the output result. represents the actual result of the charging pile load at the kth moment, represents the charging pile load prediction result at the k+2th moment, represents the charging pile load prediction result at the k+1th moment, represents the prediction model loss function, represents the weight parameter of the smoothness loss, represents the weight parameter of the weighted loss, represents the weighting coefficient at the kth moment;

[0168] The prediction model reverse parameter update is specifically to update the parameters of the internal structure MT parameters of the prediction model. The internal structure MT parameters of the prediction model include: , , , , , , and ; including the following steps:

[0169] Calculate the current gradient, specifically the gradient of the loss function with respect to the parameters. The formula used is as follows:

[0170] ;

[0171] In the formula, represents the gradient vector at the current moment, Indicates the parameters The partial derivative of Represents the model parameters at the previous moment;

[0172] To update momentum, the formula used is as follows:

[0173] ;

[0174] ;

[0175] ;

[0176] In the formula, represents the first-order momentum update value at the current moment, represents the second-order momentum update value at the current moment, represents the infinite norm momentum update value at the current moment, represents the infinite norm momentum update value at the previous moment, represents the decay rate of the first-order momentum, represents the decay rate of the second-order momentum, represents the first-order momentum update value at the previous moment, Represents the second-order momentum update value at the previous moment;

[0177] The mixed momentum is calculated using the following formula:

[0178] ;

[0179] In the formula, Represents the hybrid momentum value at the current moment, which is used to accelerate convergence and stabilize parameter updates. represents the cumulative decay of all first-order momentum at the current time t;

[0180] The dynamic adjustment learning factor is calculated using the following formula:

[0181] ;

[0182] In the formula, represents the dynamic adjustment of learning factor, t represents the number of time steps, T represents the total number of time steps, represents the cumulative decay of all second-order momentum at the current time t;

[0183] Parameter optimization update, the formula used is as follows:

[0184] ;

[0185] In the formula, Indicates the current parameter value. Indicates the parameter value at the previous moment. Represents the learning rate, which is used to control the step size of parameter updates;

[0186] Iteration judgment, specifically monitoring the change of the loss function and the current number of iterations. If the loss function is lower than the set threshold, the iterative training is stopped, otherwise the iterative training is continued. If the maximum number of iterations is reached, the iterative training is stopped, otherwise the iterative training is continued, and finally the charging pile load prediction model with updated model parameters is obtained;

[0187] The real-time prediction of the load of a single charging pile is specifically to use the real-time charging pile data in the preliminary data of the intelligent management of the charging pile as the input data of the charging pile load prediction model after the model parameters are updated, to obtain the real-time prediction result of the load of a single charging pile, and the real-time prediction result of the load of a single charging pile includes the prediction time and load power.

[0188] By performing the above operations, this solution innovatively designs charging pile load prediction activation functions, momentum update strategies, and a mechanism for dynamically adjusting learning factors to address the limitations of the activation functions in existing charging pile load prediction models under specific conditions and the overfitting problem that may occur during the training process, which leads to inaccurate output results of the prediction model. This improves the expressiveness and flexibility of the prediction model, enhances its generalization ability, and significantly improves the accuracy of the model, thereby achieving accurate prediction of the load of each charging pile.

[0189] Example 5, see Figure 1 and Figure 3 This embodiment is based on the above embodiment. The controllable load adjustment module of the charging pile includes determining the load adjustment target, establishing a controllable load adjustment model, training the controllable load adjustment model and obtaining a controllable load adjustment strategy for a single charging pile. Specifically, the following steps are included:

[0190] Determine the controllable load adjustment target of charging piles, specifically to achieve grid load balance, charging pile load scheduling, charging demand maximization and overall energy efficiency maximization;

[0191] The grid load balancing is specifically to avoid grid overload and keep the load within a reasonable range;

[0192] The charging pile load scheduling is specifically to ensure that each charging pile completes charging within a reasonable time and adjust the charging speed and power;

[0193] Maximizing charging demand specifically means meeting the charging needs of charging pile users and reducing user waiting time;

