Task Allocation Method for High-Rate Chargers of New Energy Vehicles Based on Ant Colony Algorithm

Abstract

The present invention belongs to the technical field of the cross-integration of intelligent optimization algorithms and high-rate charging of new energy sources, and provides a method for task allocation of high-rate charging piles for new energy vehicles based on the ant colony algorithm. The present invention includes the following steps: Step S1: Pre-search and filter high-rate charging piles with a danger probability greater than a threshold to form a candidate set; Step S2: Use a graph model to define a high-rate charging task set and a high-rate charging pile set; Step S3: Construct a comprehensive cost function and determine the weight coefficients of the comprehensive cost function; Step S4: The high-rate charging piles conduct multiple rounds of search; Step S5: Perform local search optimization on the optimal solution obtained by each ant; Step S6: Match and optimize to complete the allocation. By means of preprocessing to filter charging piles with too high a danger probability, a multi-round search strategy, the application of simulated pruning operations, and a local optimization algorithm, the present invention effectively solves the problems of safety and efficiency existing in traditional methods.

Description

Technical Field

[0001] The present invention belongs to the technical field of the cross - integration of intelligent optimization algorithms and new - energy high - rate charging, and particularly relates to a method for task allocation of new - energy vehicle high - rate charging piles based on the ant colony algorithm. Background Technique

[0002] High - rate charging refers to the charging current exceeding 1C of the battery's rated capacity, mainly applied to new - energy fast - charging scenarios. With the popularization of electric vehicles and the rapid development of renewable energy, high - rate charging piles are the key facilities to alleviate charging anxiety and improve charging efficiency. However, in high - rate charging scenarios, the safety problems of charging piles are becoming increasingly prominent. The main safety risks include battery thermal runaway, uneven power distribution, overheating of cables, and insufficient environmental heat dissipation. These problems may not only lead to low charging efficiency but also trigger serious safety accidents, such as charging pile fires, threatening the safety of users' lives and property.

[0003] Traditional charging pile management methods often focus on improving charging efficiency. These methods usually directly include all available charging piles in the candidate set and perform task allocation based on simple factors such as distance and price, while ignoring the key factor of the danger probability of charging piles. When using the traditional ant colony algorithm to handle high - rate charging pile task allocation, there is a lack of effective pre - processing means to filter out charging piles with too high a danger probability, resulting in the reliability and safety of task allocation results being difficult to guarantee. Therefore, how to optimize the high - rate charging pile matching algorithm and reduce the failure rate while ensuring safety has become an urgent problem to be solved in the current high - rate charging pile management field. Summary of the Invention

[0004] In view of the deficiencies of the prior art, the present invention provides a method for task allocation of new - energy vehicle high - rate charging piles based on the ant colony algorithm, aiming to solve the problems mentioned in the background technique. This method not only inherits the positive feedback and concurrency characteristics of the ant colony algorithm and can quickly find the optimal task allocation scheme but also realizes safe, flexible, and efficient task allocation.

[0005] The present invention provides a method for task allocation of new - energy vehicle high - rate charging piles based on the ant colony algorithm, including the following steps:

[0006] Step S1: Perform pre - search according to the prediction results of the deep - learning model of high - rate charging piles, filter out high - rate charging piles with a danger probability greater than the threshold, and form a candidate set;

[0007] Step S2: Based on the candidate set, use a graph model to define the high - rate charging task set of new - energy vehicles and the high - rate charging pile set; the high - rate charging task set includes several high - rate charging task nodes, and the high - rate charging pile set includes several high - rate charging pile nodes;

[0008] Step S3: Construct the comprehensive cost function of high-rate charging task nodes and high-rate charging pile nodes, analyze and streamline the parameters of high-rate charging piles in the candidate set through simulated pruning, and determine the weight coefficients of the comprehensive cost function based on the parameters of the high-rate charging piles after simulated pruning;

[0009] Step S4: Conduct multiple rounds of search on the high-rate charging piles in the candidate set through the ant colony algorithm. The process of the ant colony algorithm includes initializing pheromones, simulating the construction of matching solutions, updating pheromones according to the comprehensive cost, iterating, and obtaining the optimal solution;

[0010] Step S5: Optimize the local search for the optimal solutions obtained by each ant;

[0011] Step S6: Combine real-time data for matching optimization to complete the allocation.

[0012] Furthermore, Step S1 specifically includes the following steps:

[0013] Step S11: Collect the status data, environmental data, and battery data of high-rate charging piles, and then perform preprocessing, including data cleaning and feature engineering; among them, the status data of high-rate charging piles includes fault records, real-time temperature, and power distribution; the environmental data includes temperature, humidity, and the status of fire protection facilities; the battery data includes health status, state of charge, and number of charge and discharge cycles;

[0014] Step S12: Input the preprocessed high-rate charging pile data into the deep learning model for training, and predict the risk probability of high-rate charging piles according to historical fault cases and environmental data during the training process ;

[0015] Step S13: Draw the ROC curve, calculate the difference between the recall rate and the false alarm rate, and calculate the optimized threshold of the risk probability. The formula is expressed as:

[0016] ;

[0017] In the formula: is the vertical axis of the curve, representing the proportion of fault samples correctly predicted as high-risk by the deep learning model; is the horizontal axis of the curve, representing the proportion of normal samples wrongly predicted as high-risk by the deep learning model; is the optimized threshold; is the value of the threshold Threshold corresponding to the maximum value within the value range of the threshold Threshold;

[0018] Step S14: Screen according to the optimized threshold, filter out high-rate charging piles with a risk probability greater than the optimized threshold. The filtering formula is expressed as:

[0019] ;

[0020] When the risk probability of the high-rate charging pile ≤ the optimized threshold , the high-rate charging pile is considered safe and included in the candidate set; otherwise, the high-rate charging pile is considered dangerous and not included in the candidate set.

[0021] Further, in step S2, the high-rate charging task set and the high-rate charging pile graph model of the new energy vehicle are expressed as: G = (C, T, W, E);

[0022] Where: T is the set of high-rate charging task nodes of the new energy vehicle to be allocated; C is the set of high-rate charging pile nodes of the new energy vehicle; E is the distance from the new energy vehicle to the high-rate charging pile or the edge connecting two high-rate charging pile nodes; W is the comprehensive cost for the new energy vehicle to reach the high-rate charging pile.

[0023] Further, in step S2, the high-rate charging task set and the high-rate charging pile set of the new energy vehicle are defined using a graph model, which specifically includes the following steps:

[0024] Step S21: High-rate charging task set: T = (t1, t2,..., t n ), where t1, t2,..., t n are the first high-rate charging task node, the second high-rate charging task node, and the nth high-rate charging task node respectively; the information included in the high-rate charging task node: high-rate charging task number, high-rate charging task location, and the charging amount required for the high-rate charging task;

[0025] Step S22: High-rate charging pile set: C = (c1, c2,..., c m ), where c1, c2,..., c m are the first high-rate charging pile node, the second high-rate charging pile node, and the mth high-rate charging pile node respectively; the information included in the high-rate charging pile node: high-rate charging pile number, high-rate charging pile location, the current load of the high-rate charging pile, and the maximum load capacity of the high-rate charging pile.

[0026] Further, step S3 specifically includes the following steps:

[0027] Step S31: Analyze and streamline the high-rate charging pile parameters in the high-rate charging pile candidate set through simulation pruning to remove unnecessary high-rate charging pile parameters in the candidate set. The simulation pruning operation can be expressed as:

[0028] ;

[0029] Among them, P is a set of parameters, and w i is a parameter weight, Q is a pruning threshold, and P pruned is the set of parameters after pruning, which only contains parameters with weights not less than the pruning threshold Q;

[0030] Step S32: Use statistical methods to evaluate the correlation or influence degree between each high-rate charging pile parameter after simulated pruning and the comprehensive cost function, so as to determine the weight coefficient of each high-rate charging pile parameter after simulated pruning on the comprehensive cost function. The statistical methods include the correlation coefficient method or the regression analysis method;

[0031] Step S33: Construct the comprehensive cost function of the high-rate charging task node and the high-rate charging pile node, which is expressed by the formula:

[0032] ;

[0033] In the formula: is the comprehensive cost of using the j-th high-rate charging pile node c i at the i-th high-rate charging task node t j ; are all weight coefficients, ; is the distance from the current position of the i-th high-rate charging task node t i to the j-th high-rate charging pile node c j ; is the estimated high-rate charging time of using the j-th high-rate charging pile node c i at the i-th high-rate charging task node t j ; is the current load of the j-th high-rate charging pile node c j ;

[0034] Furthermore, step S4 specifically includes the following steps:

[0035] Step S41: Initialize the pheromone concentration of each connection between the high-rate charging task node and the high-rate charging pile node, which is expressed by the formula:

[0036] ;

[0037] In the formula: is the pheromone concentration from the i-th high-rate charging task node t i to the j-th high-rate charging pile node c j , that is, an index of the attractiveness or selection probability of this path;

[0038] Step S42: Expand multiple rounds of simulated search according to the weight coefficients of the comprehensive cost function. The specific steps of each round of simulated search are as follows:

[0039] Preset the maximum number of iterations and the pheromone threshold;

[0040] Each ant randomly selects an unassigned high-rate charging task node and a high-rate charging pile node, which is expressed by the formula:

[0041] ;

[0042] In the formula: is the probability that the ant selects to match the ith high-rate charging task node t i to the jth high-rate charging pile node c j ; represents the information for evaluating the matching of the ith high-rate charging task node t i to be assigned and the jth high-rate charging pile node c j based on other influencing factors or rules; N is the set of the ith high-rate charging task node t i and the kth high-rate charging pile node c k ;

[0043] Update the corresponding relationship between the selected high-rate charging task node and the high-rate charging pile node, and mark the high-rate charging task node as assigned;

[0044] Step S43: Calculate the comprehensive cost of the current matching solution through the comprehensive cost function, and update the pheromone concentration according to the comprehensive cost, which is expressed by the formula:

[0045] ;

[0046] In the formula, is the pheromone concentration on the path from the ith high-rate charging task node t i to the jth high-rate charging pile node c j ; is the pheromone evaporation coefficient; represents the remaining part of the original pheromone concentration; is the pheromone increment on the path from the ith high-rate charging task node t i to the jth high-rate charging pile node c j ;

[0047] Step S44: When the preset maximum number of iterations is reached or the change rate of pheromone concentration is less than the preset pheromone threshold, stop updating the pheromone and the simulated matching pheromone iteration, record the ant matching of this round of simulated search as the optimal solution, and mark the weight coefficient of the comprehensive cost function of this round of simulated search;

[0048] Step S45: Repeat Step S43 - Step S44 until the weight coefficients of the comprehensive cost function are all marked, obtain the optimal solution of each ant, that is, finally obtain the optimal solution set of multiple rounds of simulated search.

[0049] Further, Step S5 is specifically as follows. The optimal solution obtained by each ant is optimized by local search using the simulated annealing algorithm, including the following steps:

[0050] Step S51: Use the optimal solution of each ant as the initial solution, and set the initial temperature, termination temperature, and cooling rate;

[0051] Step S52: According to the high-rate charging pile j of the current solution, calculate the comprehensive score S(j) of the high-rate charging pile j; then randomly select a high-rate charging pile within the neighborhood of the current solution , and calculate the comprehensive score of the high-rate charging pile ;

[0052] The comprehensive score S formula is expressed as:

[0053]

[0054] In the formula: are the weight coefficients of Safety, Speed, Price, and Distance respectively;

[0055] Step S53: According to the Metropolis criterion, if , then accept the high-rate charging pile ; if , then calculate the acceptance probability , and the formula is expressed as:

[0056] ;

[0057] In the formula: T is the temperature; P M is the acceptance probability;

[0058] Step S54: Update the temperature according to the cooling strategy;

[0059] Step S55: Repeat Step S52 - Step S54 until the termination temperature is reached, obtain the optimal solution after local search optimization of each ant, as the result of local search optimization.

[0060] The present invention has the following beneficial effects:

[0061] (1) By virtue of the positive feedback and concurrency characteristics of the ant colony algorithm, an optimal task allocation scheme can be quickly found. The positive feedback mechanism of the ant colony algorithm can achieve load balancing for high-rate charging tasks, avoid the problem of overload during peak hours at high-rate charging stations, and improve resource utilization efficiency. Moreover, the ant colony algorithm has strong robustness and self-adaptability, and can adapt to task allocation problems of different scales and complexities. And through the continuously iterative pheromone update mechanism, the system can provide more intelligent decision-making support to help managers make more scientific resource allocation and scheduling decisions. In addition, by simulating ant colony behavior, a comprehensive cost function is constructed by combining factors such as the geographical location, charging speed, and current battery level of high-rate charging piles, and dynamic information such as user behavior prediction and environmental factor changes is incorporated, making the task allocation more accurate and real-time.

[0062] (2) A deep learning model is introduced to accurately predict the risk probability of high-rate charging piles, and the threshold of the risk probability is optimized by using the ROC curve analysis, effectively filtering out potential safety hazards and improving the safety and reliability of task allocation. At the same time, a multi-round search strategy is adopted to optimize for different weight coefficients, improving the flexibility of task allocation to ensure that the final allocation scheme can more comprehensively meet user needs and system performance requirements. In addition, through simulated pruning operations, the parameters of high-rate charging piles are deeply analyzed and streamlined, simplifying the problem complexity, improving the calculation efficiency, and optimizing the determination of weight coefficients and the construction of the comprehensive cost function. After obtaining the initial optimal solution, a local optimization algorithm is further used to finely adjust the solution, improving the accuracy and efficiency of task allocation and enhancing the adaptability to complex environments and dynamic changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] The exemplary embodiments of the present invention can be more fully understood by referring to the following drawings:

[0064] Figure 1 It is a flowchart of a method for task allocation of high-rate charging piles for new energy vehicles based on the ant colony algorithm provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0065] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0066] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this invention belongs; the terms used herein are for the purpose of describing specific embodiments only and are not intended to limit the present invention.

[0067] An embodiment of the present invention provides a high-rate charging pile task allocation method for new energy vehicles based on the ant colony algorithm, including the following steps:

[0068] Step S1: Perform pre-search according to the prediction results of the deep learning model of the high-rate charging pile, filter out high-rate charging piles with a hazard probability greater than the threshold, and form a candidate set;

[0069] Step S2: Based on the candidate set, use a graph model to define the high-rate charging task set and the high-rate charging pile set of new energy vehicles; the high-rate charging task set includes several high-rate charging task nodes, and the high-rate charging pile set includes several high-rate charging pile nodes;

[0070] Step S3: Construct a comprehensive cost function for high-rate charging task nodes and high-rate charging pile nodes, analyze and streamline the parameters of high-rate charging piles in the candidate set through simulated pruning, and determine the weight coefficient of the comprehensive cost function based on the parameters of high-rate charging piles after simulated pruning;

[0071] Step S4: Conduct multiple rounds of search on the high-rate charging piles in the candidate set through the ant colony algorithm. The process of the ant colony algorithm includes initializing pheromone, simulating the construction of matching solutions, updating pheromone according to the comprehensive cost, iterating, and obtaining the optimal solution;

[0072] Step S5: Perform local search optimization on the optimal solutions obtained by each ant;

[0073] Step S6: Combine real-time data for matching optimization to complete the allocation.

