Electric vehicle scheduling method based on multi-pheromone channel ant colony system

Abstract

The invention discloses an electric vehicle scheduling method based on a multi-pheromone channel ant colony system, and the method comprises the steps: firstly building a scheduling model on the basis of an electric vehicle service scene, and enabling the scheduling model to be related with the multi-source traffic data, such as the vehicle electric quantity and a charging pile, different from a traditional fuel vehicle scheduling model; in order to solve the proposed scheduling model, a multi-pheromone channel ant colony system algorithm is applied. According to the algorithm, on the basis of an ant colony system, a plurality of pheromone channels are established, and a pheromone updating factor is calculated based on an average absolute characteristic value of a pheromone matrix, so that the search capability of the algorithm is enhanced, and the matching degree is improved or the average waiting time is shortened. Experimental results under different scheduling scenes prove that compared with other existing scheduling methods, the electric vehicle scheduling optimization method can obtain a better matching scheme and is an effective method for solving the problem of electric vehicle scheduling.

Description

Technical Field

[0001] The present invention belongs to the technical fields of intelligent transportation and intelligent computing, and particularly relates to an electric vehicle scheduling method based on a multi-pheromone channel ant colony system. Background Art

[0002] In recent years, as an important carrier for green and low-carbon travel, electric vehicles have been widely used in scenarios such as urban transportation, shared travel, and last-mile delivery. However, the widespread deployment of electric vehicles has also brought new technical challenges to scheduling and management. Different from traditional vehicles that only take a few minutes to refuel, electric vehicles have a long charging time, limited cruising range, and uneven distribution of charging piles. This may cause vehicles to drop out of service midway due to insufficient power during the service process, resulting in longer waiting times for customers and affecting the travel experience. Especially in a complex urban environment with high-concurrency requests and a large number of vehicles, how to achieve reasonable scheduling, maximize service capacity, minimize user waiting time, and balance charging and task requirements has become an important issue to be solved urgently. Existing research covers multiple aspects such as modeling strategies, scheduling algorithms, and key constraint modeling, and a large number of representative works have emerged.

[0003] As a swarm intelligence algorithm that simulates the foraging behavior of ants in nature, the ant colony system is widely used in multiple fields such as combinatorial optimization, path planning, and scheduling problems due to its excellent global search ability and good adaptability. Since electric vehicle scheduling can be modeled as a combinatorial optimization problem of vehicle-customer matching, the ant colony system can be applied to solve this problem. At the same time, due to its involvement of multiple data sources and constraints, electric vehicle scheduling is more challenging than traditional fuel vehicle scheduling. Therefore, a more effective ant colony system algorithm needs to be designed. Summary of the Invention

[0004] The main purpose of the present invention is to overcome the disadvantages and deficiencies of the prior art, and provide an electric vehicle scheduling method based on a multi-pheromone channel ant colony system. This electric vehicle scheduling method associates multi-source traffic data such as vehicle power and charging facilities in the application scenario; at the same time, applies the ant colony system algorithm to electric vehicle scheduling, establishes multiple pheromone channels, and controls the pheromone update direction based on the average absolute eigenvalue of the pheromone matrix to calculate the pheromone update factor, so as to optimize the electric vehicle scheduling performance.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] An electric vehicle scheduling method based on a multi-pheromone channel ant colony system, the electric vehicle scheduling method includes the following steps:

[0007] S1. Perform power reachability modeling, define vehicle v j Meet customer ci Two scenarios of power constraints, where Scenario 1: The vehicle itself has enough power to send customer c i to the destination, and the service can be completed without charging, which is expressed as:

[0008] (1)

[0009] where e j is the current remaining power of vehicle v j , D(e j ) is the maximum driving distance that the current remaining power e j can support, t i is the destination of customer c i , d(c i , v j ) is the distance between customer c i and vehicle v j , d(c i , t i ) is the distance between the departure of customer c i and the destination t i , d(t i , cp(t i )) is the distance between the destination t i and the nearest charging pile, and cp(t i ) is the location of the nearest charging pile near the destination t i ; In this case, vehicle v j does not need to charge midway, and the driving path of vehicle v j is: v j → c i → t i → cp(t i );