[0194] Maximizing overall energy efficiency specifically involves improving the energy efficiency of the power grid and charging stations and reducing losses;

[0195] A controllable load regulation model is established, specifically using a deep Q network algorithm to establish the model. This model takes into account the load demand of a single charging pile while coordinating the load balance between multiple charging piles and the power grid; the following steps are included:

[0196] Define the state using the following formula:

[0197] ;

[0198] In the formula, Indicates the current state. Indicates the current total load of the power grid. represents the predicted load of the i-th charging pile, represents the power demanded by the user of the i-th charging pile, represents the load forecast value of the power grid at time t, represents the health status of the power grid at time t, represents the environmental data at time t, represents the charging price of the i-th charging pile;

[0199] Define the action space, the formula used is as follows:

[0200] ;

[0201] In the formula, Indicates the charging power limit. Indicates charging power delay, Indicates that the charging power is increased. It means lowering the electricity price of charging piles. represents the increase of charging pile electricity price, and A represents the action space;

[0202] Design the reward function, the formula used is as follows:

[0203] ;

[0204] In the formula, represents the current reward function value, Indicates that the charging demand meets the reward. represents the overall energy efficiency reward, , , Represents the weight parameters controlling each target;

[0205] Interaction and storage experience, depending on the current state , using the ε-greedy strategy to select actions , perform the action , record the next state and rewards ; Experience Deposit into the experience replay pool;

[0206] To calculate the target value, the formula used is as follows:

[0207] ;

[0208] In the formula, represents the target value, represents the discount factor, represents the target network, Indicates possible action at the next moment. represents the target network parameters;

[0209] Training the controllable load regulation model, specifically using the historical charging pile controllable regulation data in the preliminary data of the charging pile intelligent management to train the controllable load regulation model, to obtain a trained controllable load regulation model;

[0210] A controllable load regulation strategy for a single charging pile is obtained, specifically, the real-time controllable regulation data of the charging pile in the preliminary data of intelligent management of the charging pile and the real-time prediction result of the load of the single charging pile are used as input data of the trained controllable load regulation model to obtain the controllable load regulation strategy for a single charging pile.

[0211] Example 6, see Figure 1 and Figure 4 This embodiment is based on the above embodiment, and the model hyperparameter optimization module specifically obtains the optimal parameter combination of the controllable load regulation model through a seed genetic algorithm; and includes the following steps:

[0212] Initialization parameters, specifically, by constructing the initial algorithm parameters; the initial algorithm parameters include population size N, initial crossover probability interval, initial mutation probability interval and maximum number of iterations ;

[0213] Initializing the population, specifically, initializing the population by generating chaotic sequence values ​​through nonlinear sinusoidal chaotic mapping, including the following steps:

[0214] Generate N chaotic sequence values ​​and initialize the random value z 0 ; The formula used is as follows:

[0215] ;

[0216] In the formula, represents the i-th chaotic sequence value, represents the i+1th chaotic sequence value, Represents adjustment parameters, which are used to adjust the mapping behavior. Indicates a parameter to avoid the denominator being 0;

[0217] The chaotic sequence value transforms the seed genetic individual, and the formula is expressed as follows:

[0218] ;

[0219] In the formula, represents the initial genetic individual gene of the i-th individual, and represent the lower and upper genetic limits of the population, respectively;

[0220] Calculate the fitness value, specifically calculate the individual fitness value f in the population i ; The performance of the controllable load regulation model based on the individual position is used as the individual fitness value;

[0221] The superior product retention strategy ranks the fitness values, selects the top 5% of genetic individuals with the highest fitness values, and inserts them into the next generation population. No individual gene crossover operation or individual genetic mutation operation is performed. During each iteration, the size of the population is guaranteed to be the same as the number of genetic individuals in the initial population.