[0074] In some embodiments, step S1 specifically includes the following steps:

[0075] Step S11: Collect high-rate charging pile status data, environmental data, and battery data, and then perform preprocessing, which includes data cleaning and feature engineering; among them, the high-rate charging pile status data includes fault records, real-time temperature, and power distribution; the environmental data includes temperature, humidity, and fire protection facility status; the battery data includes state of health (SOH), state of charge (SOC), and number of charge and discharge cycles;

[0076] Specifically, during preprocessing, data cleaning is first performed to remove missing values and outliers. For example, data with a temperature exceeding 100 °C is removed. Then, feature engineering is carried out to extract time series features, such as the temperature volatility in the past 24 hours. Composite features are calculated, such as power density and environmental heat dissipation efficiency. A dynamic temperature threshold based on the battery SOH can also be constructed.

[0077] Step S12: Input the preprocessed high-rate charging pile data into the deep learning model for training. During the training process, based on historical fault cases and environmental data, predict the risk probability of the high-rate charging pile. ;

[0078] Specifically, the deep learning model includes an input layer, a hidden layer, and an output layer. In the input layer, a multi-dimensional feature vector composed of static features such as the high-rate charging pile model and dynamic features such as real-time temperature is input. In the hidden layer, a multi-layer perceptron or a graph neural network is used to capture the spatial dependence of the high-rate charging pile, and the attention mechanism is used to weight the importance of different features. In the output layer, a single neuron outputs the risk probability. In the deep learning model, the binary cross-entropy is used as the training objective of the loss function, labels are set according to historical fault records, data augmentation is performed while processing data, and noise is added to the fault data to improve the robustness of the model.

[0079] Step S13: Plot the ROC curve, calculate the difference between the recall rate and the false alarm rate, and calculate the optimized threshold of the risk probability. The formula is expressed as:

[0080] ;

[0081] In the formula: represents the vertical axis of the curve, indicating the proportion of fault samples correctly predicted as high-risk by the deep learning model; represents the horizontal axis of the curve, indicating the proportion of normal samples mispredicted as high-risk by the deep learning model; is the optimized threshold; is the value of the threshold Threshold corresponding to the maximum value within the value range of the threshold Threshold;

[0082] Specifically, using 100,000 historical fault cases, analyze the TPR (true positive rate) and FPR (false positive rate) at different thresholds through the ROC curve. The optimization goal is to maximize TPR - FPR, ensuring a high recall rate (e.g., TPR > 95%) while controlling the false alarm rate (e.g., FPR < 10%). In addition, the threshold is dynamically adjusted according to the environmental temperature. When the environmental temperature < 35 °C, the threshold is 0.3. When the environmental temperature ≥ 35 °C, the threshold is 0.2.

[0083] Step S14: Filter according to the optimized threshold, and filter out high-rate charging piles with a risk probability greater than the optimized threshold. The filtering formula is expressed as:

[0084] ;

[0085] When the risk probability of the high-rate charging pile ≤ the optimized threshold , the high-rate charging pile is considered safe and included in the candidate set; otherwise, the high-rate charging pile is considered dangerous and not included in the candidate set.

[0086] Specifically, filter the high-rate charging piles according to the optimized threshold, and exclude the high-rate charging piles with a risk probability > threshold to form a candidate set; for example, when the threshold is 0.3, the candidate set is the set of high-rate charging piles with a risk probability ≤ 0.3; the candidate set is also rechecked, and the temperature of the battery and the cable is monitored in real time. If the cable temperature continues to rise, it is excluded.

[0087] In some embodiments, in step S2, the high-rate charging task set and the high-rate charging pile graph model of the new energy vehicle are represented as: G=(C,T,W,E);

[0088] In the formula: T is the set of high-rate charging task nodes of the new energy vehicle to be allocated; C is the set of high-rate charging pile nodes of the new energy vehicle; E is the distance from the new energy vehicle to the high-rate charging pile or the edge connecting two high-rate charging pile nodes; W is the comprehensive cost for the new energy vehicle to reach the high-rate charging pile.

[0089] In some embodiments, in step S2, a graph model is used to define the high-rate charging task set and the high-rate charging pile set of the new energy vehicle, which specifically includes the following steps:

[0090] Step S21: High-rate charging task set: T=(t1,t2,...,t n ), where t1,t2,...,t n are the first high-rate charging task node, the second high-rate charging task node, and the nth high-rate charging task node respectively; the information included in the high-rate charging task node: high-rate charging task number, high-rate charging task location, and charging amount required for the high-rate charging task;

[0091] Step S22: High-rate charging pile set: C=(c1,c2,...,c m ), where c1,c2,...,c mThey are the first high-power charging pile node, the second high-power charging pile node, and the m-th high-power charging pile node respectively; the information included in the high-power charging pile node: high-power charging pile number, high-power charging pile location, current load of the high-power charging pile, and maximum load capacity of the high-power charging pile.