[0010] Scenario 2: Midway charging is required to meet the service path. If the current remaining power e j of vehicle v j is not enough to complete the service path, but it can reach the nearest charging pile for charging, then midway charging is required. In this case, it is necessary to first go to the nearest charging pile p k to fully charge vehicle v j , and then continue to execute the service. At this time, the driving path of vehicle v j is: v j → p k → c i → t i , and Scenario 2 needs to meet the following conditions:

[0011] (1)Reachable charging pile: Vehicle v jMust be able to reach the nearest charging pile. The specific electricity availability conditions are as follows:

[0012] (2)

[0013] d(v j , p k ) is the distance between the vehicle v j and the nearest charging pile p k . At this time, the maximum driving distance that can be supported by the current remaining electricity e j must be sufficient to support the vehicle v j to reach the charging pile p k from its current position;

[0014] (2)Charging and service time limit: The charging time and service time of the vehicle v j should meet the maximum waiting time tolerance of the customer. The sum of the charging time and service time shall not exceed the maximum waiting time tolerated by the customer. The sum of the charging time and service time is:

[0015] (3)

[0016] where e need is the electricity required to complete the service, d(p k , c i ) is the distance between the nearest charging pile p k and the customer c i , cr is the charging efficiency, s is the driving speed of the vehicle v j , e need -e j is the difference between the electricity required to complete the service e need and the current remaining electricity e j , indicating that additional electricity needs to be charged to complete the current service. D(e need -e j ) is the maximum driving distance that can be supported by the additional electricity needed to be charged for the current service, and tl is the maximum waiting time tolerated by the customer;

[0017] S2. Model the customer's waiting time, which is divided into the following two cases. Case 1: The vehicle v j does not need to be charged. Assuming it departs from the current position and directly goes to the position of the customer c i , the customer's waiting time is calculated by the following formula:

[0018] (4)

[0019] Case 2: The vehicle v j needs to be charged. The customer's waiting time Including vehicle v j The charging and driving times, expressed as:

[0020] (5)

[0021] S3. Conduct a bi-objective optimization modeling, introducing a binary decision variable x ij ∈ {0, 1}, which represents whether to assign vehicle v j to customer c i , x ij = 1 for assignment; x ij = 0 for non-assignment; The solution of the ant is represented as a 0-1 matrix, which can not only simplify the implementation of the optimization algorithm, but also significantly improve the operation efficiency; The parallelism and batch processing ability of matrix operations make it more efficient when solving large-scale problems. Especially in the scheduling problems involving a large number of ants, customers and electric vehicles, the matrix representation method can effectively shorten the calculation time and improve the optimization performance; The main objective is to maximize the total number of successfully completed customer service requests, expressed as follows:

[0022] (6)

[0023] where S is the solution, f1() is the main objective function, f1(S) is the result of the solution S calculated under the main objective function f1(), max is the operation of taking the maximum value, N is the number of customers, and NP is the number of electric vehicles;

[0024] On the basis of maximizing the service matching number, introduce a secondary objective to minimize the overall customer waiting time of the system:

[0025] (7)

[0026] where f2() is the secondary objective function, f2(S) is the result of the solution S calculated under the secondary objective function f2(), min is the operation of taking the minimum value, and x ij ∙ is the product of the decision variable and the customer waiting time;

[0027] The constraint conditions to be satisfied are:

[0028] (8)

[0029] (9)

[0030] This model demonstrates significant advantages in multi-objective optimization environments. By introducing the dual objectives of matching rate and customer waiting time, it avoids the one-sidedness of a single metric and makes the model more aligned with the actual operational needs of electric vehicle dispatch tasks. Furthermore, the model exhibits excellent scalability and can be flexibly expanded to three-objective or even multi-objective scenarios (such as introducing service fairness and vehicle energy consumption) to meet optimization needs in more complex dispatch environments.