[0222] The selection operation is to sort the calculated fitness values ​​and select 80% of the genetic individuals for individual gene crossover operation and individual genetic mutation operation according to the roulette wheel selection probability; the formula used is as follows:

[0223] ;

[0224] In the formula, represents the probability of roulette selection for the i-th individual, represents the fitness function of the individual;

[0225] Calculate the adaptive individual genetic probability, specifically calculate the individual gene crossover probability and individual mutation probability, the formula used is as follows:

[0226] ;

[0227] ;

[0228] In the formula, represents the latest value of the crossover probability, represents the latest value of the mutation probability, and represents the maximum and minimum crossover probability, and represents the maximum and minimum values ​​of mutation probability, represents the current value of the crossover probability, represents the current value of the mutation probability, represents the average value of the population fitness, The maximum value of the population fitness;

[0229] Individual gene crossover operation, specifically if , then two random parent individuals are randomly selected from the population for gene crossover to generate two offspring individuals; the formula used is as follows:

[0230] ;

[0231] ;

[0232] In the formula, and They represent the individual genes inherited by the two offspring after gene crossover. and Represent the gene values ​​of two parent individuals, represents a random number in the range [0,1], c represents the intersection of the gene, k represents the length of the gene, and j represents the index of the gene;

[0233] Individual genetic variation operations are performed using the following formula:

[0234] ;

[0235] In the formula, Indicates the mutated genetic individual genes in the t+1 generation, represents the genetic individual gene with the highest fitness value in the tth generation, represents a random number in the range [0,1], represents the Cauchy probability distribution;

[0236] Genetic determination, specifically, by constructing genetic termination conditions, performing genetic determination on the optimal individual with the highest fitness value, and obtaining the genetic individual with the optimal fitness value in the genetic process;

[0237] The genetic termination conditions include threshold termination and iteration termination;

[0238] The threshold termination is specifically to set the fitness threshold, when the individual fitness value f i When the fitness threshold is exceeded, inheritance is completed;

[0239] The iteration termination specifically refers to terminating the iteration and obtaining the optimal genetic individual gene when the maximum number of iterations is reached;

[0240] The optimal genetic individual gene is specifically a controllable load regulation model parameter combination.

[0241] By performing the above operations, in order to solve the technical problems of improper parameter settings in traditional models and weak ability of parameter optimization algorithms to obtain global optimal solutions, which lead to inaccurate final results, this solution uses nonlinear sinusoidal chaos mapping, adaptive individual genetic probability and superior product retention strategy to enhance the global optimal solution search and improve the robustness of the algorithm, thereby obtaining the optimal parameter values ​​and solving the problem of inaccurate output results of the final adjustment model.

[0242] Embodiment 7, see Figure 1 This embodiment is based on the above embodiment. The charging pile intelligent management module specifically dynamically adjusts the charging power and charging price of each charging pile according to the controllable load adjustment strategy of the single charging pile, so as to realize intelligent management and scheduling of charging piles in the area.

[0243] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0244] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0245] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A charging pile intelligent management system based on big data, characterized by: It includes data acquisition module, data preprocessing module, charging pile load prediction module, charging pile controllable load adjustment module, model hyperparameter optimization module and charging pile intelligent management module; The data acquisition module obtains the original data of intelligent management of charging piles by collecting data from the national power grid system; The data preprocessing module performs data cleaning, standardization processing and feature selection on the charging pile intelligent management raw data to obtain the charging pile intelligent management preliminary data; The charging pile load prediction module establishes a charging pile load prediction model by designing a charging pile load prediction activation function, adopts a momentum update strategy and a mechanism of dynamically adjusting a learning factor during the model training process, performs model training on the charging pile load prediction model, and inputs real-time data into the trained charging pile load prediction model to achieve real-time prediction of a single charging pile load, thereby obtaining a real-time prediction result of a single charging pile load; The controllable load regulation module of the charging pile determines the load regulation target and uses the deep Q network algorithm to establish a controllable load regulation model in combination with the overall load condition of the power grid, performs model training on the controllable load regulation model, and inputs the real-time controllable regulation data of the charging pile in the preliminary data of intelligent management of the charging pile and the real-time prediction result of the load of the single charging pile into the trained controllable load regulation model to obtain a controllable load regulation strategy for a single charging pile; The model hyperparameter optimization module obtains the optimal parameter combination of the controllable load regulation model by improving the seed genetic algorithm through nonlinear sinusoidal chaos mapping, adaptive individual genetic probability and superior product retention strategy; The charging pile intelligent management module realizes intelligent management and scheduling of charging piles in the area based on the output result of the charging pile controllable load adjustment module.