[0092] In some embodiments, step S3 specifically includes the following steps:

[0093] Step S31: Analyze and streamline the high-power charging pile parameters in the high-power charging pile candidate set through simulated pruning to remove unnecessary high-power charging pile parameters in the candidate set. The simulated pruning operation can be expressed as:

[0094] ;

[0095] where P is the parameter set, w i is the weight of the parameter , Q is the pruning threshold, and P pruned is the parameter set after pruning, which only contains parameters with weights not less than the pruning threshold Q;

[0096] Specifically, simulated pruning will follow certain rules, such as deleting redundant information or adjusting parameters;

[0097] Step S32: Use statistical methods to evaluate the correlation or influence degree between each high-power charging pile parameter after simulated pruning and the comprehensive cost function to determine the weight coefficient of each high-power charging pile parameter after simulated pruning on the comprehensive cost function. The statistical methods include the correlation coefficient method or the regression analysis method;

[0098] Specifically, taking parameters such as distance, estimated charging time, and current load as examples for weight coefficient definition:

[0099] Generally, the value of the distance weight coefficient is 0.3 - 0.5. If the distance factor has a greater impact on the comprehensive cost in the charging scenario, for example, in long-distance transportation or when high-power charging piles are relatively scattered in the city, a larger value can be taken, such as 0.45 - 0.5; if the distance factor has a relatively small impact, such as when high-power charging piles are relatively dense in the city center area and the vehicle driving distance is generally short, a smaller value can be taken, such as 0.3 - 0.35;

[0100] The value range of the expected charging time weight coefficient β is 0.2 - 0.4. When the charging powers of high-rate charging piles vary greatly and users are more sensitive to charging time, β can take a larger value, such as 0.35 - 0.4; if the charging powers of high-rate charging piles are relatively stable and users have less strict requirements for charging time, β can take a smaller value, such as 0.2 - 0.25;

[0101] The value range of the current load weight coefficient γ is 0.1 - 0.3. During the charging peak period, the loads of high-rate charging piles are generally high, and γ can take a larger value, such as 0.25 - 0.3. During the non-peak period, the loads of high-rate charging piles are low, and γ can take a smaller value, such as 0.1 - 0.15;

[0102] Meanwhile, to ensure the rationality of the weight coefficients, it is necessary to satisfy α + β + γ = 1;

[0103] Step S33: Construct the comprehensive cost function of high-rate charging task nodes and high-rate charging pile nodes, which is expressed by the formula:

[0104] ;

[0105] In the formula: is the comprehensive cost of using the j-th high-rate charging pile node c i at the i-th high-rate charging task node t j ; are all weight coefficients, ; is the distance from the current position of the i-th high-rate charging task node t i to the j-th high-rate charging pile node c j ; is the expected high-rate charging time of using the j-th high-rate charging pile node c i at the i-th high-rate charging task node t j ; is the current load of the j-th high-rate charging pile node c j ;

[0106] Specifically, when constructing the comprehensive cost function, safety parameters, price cost parameters, distance parameters, charging speed parameters, etc. are considered; among them, safety parameters include the risk of battery thermal runaway, the risk of uneven power distribution, the risk of cable overheating, the risk of insufficient environmental heat dissipation, etc.; charging speed parameters include the output power of the charging pile, etc.

[0107] In some embodiments, step S4 specifically includes the following steps:

[0108] Step S41: Initialize the pheromone concentration of each connection between high-rate charging task nodes and high-rate charging pile nodes, which is expressed by the formula:

[0109] ;

[0110] In the formula: is the pheromone concentration from the i-th high-rate charging task node t i to the j-th high-rate charging pile node c j which is an indicator of the attractiveness or selection probability of this path;

[0111] It can be understood that each ant selects a path according to the probability, which is determined by the pheromone concentration and the attractiveness of the path. The path attractiveness is determined by the predicted high-rate charging demand and the availability of high-rate charging piles. In practical applications, the initial pheromone concentration of each pair of high-rate charging task node - high-rate charging pile node can be set according to historical data. For example, if a certain high-rate charging pile has been frequently used by a certain type of task in the past, a higher initial pheromone concentration can be set to reflect its attractiveness;

[0112] Step S42: Conduct multiple rounds of simulated search according to the weight coefficients of the comprehensive cost function. Each round of simulated search is carried out around different combinations of weight coefficients, aiming to obtain the optimal solutions under different influencing factors. The specific steps of each round of simulated search are as follows:

[0113] Preset the maximum number of iterations MAX_ITER and the pheromone threshold ;

[0114] The maximum number of iterations MAX_ITER indicates that the ant will try to match the high-rate charging task node - high-rate charging pile node for at most MAX_ITER rounds; the pheromone threshold , which is used to judge whether the pheromone is high enough to affect the selection;