[0031] Among them, constraint (8) ensures that at most one customer can be assigned to one car, and constraint (9) ensures that at most one car can be assigned to one customer;

[0032] S4. Perform initialization settings and set the initial pheromone τ0 = 1 / (N ∙ min(LD)), where min(LD) is the smallest element in LD. This setting method ensures that the initial pheromone value has a uniform scale and is linked to the scheduling domain size and the shortest service cost to prevent the pheromone from being too large or too small and affecting the search process; set the heuristic pheromone matrix Heu = 1 / (LD +1), where LD is the customer c i With car v j The logical distance matrix between them, LD + 1 can avoid the denominator being zero, while retaining the sensitivity of logical distance to path selection, guiding ants to give priority to the path of "close distance and high service willingness"; i,j is the element in row i and column j of the logical distance matrix LD, and is calculated as:

[0033] (10)

[0034] Among them, NA is a null value;

[0035] S5. Initialize the multi-channel pheromone matrix, design the pheromone channels in four dimensions, the service willingness pheromone matrix Pher1, the waiting time feedback pheromone matrix Pher2, the power feasibility pheromone matrix Pher3, the global reinforcement pheromone matrix Pher g , the dimensions of the four pheromone matrices are all N × NP, and the initial pheromone of the four pheromone matrices is uniformly set to τ0;

[0036] S6. For each ant, in each round of allocation, according to the logical distance matrix LD and the current remaining power status of the car, the ant that is most suitable for customer c is selected. i The set of reachable and serviceable candidate electric vehicles J i ,This process is dynamically updated in each ant and each round of customer allocation,,ensuring that the search path is always within the feasible solution space;

[0037] S7. Calculate the probability of a candidate electric vehicle being assigned to a customer, select a suitable vehicle for the corresponding customer according to the state transition rule, and calculate the pheromone matrix Pher of this matching scheme using the following formula:

[0038] (11)

[0039] The multi-pheromone channel fusion mechanism can achieve better scheduling and allocation results in multi-objective optimization scenarios such as service willingness, power feasibility, and customer waiting time. By comprehensively considering the "preference characteristics" reflected in each pheromone matrix, this mechanism realizes multi-objective trade-off and collaborative guidance in the path selection process, thereby improving the quality of the overall scheduling solution and the achievement degree of service objectives.

[0040] S8. Execute the local pheromone update rule. The local update mechanism prevents a certain path from being monopolized early and guides the ants to explore more solutions:

[0041] (12)

[0042] where ξ is the local pheromone update rate;

[0043] S9. Evaluate the fitness value, evaluate the fitness value of each ant in the ant colony system algorithm and record the optimal solution. Considering the optimization objectives of the matching rate and customer waiting time, when evaluating two scheduling results, first compare the f1(S) of the two scheduling results, and the solution with the larger value is better; if the two scheduling results have the same f1(S) value, then compare the f2(S) of the two scheduling results, and the solution with the larger f2(S) value is better; this discriminant rule with clear priority strengthens the identification ability of the global optimal solution, enabling the accurate selection of the optimal solution among multiple feasible solutions; at the same time, this mechanism enhances the goal orientation of the algorithm, guiding the search process to give priority to improving the service matching rate while effectively controlling the customer waiting time, achieving a balance between service efficiency and user experience, helping to guide the ants to focus on the high-quality solution area, thereby optimizing the convergence path and search accuracy of the algorithm and improving the overall solution performance.

[0044] S10. Execute the global pheromone update rule. By gradually accumulating pheromones, significantly enhance the attractiveness of high-quality matching schemes in the solution space;

[0045] S11. End condition: When the maximum number of generations is reached, end the electric vehicle scheduling; otherwise, go to step S6 and continue the optimization.

[0046] Further, in step S8, ξ is set to 0.1, that is, the pheromone only decreases by 10% after each local update, maintaining a certain degree of pheromone residue to avoid rapid erasure of path information. During the process of individual path construction, the pheromone intensity decays moderately to prevent premature convergence of path selection, which can improve the stability and continuity of the algorithm in the local path exploration stage;

[0047] Further, in step S10, ω1, ω2, ω3, ω g are all set to 0.25 to ensure equal weights for each optimization objective during path selection and avoid solution deviation caused by bias towards a certain objective; AAEPM and ε are introduced to provide reliable and dynamic convergence criteria to effectively prevent premature convergence. ε is set to 0.01, and a smaller threshold is set to judge the convergence degree of the pheromone matrix. When AAEPM is lower than this value, it is considered that the system has approached a stable state; ρ is set to 0.1, enabling historical excellent solutions to guide the search for a long time and improving the quality and stability of the solutions.