2. According to the big data-based charging pile intelligent management system of claim 1, it is characterized by: The charging pile load prediction module includes establishing a charging pile load prediction model, charging pile load prediction model training and single charging pile load real-time prediction; specifically includes the following steps: Establishing a charging pile load prediction model includes the following steps: Design the charging pile load prediction activation function, specifically design the prediction model input activation function and the prediction model output activation function. The formula used is as follows: ； ； In the formula, represents the prediction model input activation function, represents the output activation function of the prediction model, x represents the input variable of the activation function, represents the parameter of the prediction model input activation function controlling the nonlinear shape, It represents the parameter of the output activation function of the prediction model to control the nonlinear shape, and p represents the dynamic adjustment parameter; Initialize the internal structure MT of the prediction model, which includes the following steps: Input data processing, the formula used is as follows: ； In the formula, Represents the output value after input data processing, Represents the weight matrix corresponding to the input data processing, represents the bias parameter for input data processing, It indicates input data. Indicates the hidden state at the last moment; The historical information is forgotten and the formula used is as follows: ； In the formula, Represents the output after the historical information is forgotten. Represents the weight matrix corresponding to the historical information forgetting process, The bias parameter representing the forgetting process of historical information; Output data calculation, the formula used is as follows: ； In the formula, Represents the result of the output data calculation, represents the weight matrix corresponding to the output data calculation, Represents the bias parameter for output data calculation; The status update candidate value calculation uses the following formula: ； In the formula, Indicates the new state update candidate value, represents the weight matrix corresponding to the calculation of the state update candidate value, Represents the bias parameter for calculating the candidate value for state update; The current status is obtained using the following formula: ； In the formula, c k Indicates the current state, c k-1 Indicates the state at the previous moment. represents element-wise multiplication; Get the current hidden state. The formula used is as follows: ； In the formula, h k Indicates the hidden state at the current moment; The final output result is obtained using the following formula: ； In the formula, represents the charging pile load prediction result at the kth moment, represents the weight matrix corresponding to the output, Represents the bias parameter of the output layer; The charging pile load prediction model training is specifically to use the historical charging pile data in the preliminary data of the charging pile intelligent management as input data to perform model parameter update training on the charging pile load prediction model to obtain the charging pile load prediction model after the model parameters are updated; comprising the following steps: Design the prediction model loss function, the formula used is as follows: ； Where N represents the total number of time steps of the output result. represents the actual result of the charging pile load at the kth moment, represents the charging pile load prediction result at the k+2th moment, represents the charging pile load prediction result at the k+1th moment, represents the prediction model loss function, represents the weight parameter of the smoothness loss, represents the weight parameter of the weighted loss, represents the weighting coefficient at the kth moment; The prediction model reverse parameter update is specifically to update the parameters of the internal structure MT parameters of the prediction model. The internal structure MT parameters of the prediction model include: , , , , , , and ; including the following steps: Calculate the current gradient, specifically the gradient of the loss function with respect to the parameters. The formula used is as follows: ； In the formula, represents the gradient vector at the current moment, Indicates the parameters The partial derivative of Represents the model parameters at the previous moment; To update momentum, the formula used is as follows: ； ； ； In the formula, represents the first-order momentum update value at the current moment, represents the second-order momentum update value at the current moment, represents the infinite norm momentum update value at the current moment, represents the infinite norm momentum update value at the previous moment, represents the decay rate of the first-order momentum, represents the decay rate of the second-order momentum, represents the first-order momentum update value at the previous moment, Represents the second-order momentum update value at the previous moment; The mixed momentum is calculated using the following formula: ； In the formula, Represents the hybrid momentum value at the current moment, which is used to accelerate convergence and stabilize parameter updates. represents the cumulative decay of all first-order momentum at the current time t; The dynamic adjustment learning factor is calculated using the following formula: ； In the formula, represents the dynamic adjustment of learning factor, t represents the number of time steps, T represents the total number of time steps, represents the cumulative decay of all second-order momentum at the current time t; Parameter optimization update, the formula used is as follows: ； In the formula, Indicates the current parameter value. Indicates the parameter value at the previous moment. Represents the learning rate, which is used to control the step size of parameter updates; Iteration judgment, specifically monitoring the change of the loss function and the current number of iterations. If the loss function is lower than the set threshold, the iterative training is stopped, otherwise the iterative training is continued. If the maximum number of iterations is reached, the iterative training is stopped, otherwise the iterative training is continued, and finally the charging pile load prediction model with updated model parameters is obtained; The real-time prediction of the load of a single charging pile is specifically to use the real-time charging pile data in the preliminary data of the intelligent management of the charging pile as the input data of the charging pile load prediction model after the model parameters are updated, to obtain the real-time prediction result of the load of a single charging pile, and the real-time prediction result of the load of a single charging pile includes the prediction time and load power.

