Interactive ant colony optimization method and system for multi-compartment vehicle scheduling

Abstract

The invention discloses an interactive ant colony optimization method for multi-cabin vehicle scheduling, and the method comprises the steps: constructing a multi-cabin vehicle scheduling problem optimization model according to a multi-cabin vehicle transportation process; respectively generating a population 1 and a population 2 by adopting a preset encoding and decoding mode according to multiple rules; initializing pheromone matrixes 1 and 2 according to the populations 1 and 2; generating a first population and a second population based on the pheromone matrixes 1 and 2; evaluating the first population according to a first optimization target in the optimization model to obtain an external file set 1; evaluating the second population according to a second optimization target in the optimization model to obtain an external file set 2; interacting the external file set 1 and the external file set 2 to obtain a third population; evaluating the third population according to a first optimization target and a second optimization target in the optimization model to determine a non-inferior solution set and performing local operation to obtain an external file set 3; respectively updating the pheromone matrix 1 and the pheromone matrix 2 by using the individuals of the external file set 3; and outputting the current optimal non-inferior solution until the end condition is reached. Compared with a traditional optimization method, the method has remarkable advantages.

Description

Technical Field

[0001] The present invention relates to an interactive ant colony optimization method and system for multi-compartment vehicle scheduling, and belongs to the technical field of intelligent optimization scheduling of multi-compartment vehicle routing problems. Background Art

[0002] In recent years, with the rapid development of industries such as petrochemical and transportation, the scheduling of multi-compartment vehicles in the transportation process has become a problem to be solved. For example, as the social demand for fuel products continues to grow, in order to meet the precise distribution needs of customers for different types of oil products such as gasoline and diesel, special transport vehicles equipped with multiple independent oil tank compartments have been widely used. The routing planning of such vehicles is a key link in the refined oil supply chain. If scientific optimization can be achieved, it can not only help energy enterprises reduce transportation costs and improve service response speed, but also effectively improve the safety factor and resource utilization rate of hazardous chemical transportation. Therefore, researching intelligent scheduling strategies for multi-compartment oil tank trucks has significant commercial value and social benefits.

[0003] Nowadays, with the increasingly fierce competition in the logistics industry, in order to enhance the competitiveness and sustainable development of enterprises, in addition to pursuing low costs, customer satisfaction has become an important means for logistics enterprises to improve their competitive advantages. Therefore, in addition to the goal of transportation costs, customer satisfaction should also be optimized as an enterprise's daily goal. In the process of transporting multi-compartment vehicles such as fuel, the multi-compartment vehicle routing problem with the optimization goal of minimizing transportation costs and maximizing total customer satisfaction belongs to the NP-Hard problem, and it is impossible to obtain the optimal solution for such complex problems in a short time. Therefore, an efficient and reliable algorithm needs to be designed to solve this problem. Summary of the Invention

[0004] The present invention provides an interactive ant colony optimization method and system for multi-compartment vehicle scheduling. On the one hand, a scheduling model is established with the optimization goal of minimizing the total transportation cost and maximizing the total customer satisfaction. On the other hand, a dual-population cooperative search mechanism is adopted for global search, and a variable neighborhood descent algorithm is used for local search operations in combination with the problem characteristics; through the method of the present invention, a high-quality non-dominated solution set for the multi-compartment vehicle scheduling problem in the process of fuel transportation and the like can be obtained in a short time.

[0005] The technical solution of the present invention is as follows:

[0006] According to the first aspect of the present invention, there is provided an interactive ant colony optimization method for multi-compartment vehicle scheduling, including the following steps:

[0007] Step 1. Based on the multi-compartment vehicle transportation process, construct an optimization model for the multi-compartment vehicle scheduling problem; the specific multi-compartment vehicle transportation process is as follows: each multi-compartment vehicle starts from the cargo collection center, returns to the cargo collection center after serving each delivery customer, and at the same time, the load of each vehicle compartment meets the capacity constraint.

[0008] Step 2. Set the main parameters.

[0009] Step 3. Encode the numbers of the cargo collection center and the delivery customers.

[0010] Step 4. Generate population 1 and population 2 with a population size of popsize respectively according to the multi-rules using the encoding and decoding method in Step 3; initialize pheromone matrix 1 based on population 1, and initialize pheromone matrix 2 based on population 2.

[0011] Step 5. Generate the first population based on pheromone matrix 1; evaluate the individuals in the first population according to the first optimization objective in the multi-compartment vehicle scheduling problem optimization model, and use the Pareto non-dominated relationship to determine the non-dominated solution set, which is stored in the external archive set 1; generate the second population based on pheromone matrix 2; evaluate the individuals in the second population according to the second optimization objective in the multi-compartment vehicle scheduling problem optimization model, and use the Pareto non-dominated relationship to determine the non-dominated solution set, which is stored in the external archive set 2.

[0012] Step 6. Interact the external archive set 1 and the external archive set 2 to obtain the third population.

[0013] Step 7. Evaluate the individuals in the third population according to the first and second optimization objectives in the multi-compartment vehicle scheduling problem optimization model, use the Pareto non-dominated relationship to determine the non-dominated solution set and perform local operations. In the local operations, evaluate according to the first and second optimization objectives in the multi-compartment vehicle scheduling problem optimization model to obtain the updated non-dominated solution set, which is stored in the external archive set 3; use the individuals in the external archive set 3 to update pheromone matrix 1 and pheromone matrix 2 respectively.