[0048] Further, the formula for the global pheromone update rule is defined as follows:

[0049] (13)

[0050] (14)

[0051] (15)

[0052] Among them, ω m is the weight parameter of the pheromone channel, AAEPM m is the average absolute eigenvalue of the pheromone matrix, eig() is the operation of taking eigenvalues, abs() is the operation of taking absolute values, mean() is the operation of taking averages, ψ m is the pheromone update factor guided by AAEPM m , ε is a set threshold parameter, and ρ is the pheromone evaporation rate; by introducing AAEPM, the distribution characteristics of the pheromone matrix are comprehensively characterized, which can improve the perception ability of the pheromone evolution trend. According to the scale differences of pheromone evolution in different scheduling scenarios, the convergence states of each channel are dynamically judged; further combined with the feedback adjustment mechanism, the intensity of its ψ m is adaptively adjusted according to the change amplitude of AAEPM m of each channel, making the pheromone update process more focused on active channels, thereby realizing more effective search direction guidance and resource allocation optimization.

[0053] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0054] (1) The present invention uses the ant colony system algorithm to associate multi-source data to solve the scheduling model: Since the electric vehicle scheduling model is a complex discrete combinatorial optimization problem, it is necessary to comprehensively consider multi-source traffic data such as vehicle battery power and charging facilities, and the ant colony system algorithm is considered for solution. Each step of the ant finding its way corresponds to finding a suitable electric vehicle for each customer. The specific process is to first calculate the candidate vehicle set corresponding to each customer and the probability of allocating the candidate vehicle to this customer. Then, the ant selects a vehicle according to the state transition rule.

[0055] (2) The present invention uses multi-pheromone channels: The pheromone is no longer expressed in the traditional one-dimensional form, but is extended to a multi-dimensional matrix structure, which is used to describe four scheduling characteristics: service willingness, waiting feedback, battery feasibility, and global reinforcement. This setting method improves the accuracy and flexibility of the scheduling system, enabling it to respond more accurately to user needs and environmental changes, optimize service quality and user experience. Through refined scheduling, the system can reduce the user waiting time and achieve efficient resource allocation at the global level. At the same time, the system can monitor the battery state in real time to avoid service interruption caused by insufficient battery power. The structure of the multi-pheromone channels also has strong scalability, adapts to a variety of different application scenarios, and improves the versatility and efficiency of the scheduling system.

[0056] (3) The present invention uses the average absolute eigenvalue of the pheromone matrix to guide pheromone update: By real-time monitoring the evolution state of the pheromone in each channel, the average absolute eigenvalue of the pheromone matrix can be used to dynamically adjust the update intensity and fusion weight of the pheromone channel, realize fine control of the ant colony behavior, improve the solution ability of the ant colony algorithm, and enable it to search more effectively and find the optimal solution in complex problems.

[0057] (4) The present invention uses matrix operations to reconstruct some operators: For the core calculation module in the algorithm, the present invention introduces matrix operations to reconstruct key operators such as path selection, probability calculation, and pheromone update. By converting the original element-by-element loop operations into vectorized or matrixized processing, the calculation efficiency and parallel processing ability are significantly improved. This not only effectively reduces the running time of the algorithm in large-scale scheduling scenarios, but also enhances the scalability and stability of the algorithm in high-dimensional data and large-scale ant colony environments. Description of the Drawings

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained according to these drawings.

[0059] Figure 1It is an example diagram of electric vehicle scheduling in an embodiment of the present invention;

[0060] Figure 2 It is a flowchart of an electric vehicle scheduling method based on a multi-pheromone channel ant colony system in an embodiment of the present invention. Detailed implementation manners

[0061] In order to enable those skilled in the art of this technology to better understand the solution of this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative efforts fall within the protection scope of this application.