3. According to the big data-based charging pile intelligent management system of claim 1, it is characterized by: The controllable load regulation module of the charging pile includes determining the load regulation target, establishing a controllable load regulation model, training the controllable load regulation model and obtaining a controllable load regulation strategy for a single charging pile; specifically, the following steps are included: Determine the controllable load adjustment target of charging piles, specifically to achieve grid load balance, charging pile load scheduling, charging demand maximization and overall energy efficiency maximization; A controllable load regulation model is established, specifically using a deep Q network algorithm to establish the model, including the following steps: Define the state using the following formula: ； In the formula, Indicates the current state. Indicates the current total load of the power grid. represents the predicted load of the i-th charging pile, represents the power demanded by the user of the i-th charging pile, represents the load forecast value of the power grid at time t, represents the health status of the power grid at time t, represents the environmental data at time t, represents the charging price of the i-th charging pile; Define the action space, the formula used is as follows: ； In the formula, Indicates the charging power limit. Indicates charging power delay, Indicates that the charging power is increased. It means lowering the electricity price of charging piles. represents the increase of charging pile electricity price, and A represents the action space; Design the reward function and use the following formula: ； In the formula, represents the current reward function value, Indicates that the charging demand meets the reward. represents the overall energy efficiency reward, , , Represents the weight parameters controlling each target; Interaction and storage experience, depending on the current state , using the ε-greedy strategy to select actions , perform the action , record the next state and rewards ; Experience Deposit into the experience replay pool; To calculate the target value, the formula used is as follows: ； In the formula, represents the target value, represents the discount factor, represents the target network, Indicates possible action at the next moment. represents the target network parameters; Training the controllable load regulation model, specifically using the historical charging pile controllable regulation data in the preliminary data of the charging pile intelligent management to train the controllable load regulation model, to obtain a trained controllable load regulation model; A controllable load regulation strategy for a single charging pile is obtained, specifically, the real-time controllable regulation data of the charging pile in the preliminary data of intelligent management of the charging pile and the real-time prediction result of the load of the single charging pile are used as input data of the trained controllable load regulation model to obtain the controllable load regulation strategy for a single charging pile.