[0115] Each ant randomly selects an unassigned high-rate charging task node and high-rate charging pile node, which is expressed by the formula:

[0116] ;

[0117] In the formula: is the probability that the ant selects to match the i-th high-rate charging task node t i to the j-th high-rate charging pile node c j ; is the information indicating the matching of the i-th high-rate charging task node t i to be assigned and the j-th high-rate charging pile node c j evaluated based on other influencing factors or rules; N is the set of the i-th high-rate charging task node t i and the k-th high-rate charging pile node c k ;

[0118] Update the correspondence between the selected high-rate charging task nodes and high-rate charging pile nodes, and mark the high-rate charging task nodes as allocated;

[0119] Specifically, the ant randomly selects a high-rate charging task node from the unallocated high-rate charging tasks. For example, electric vehicle A needs high-rate charging, and then randomly selects one from the available high-rate charging piles (such as high-rate charging pile 1). The selection probability is jointly determined by the pheromone concentration and heuristic information (such as the idle state of the high-rate charging pile, the estimated high-rate charging time, etc.);

[0120] Once the high-rate charging task node and the high-rate charging pile node are selected, then update the correspondence between them, record that electric vehicle A will use high-rate charging pile 1, update the status of electric vehicle A to allocated, and remove it from the unallocated task list to avoid repeated allocation;

[0121] Step S43: Calculate the comprehensive cost of the current matching solution through the comprehensive cost function, and update the pheromone concentration according to the comprehensive cost. The formula is expressed as:

[0122] ;

[0123] In the formula, is the pheromone concentration on the path from the i-th high-rate charging task node t i to the j-th high-rate charging pile node c j ; is the pheromone evaporation coefficient; represents the remaining part of the original pheromone concentration; is the pheromone increment on the path from the i-th high-rate charging task node t i to the j-th high-rate charging pile node c j ;

[0124] Specifically, takes values between 0 and 1, and is jointly determined by the comprehensive cost function and heuristic information (such as the idle state of the high-rate charging pile, the estimated high-rate charging time, etc.); in practical applications, whenever an ant completes a high-rate charging task allocation, the comprehensive cost of this path will be calculated to update the pheromone concentration; thus, better paths will obtain more pheromones, making subsequent ants more inclined to choose these paths, which helps subsequent ants to more effectively utilize the previous selection results;

[0125] Step S44: When the preset maximum number of iterations is reached or the change rate of pheromone concentration is less than the preset pheromone threshold, stop updating the pheromone and the simulated matching pheromone iteration, record the ant matching in this round of simulated search as the optimal solution, and mark the weight coefficient of the comprehensive cost function in this round of simulated search;

[0126] Step S45: Repeat Step S43 - Step S44 until the weight coefficients of the comprehensive cost function are all marked, obtain the optimal solution for each ant, that is, finally obtain the optimal solution set of multiple rounds of simulated search.

[0127] In some embodiments, Step S5 is specifically to perform local search optimization on the optimal solution obtained by each ant using the simulated annealing algorithm;

[0128] Specifically, after the ant colony algorithm search, 3 ants obtained 3 different optimal solutions, which are respectively:

[0129] Solution A: Recommend the high - magnification charging pile P1, which has a high safety factor, a moderate charging price, but is far from the user;

[0130] Solution B: Recommend the high - magnification charging pile P2, which has a fast charging speed, but a relatively high price and an average safety record;

[0131] Solution C: Recommend the high - magnification charging pile P3, which is the closest to the user, has a relatively low price, but a slow charging speed;

[0132] In some embodiments, Step S5 includes the following steps:

[0133] Step S51: Take the optimal solution of each ant as the initial solution, set the initial temperature T0, the termination temperature end and the cooling rate ;

[0134] For example, set the initial parameters to take Solution A, Solution B, and Solution C as the initial solutions respectively, set the initial temperature T0 = 100, the termination temperature T end = 1, and the cooling rate = 0.9;

[0135] Step S52: According to the high - magnification charging pile j of the current solution, calculate the comprehensive score S(j) of the high - magnification charging pile j; then randomly select a high - magnification charging pile in the neighborhood of the current solution, and calculate the comprehensive score of the high - magnification charging pile ;

[0136] The comprehensive score S formula is expressed as:

[0137]

[0138] In the formula: are the weight coefficients of Safety, Speed, Price, and Distance respectively;

[0139] Specifically, taking solution A as an example, a high-rate charging pile is randomly selected within its neighborhood, where w1 = 0.3, w2 = 0.2, w3 = 0.2, w4 = 0.3, and the comprehensive score S1 of the high-rate charging pile of the current solution is calculated as 80; the comprehensive score S4 of a high-rate charging pile randomly selected within the neighborhood of the current solution is 82;

[0140] Step S53: According to the Metropolis criterion, if , then accept the high-rate charging pile ; if , then calculate the acceptance probability , and the formula is expressed as:

[0141] ;

[0142] In the formula: T is the temperature; P M is the acceptance probability;

[0143] Step S54: Update the temperature according to the cooling strategy , and the formula is expressed as:

[0144] ;

[0145] Step S55: Repeat steps S52 - S54 until the termination temperature is reached, and the optimal solution after local search optimization for each ant is obtained as the result of local search optimization.