[0062] Referring to "embodiment" in this application means that the specific features, structures or characteristics described in connection with the embodiment may be included in at least one embodiment of this application. The phrase appears in various positions in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art explicitly and implicitly understand that the embodiments described in this application can be combined with other embodiments.

[0063] Embodiment 1

[0064] This embodiment introduces an electric vehicle scheduling method based on a multi-pheromone channel ant colony system, and simulates the scheduling tasks under different resource scenarios. The specific test data is shown in Table 1.

[0065] Table 1. Test data under different resource scenarios

[0066]

[0067] The operation steps of the electric vehicle scheduling method based on the multi-pheromone channel ant colony system are as follows:

[0068] S1. Perform modeling according to the test data and calculate LD.

[0069] S2. Perform initialization settings, set the initial pheromone τ0 = 1 / (N ∙ min(LD)), and set the heuristic pheromone matrix Heu = 1 / (LD + 1).

[0070] S3. Initialize the multi-channel pheromone matrix, design four-dimensional pheromone channels, the service willingness pheromone matrix Pher1, the waiting time feedback pheromone matrix Pher2, the power feasibility pheromone matrix Pher3, and the global reinforcement pheromone matrix Pher g, all four pheromone matrices are of dimension N × NP, and the initial pheromone of the four pheromone matrices is uniformly set to τ0.

[0071] S4. For each ant, during each round of allocation, based on the logical distance matrix LD and the remaining battery state of the current vehicle, filter out the set J of candidate electric vehicles that are reachable and serviceable for customer c i reachable and serviceable candidate electric vehicle set J i .

[0072] S5. Calculate the probability of the candidate electric vehicle for this customer according to the pheromone matrix Pher, and select a suitable vehicle for the corresponding customer according to the state transition rule.

[0073] S6. After all ants have completed the allocation, which means the end of this round of allocation, execute the local pheromone update rule.

[0074] S7. After all ants have constructed a complete solution, evaluate the solution quality and execute the global pheromone update rule.

[0075] S8. Determine whether the maximum number of iterations, 100 generations, has been reached. If so, terminate the algorithm and output the result; otherwise, return to step S4 to continue the optimization.

[0076] To compare and illustrate the advantages of the method of this application over the prior art, this invention is named AMPACS, and it is compared with the existing ant colony algorithm MACS on the test group data. Each method runs independently 10 times and the average value is taken. The final experimental results are shown in Table 2, and the optimal results are marked in bold:

[0077] Table 2. Experimental comparison result table of Example 1

[0078]

[0079] As can be seen from Table 2, in test groups 1, 2, 3, 4, 7, and 8, the matching rates of AMPACS and MACS are the same, but the average waiting time of AMPACS is lower than that of MACS. In test groups 5 and 6, the matching rate of AMPACS is higher than that of MACS, which indicates that the method proposed in this invention can achieve better results in the main objective, and when the main objective is the same as that of the comparative algorithm, it can achieve better results in the secondary objective. Under comprehensive comparison, the method proposed in this invention can find a better solution.

[0080] Example 2

[0081] To simulate the scenario of customers gathering in a real - world scenario, based on the scenario of Example 1, the K - Means clustering method is used to group user locations, and local random perturbations are performed within each clustering cluster, so as to simulate the uncertainty of the real - world customer request order while maintaining the overall geographical distribution characteristics. The number of clusters K is set to 5. The AMPACS with the shuffling mechanism is called AMPACS - s, and it is compared with AMPACS. The implementation steps refer to Example 1. The final experimental results are shown in Table 3, and the optimal results are marked in bold:

[0082] Table 3. Experimental comparison results table of Example 2

[0083]

[0084] As can be seen from Table 3, the results of AMPACS - s in test groups 1, 2, 3, 4, 5, and 8 are better than those of AMPACS, indicating that in the scenario of simulating customer gathering, the method proposed by the present invention shows better solution performance.