4. The charging pile intelligent management system based on big data according to claim 1 is characterized by: The model hyperparameter optimization module specifically obtains the optimal parameter combination of the controllable load regulation model through a seed genetic algorithm; The following steps are involved: Initialization parameters, specifically, by constructing the initial algorithm parameters; the initial algorithm parameters include population size N, initial crossover probability interval, initial mutation probability interval and maximum number of iterations ; Initializing the population, specifically, initializing the population by generating chaotic sequence values ​​through nonlinear sinusoidal chaotic mapping, including the following steps: Generate N chaotic sequence values ​​and initialize the random value z0; the formula used is as follows: ； In the formula, represents the i-th chaotic sequence value, represents the i+1th chaotic sequence value, Represents adjustment parameters, which are used to adjust the mapping behavior. Indicates the parameter to avoid the denominator being 0; The chaotic sequence value transforms the seed genetic individual, and the formula is expressed as follows: ； In the formula, represents the initial genetic individual gene of the i-th individual, and represent the lower and upper genetic limits of the population, respectively; Calculate the fitness value, specifically calculate the individual fitness value f in the population i ; The performance of the controllable load regulation model based on the individual position is used as the individual fitness value; The superior product retention strategy ranks the fitness values, selects the top 5% of genetic individuals with the highest fitness values, and inserts them into the next generation population. No individual gene crossover operation or individual genetic mutation operation is performed. During each iteration, the size of the population is guaranteed to be the same as the number of genetic individuals in the initial population. The selection operation is to sort the calculated fitness values ​​and select 80% of the genetic individuals for individual gene crossover operation and individual genetic mutation operation according to the roulette wheel selection probability; the formula used is as follows: ； In the formula, represents the probability of roulette selection for the i-th individual, represents the fitness function of the individual; Calculate the adaptive individual genetic probability, specifically calculate the individual gene crossover probability and individual mutation probability, the formula used is as follows: ； ； In the formula, represents the latest value of the crossover probability, represents the latest value of the mutation probability, and represents the maximum and minimum crossover probability, and represents the maximum and minimum values ​​of mutation probability, represents the current value of the crossover probability, represents the current value of the mutation probability, represents the average value of the population fitness, The maximum value of the population fitness; Individual gene crossover operation, specifically if , then two random parent individuals are randomly selected from the population for gene crossover to generate two offspring individuals; the formula used is as follows: ； ； In the formula, and They represent the individual genes inherited by the two offspring after gene crossover. and Represent the gene values ​​of two parent individuals, represents a random number in the range [0,1], c represents the intersection of the gene, k represents the length of the gene, and j represents the index of the gene; Individual genetic variation operations are performed using the following formula: ； In the formula, Indicates the mutated genetic individual genes in the t+1 generation, represents the genetic individual gene with the highest fitness value in the tth generation, represents a random number in the range [0,1], represents the Cauchy probability distribution; Genetic determination, specifically, by constructing genetic termination conditions, performing genetic determination on the optimal individual with the highest fitness value, and obtaining the genetic individual with the optimal fitness value in the genetic process; The genetic termination conditions include threshold termination and iteration termination; The threshold termination is specifically to set the fitness threshold, when the individual fitness value f i When the fitness threshold is exceeded, inheritance is completed; The iteration termination specifically refers to terminating the iteration and obtaining the optimal genetic individual gene when the maximum number of iterations is reached; The optimal genetic individual gene is specifically a controllable load regulation model parameter combination.

5. The charging pile intelligent management system based on big data according to claim 1 is characterized by: The charging pile intelligent management module specifically dynamically adjusts the charging power and charging price of each charging pile according to the controllable load adjustment strategy of the single charging pile, so as to realize intelligent management and scheduling of charging piles in the area.

6. The charging pile intelligent management system based on big data according to claim 1 is characterized by: The data acquisition module specifically acquires the original data of intelligent management of charging piles from the national grid system by acquisition; the original data of intelligent management of charging piles includes historical charging pile data, real-time charging pile data, historical charging pile controllable adjustment data and real-time charging pile controllable adjustment data.

7. The charging pile intelligent management system based on big data according to claim 1 is characterized by: The data preprocessing module specifically performs data cleaning, standardization processing and feature selection on the original data of the charging pile intelligent management to obtain preliminary data of the charging pile intelligent management; the data cleaning is to process missing values, abnormal values ​​and duplicate values; the standardization processing is to standardize the data based on the maximum and minimum normalization method; The feature selection is to use a correlation analysis method to screen out features related to charging pile load prediction and charging pile controllable load regulation.

Images