[0146] Specifically, the optimized solutions A', B', and C' are obtained as the results of local search optimization.

[0147] In some embodiments, step S6 is specifically: combining with the user's real-time requirements, selecting the solution with the highest matching degree as the plan to complete the high-rate charging pile task allocation;

[0148] Specifically, the user's real-time requirements are that they hope the charging speed is as fast as possible and the price is not too high; the system evaluates the matching degree of the optimized solutions A', B', and C' according to the user's real-time requirements;

[0149] Solution A': Recommend charging pile P4, with a general charging speed and a moderate price;

[0150] Solution B': Recommend charging pile P5, with a fast charging speed and a slightly higher price;

[0151] Solution C': Recommend charging pile P6, with slow charging speed and low price;

[0152] By calculating the matching degree between each solution and the user's needs, the system finds that the matching degree of solution B' is the highest. Therefore, solution B' is taken as the final solution to complete the task allocation of high-rate charging piles and recommend high-rate charging piles to the user.

[0153] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A task allocation method for high-rate charging piles of new energy vehicles based on the ant colony algorithm, characterized in that It includes the following steps: Step S1: Perform pre-search according to the prediction results of the deep learning model of the high-rate charging pile, filter out the high-rate charging piles with a danger probability greater than the threshold, and form a candidate set; Specifically, it includes the following steps: Step S11: Collect the status data, environmental data, and battery data of the high-rate charging pile, and then perform preprocessing, which includes data cleaning and feature engineering; among them, the status data of the high-rate charging pile includes fault records, real-time temperature, and power distribution; the environmental data includes temperature, humidity, and the status of fire-fighting facilities; the battery data includes health status, state of charge, and number of charge and discharge cycles; Step S12: Input the preprocessed high-rate charging pile data into the deep learning model for training. During the training process, predict the risk probability of the high-rate charging pile based on historical fault cases and environmental data ; Step S13: Draw an ROC curve, calculate the difference between the recall rate and the false alarm rate, and calculate the optimized threshold of the danger probability. The formula is expressed as: ； Wherein: is the vertical axis of the curve, representing the proportion of high-risk cases correctly predicted by the deep learning model in the fault samples; is the horizontal axis of the curve, representing the proportion of normal samples mispredicted as high-risk by the deep learning model; is the optimized threshold; is the value of the threshold Threshold corresponding to the maximum value within the value range of the threshold Threshold; Step S14: Perform screening according to the optimized threshold, and filter out the high-rate charging piles with a danger probability greater than the optimized threshold. The filtering formula is expressed as: ； When the risk probability of the high-power charging pile ≤ the optimized threshold , the high-power charging pile is considered safe and included in the candidate set; otherwise, the high-power charging pile is considered dangerous and not included in the candidate set; Step S2: Based on the candidate set, use a graph model to define the high-rate charging task set and high-rate charging pile set of new energy vehicles; the high-rate charging task set includes several high-rate charging task nodes, and the high-rate charging pile set includes several high-rate charging pile nodes; Step S3: Construct a comprehensive cost function for the high-rate charging task node and the high-rate charging pile node, analyze and streamline the high-rate charging pile parameters in the candidate set through simulated pruning analysis, and determine the weight coefficient of the comprehensive cost function based on the high-rate charging pile parameters after simulated pruning; Specifically, it includes the following steps: Step S31: Analyze and streamline the high-rate charging pile parameters in the high-rate charging pile candidate set through simulated pruning to remove unnecessary high-rate charging pile parameters in the candidate set. The simulated pruning operation can be expressed as: ； Among them, P is a set of parameters, and w i is a parameter weight, Q is a pruning threshold, and P pruned is the set of parameters after pruning, which only contains parameters whose weights are not less than the pruning threshold Q; Step S32: Use statistical methods to evaluate the correlation or influence degree between each high-rate charging pile parameter after simulated pruning and the comprehensive cost function to determine the weight coefficient of each high-rate charging pile parameter after simulated pruning on the comprehensive cost function. The statistical methods include the correlation coefficient method or the regression analysis method; Step S33: Construct a comprehensive cost function for the high-rate charging task node and the high-rate charging pile node. The formula is expressed as: ； Wherein: is the comprehensive cost of using the j-th high-rate charging pile node c i at the i-th high-rate charging task node t j ; are all weight coefficients, ; is the distance from the current position of the i-th high-rate charging task node t i to the j-th high-rate charging pile node c j ; is the estimated high-rate charging time of using the j-th high-rate charging pile node c i at the i-th high-rate charging task node t j ; is the current load of the j-th high-rate charging pile node c j ; Step S4: Conduct multiple rounds of search for the high-rate charging piles in the candidate set through the ant colony algorithm. The process of the ant colony algorithm includes initializing pheromone, simulating the construction of matching solutions, updating pheromone according to the comprehensive cost, iterating, and obtaining the optimal solution; Step S5: Use the simulated annealing algorithm to perform local search optimization on the optimal solution obtained by each ant; Specifically, it includes the following steps: Step S51: Use the optimal solution of each ant as the initial solution, and set the initial temperature, termination temperature, and cooling rate; Step S52: Calculate the comprehensive score S(j) of the high-rate charging pile j according to the current solution; then randomly select a high-rate charging pile within the neighborhood of the current solution , calculate the high-rate charging pile 's comprehensive score ; The formula for the comprehensive score S is expressed as: ； Wherein: are the weight coefficients of Safety, Speed, Price, and Distance, respectively; Step S53: According to the Metropolis criterion, if , then accept the high-power charging pile ; if , then calculate the acceptance probability , which is expressed by the formula: ； Where: T is the temperature; P M is the acceptance probability; Step S54: Update the temperature according to the cooling strategy; Step S55: Repeat Step S52 - Step S54 until the termination temperature is reached, and obtain the optimal solution after local search optimization of each ant as the result of local search optimization; Step S6: Combine the real-time needs of the user, select the solution with the highest matching degree as the plan, and complete the high-rate charging pile task allocation.