[0085] Example 3

[0086] In this embodiment, the calculation method of the initial pheromone τ0 in implementation step S2 is changed to τ0 = 1 / (N ∙ max(LD)), where max(LD) is the largest element in LD. The remaining implementation steps refer to Example 1.

[0087] When τ0 = 1 / (N ∙ min(LD)), this setting is based on the shortest effective path distance, making the initial pheromone relatively large, which is beneficial to enhancing the ant's preference for short paths and improving the early path convergence speed. The AMPACS using this setting method is called AMPACS - 1.

[0088] When τ0 = 1 / (N ∙ max(LD)), this setting is based on the longest effective path distance, making the initial pheromone relatively small, which helps to delay convergence, enhance search diversity, and improve the global exploration ability. The AMPACS using this setting method is called AMPACS - 2.

[0089] Compare AMPACS - 1 and AMPACS - 2. The final experimental results are shown in Table 4, and the optimal results are marked in bold:

[0090] Table 4. Experimental comparison results table of Example 3

[0091]

[0092] As can be seen from Table 4, AMPACS-2 performs better in Test Groups 1, 2, 3, 4, 5, 6, and 7. It is recommended to set the calculation method of τ0 to 1 / (N ∙ max(LD)). Under this setting method, the method proposed by the present invention can find a better solution.

[0093] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0094] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications made without departing from the spirit and principle of the present invention shall be equivalent replacement methods and are all included in the protection scope of the present invention.

Claims

1. An electric vehicle scheduling method based on a multi-pheromone channel ant colony system, characterized in that, The electric vehicle scheduling method includes the following steps: S1. Conduct power reachability modeling and define vehicle v j to satisfy customer c i with two scenarios that meet the power constraint. Among them, Scenario 1: The vehicle itself has enough power to send customer c i to the destination and can complete the service without charging, which is expressed as: （1） where, e j is the current remaining power of vehicle v j , D(e j ) is the maximum driving distance that the current remaining power e j can support, t i is the destination of customer c i , d(c i , v j ) is the distance between customer c i and vehicle v j , d(c i , t i ) is the distance between the departure of customer c i and the destination t i , d(t i , cp(t i )) is the distance between the destination t i and the nearest charging pile, and cp(t i ) is the location of the nearest charging pile near the destination t i ; in this case, vehicle v j does not need to charge midway, and the driving route of vehicle v j is: v j → c i → t i → cp(t i ); Scenario 2: Recharge during the service to meet the service route. If the current remaining power e j of vehicle v j is not enough to complete the service route but can reach the nearest charging pile for charging, then mid-way charging is required. In this case, it is necessary to first go to the nearest charging pile p k to fully charge vehicle v j , and then continue to execute the service. At this time, the driving route of vehicle v j is: v j → p k → c i → t i . Scenario 2 needs to meet the following conditions: (1) Reachable charging pile: Vehicle v j must be able to reach the nearest charging pile. The specific power reachability conditions are as follows: （2） d(v j , p k ) is the distance between vehicle v j and the nearest charging pile p k , where the maximum driving distance that can be supported by the current remaining battery level e j must be sufficient to support driving from the current location of vehicle v j to the charging pile p k ; (2) Charging and service time limit: Vehicle v j The time required for charging and the service time shall meet the maximum waiting time tolerance of the customer. The sum of the charging time and the service time shall not exceed the maximum waiting time tolerated by the customer. The sum of the charging time and the service time is: （3） Among them, e need is the electricity required to complete the service, d(p k , c i ) is the distance between the nearest charging pile p k and the customer c i ; cr is the charging efficiency, s is the driving speed of the vehicle v j ; e need -e j is the difference between the electricity required e need to complete the service and the current remaining electricity e j , indicating that charging is needed to supplement the electricity to complete the current service. D(e need -e j ) is the maximum driving distance that can be supported by the electricity to be charged to complete the current service; tl is the maximum waiting time tolerated by the customer. S2. Model the customer waiting time, which is divided into the following two cases. Case 1: Vehicle v j does not need to be charged. Assume that starting from the current location, it directly goes to the location of customer c i . The waiting time of the customer is calculated by the following formula: （4） Case 2: Vehicle v j needs to be charged, and the waiting time of the customer includes Vehicle v j the charging and driving time, expressed as: （5） S3. Conduct bi-objective optimization modeling and introduce a binary decision variable x ij ∈ {0, 1}, which indicates whether to allocate vehicle v j to customer c i . When x ij = 1, it is allocated; when x ij = 0, it is not allocated. The main objective is to maximize the total number of successfully completed customer service requests, which is expressed as follows: （6） Where S is the solution, f1() is the main objective function, f1(S) is the result of the solution S calculated under the main objective function f1(), max is the operation of taking the maximum value, N is the number of customers, and NP is the number of electric vehicles; On the basis of maximizing the service matching number, a secondary objective is introduced to minimize the overall customer waiting time of the system: （7） Among them, f2() is the sub-objective function, f2(S) is the result of the solution S under the calculation of the sub-objective function f2(), min is the minimum value operation, and x ij ∙ is the product of the decision variable and the customer waiting time; The constraint conditions to be satisfied are: （8） （9） Among them, constraint condition (8) ensures that at most one customer can be assigned to one vehicle, and constraint condition (9) ensures that at most one vehicle can be assigned to one customer; S4. Perform initialization settings, set the initial pheromone τ0 = 1 / (N ∙ min(LD)), where min(LD) is the minimum element in LD, and set the heuristic pheromone matrix Heu = 1 / (LD + 1), where LD is the logical distance matrix between customer c i and vehicle v j , and ld i,j is the element in the i-th row and j-th column of the logical distance matrix LD, and the calculation method is as follows: （10） Where NA is a null value; S5. Initialize the multi-channel pheromone matrix, design pheromone channels in 4 dimensions, the service willingness pheromone matrix Pher1, the waiting time feedback pheromone matrix Pher2, the power feasibility pheromone matrix Pher3, and the global reinforcement pheromone matrix Pher g , and the 4 pheromone matrices are all N × NP in dimension. Uniformly set the initial pheromone of the 4 pheromone matrices to τ0; S6. For each ant, in each round of allocation, according to the logical distance matrix LD and the remaining power state of the current vehicle, filter out the candidate electric vehicle set J i reachable and serviceable for customer c i ; S7. Calculate the probability of the candidate electric vehicle for this customer, select a suitable vehicle for the corresponding customer according to the state transition rule, and calculate the pheromone matrix Pher of this matching scheme using the following formula: （11） S8. Execute the local pheromone update rule: （12） Where ξ is the local pheromone update rate; S9. Evaluate the fitness value, evaluate the fitness value of each ant in the ant colony system algorithm and record the optimal solution. Considering the optimization objectives of the matching rate and the customer waiting time comprehensively, when evaluating two scheduling results, first compare the f1(S) of the two scheduling results, and the solution with the larger value is better; if the two scheduling results have the same f1(S) value, then compare the f2(S) of the two scheduling results, and the solution with the larger f2(S) value is better; S10. Execute the global pheromone update rule; S11. End condition: When the maximum number of generations is reached, end the electric vehicle scheduling; otherwise, go to step S6 and continue the optimization.

2. The electric vehicle scheduling method based on the multi-pheromone channel ant colony system according to claim 1, wherein In step S8 described above, ξ is set to 0.

1.

3. The electric vehicle scheduling method based on the multi-pheromone channel ant colony system according to claim 1, wherein, In the said step S10, ω1, ω2, ω3, ω g are all set to 0.25, ε is set to 0.01, and ρ is set to 0.

1.

4. The electric vehicle scheduling method based on the multi-pheromone channel ant colony system according to claim 1, characterized in that, The formula definition of the global pheromone update rule is as follows: （13） （14） （15） Among them, ω m is the weight parameter of the pheromone channel, and AAEPM m is the average absolute eigenvalue of the pheromone matrix, eig() is the eigenvalue extraction operation, abs() is the absolute value extraction operation, mean() is the average value extraction operation, and ψ m is the pheromone update factor guided by AAEPM m The parameter ε is a set threshold parameter, and ρ is the pheromone evaporation rate.

Images