2. The high-rate charging pile task allocation method for new energy vehicles based on the ant colony algorithm according to claim 1, characterized in that: In step S2, the high-rate charging task set and high-rate charging pile graph model of new energy vehicles are represented as: G = (C, T, W, E); Where: T is the set of high-rate charging task nodes of new energy vehicles to be allocated; C is the set of high-rate charging pile nodes of new energy vehicles; E is the distance from a new energy vehicle to a high-rate charging pile or the edge connecting two high-rate charging pile nodes; W is the comprehensive cost for a new energy vehicle to reach a high-rate charging pile.

3. The task allocation method for high-rate charging piles of new energy vehicles based on the ant colony algorithm according to claim 2, wherein In step S2, the high-rate charging task set and high-rate charging pile set of new energy vehicles are defined using a graph model, which specifically includes the following steps: Step S21: High-rate charging task set: T = (t1, t2,..., t n ), where t1, t2,..., t n are the first high-rate charging task node, the second high-rate charging task node, and the nth high-rate charging task node respectively; the information included in the high-rate charging task node: high-rate charging task number, high-rate charging task location, and the charging amount required for the high-rate charging task; Step S22: Aggregation of high-power charging piles: C=(c1,c2,...,c m ), where c1, c2,..., c m are the first high-power charging pile node, the second high-power charging pile node, and the m-th high-power charging pile node respectively; the information contained in the high-power charging pile node: high-power charging pile number, high-power charging pile location, current load of the high-power charging pile, and maximum load capacity of the high-power charging pile.

4. The high-rate charging pile task allocation method for new energy vehicles based on the ant colony algorithm according to claim 3, characterized in that, Step S4 specifically includes the following steps: Step S41: Initialize the pheromone concentration of each pair of high-rate charging task node - high-rate charging pile node connections, and the formula is expressed as: ； Wherein: is the pheromone concentration from the i-th high-rate charging task node t i to the j-th high-rate charging pile node c j which is an index of the attractiveness or selection probability of this path; Step S42: Conduct multiple rounds of simulated search based on the weight coefficient of the comprehensive cost function. The specific steps of each round of simulated search are as follows: Preset the maximum number of iterations and the pheromone threshold; Each ant randomly selects an unallocated high-rate charging task node and high-rate charging pile node, and the formula is expressed as: ； Wherein: is the probability that the ant chooses to match the \(i\)-th high-rate charging task node \(t\) i to the \(j\)-th high-rate charging pile node \(c\); j The probability; is the information indicating the matching of the \(i\)-th high-rate charging task node \(t\) to be allocated i and the \(j\)-th high-rate charging pile node \(c\) j evaluated based on other influencing factors or rules; \(N\) is the set of the \(i\)-th high-rate charging task node \(t\) i and the \(k\)-th high-rate charging pile node \(c\); k The set; Update the corresponding relationship between the selected high-rate charging task node and high-rate charging pile node, and mark the high-rate charging task node as allocated; Step S43: Calculate the comprehensive cost of the current matching solution through the comprehensive cost function, and update the pheromone concentration according to the comprehensive cost. The formula is expressed as: ； In the formula, is the pheromone concentration on the path from the ith high-rate charging task node t i to the jth high-rate charging pile node c j ; is the pheromone evaporation coefficient; represents the remaining part of the original pheromone concentration; is the pheromone increment of the path from the ith high-rate charging task node t i to the jth high-rate charging pile node c j ; Step S44: When the preset maximum number of iterations is reached or the change rate of the pheromone concentration is less than the preset pheromone threshold, stop updating the pheromone and simulating the matching pheromone iteration, record the ant matching of this round of simulated search as the optimal solution, and mark the weight coefficient of the comprehensive cost function of this round of simulated search; Step S45: Repeat step S43 - step S44 until the weight coefficients of the comprehensive cost function are all marked, obtain the optimal solution for each ant, that is, finally obtain the optimal solution set of multiple rounds of simulated search.

Images