[0014] Step 8. Determine whether the termination condition is met: if it is met, output the current optimal non-dominated solution; otherwise, jump to Step 5 and continue to execute.

[0015] Furthermore, the multi-rules include: the first rule and the second rule; the specific method of generating population 1 and population 2 with a population size of popsize respectively according to the multi-rules is as follows: generate population 1 with a population size of popsize according to the first rule and the second rule, and generate population 2 with a population size of popsize according to the first rule and the second rule; among them, the number of individuals generated according to the first rule in population 1 is equal to the number of individuals generated according to the first rule in population 2, the second rule for generating population 1 is established using the first optimization objective, and the second rule for generating population 2 is established using the second optimization objective.

[0016] Furthermore, the optimization model for the multi-compartment vehicle scheduling problem is specifically as follows: taking the minimization of the total transportation cost as the first optimization objective and the maximization of the total customer satisfaction as the second optimization objective, and constructing constraint conditions.

[0017] Furthermore, the first rule is the random rule and the second rule is the greedy rule; the greedy objective adopted by the greedy rule for generating population 1 is the first optimization objective, and the greedy objective adopted by the greedy rule for generating population 2 is the second optimization objective; the number of individuals generated according to the greedy rule in population 1 and population 2 is 1.

[0018] Furthermore, the interaction between the external archive set 1 and the external archive set 2 is specifically as follows: merging the external archive set 1 and the external archive set 2 as the third population.

[0019] Furthermore, the local operation is specifically as follows: successively adopting the "Insert", "Swap", and "Reverse" neighborhood operations based on the inside / outside of the sub-path, and adopting the variable neighborhood descent strategy during the execution, that is, if the current individual becomes better when performing any one of the neighborhood operations in the local operation, then the individual will re-execute from the "Insert" neighborhood operation; otherwise, continue to execute the next neighborhood operation until all neighborhood operations are completed.

[0020] According to the second aspect of the present invention, there is provided an interactive ant colony optimization system for multi-compartment vehicle scheduling, including the module of the interactive ant colony optimization method for multi-compartment vehicle scheduling described in any one of the above.

[0021] According to the third aspect of the present invention, there is provided a processor, which is used to run a program, wherein the program, when running, executes the interactive ant colony optimization method for multi-compartment vehicle scheduling described in any one of the above.

[0022] According to the fourth aspect of the present invention, there is provided a computer-readable storage medium, which includes a stored program, wherein, when the program runs, it controls the device where the computer-readable storage medium is located to execute the interactive ant colony optimization method for multi-compartment vehicle scheduling described in any one of the above.

[0023] The beneficial effects of the present invention are as follows: Firstly, the present invention adopts the greedy rule and the random rule for population initialization, which improves the diversity of the population; Secondly, the evolution mechanism of two independent ant colonies is adopted, which can optimize each single objective simultaneously and effectively search both ends of the Pareto front; At the same time, an interaction mechanism based on the ant colonies is designed, which enables the offspring of the two independent groups to gradually approach the center of the Pareto front instead of always concentrating at both ends of the Pareto front; Finally, a variable neighborhood descent algorithm with three neighborhood operators is adopted for local search, which improves the search depth of the algorithm. This method can obtain a high-quality non-dominated solution set for the multi-compartment vehicle scheduling problem in the transportation process of fuels, etc. in a short time, providing a high-quality decision-making basis for transportation enterprises. Further, through simulation verification, it is further shown that the method of the present invention has significant advantages over traditional optimization methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 is the overall algorithm flow chart of the present invention;

[0025] Figure 2 is the schematic diagram of the multi-compartment vehicle scheduling path of the present invention;

[0026] Figure 3 is the schematic diagram of the "Insert", "Swap" and "Reverse" neighborhood operations based on the inside of the sub-path of the present invention;

[0027] Figure 4 is the schematic diagram of the "Insert", "Swap" and "Reverse" neighborhood operations based on the outside of the sub-path of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0028] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention. It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments can be combined with each other arbitrarily.

[0029] Embodiment 1: As Figures 1-4 shown, an interactive ant colony optimization method for multi-compartment vehicle scheduling includes the following steps:

[0030] Step1. According to the multi-compartment vehicle transportation process, construct an optimization model for the multi-compartment vehicle scheduling problem; The multi-compartment vehicle transportation process is specifically as follows: Each multi-compartment vehicle starts from the goods collection center, returns to the goods collection center after serving each transportation customer, and at the same time, the load of each compartment of the vehicle satisfies the capacity constraint;

[0031] Step 2: Set the main parameters. Among them, the main parameters include the population size popsize, the pheromone importance factor α, the heuristic function importance factor β, and the pheromone evaporation coefficient ρ.

[0032] Exemplarily, the initialization of the main parameters includes: obtaining the values of the main parameters through the orthogonal experimental design method. In the example of the present invention, the values of the four main parameters are: popsize = 30, α = 2, β = 1, ρ = 0.5.

[0033] Step 3: Use decimal coding for the numbering of the collection center and the delivery customers.

[0034] Step 4: Generate population 1 and population 2 with the population size popsize respectively according to the multi - rules using the encoding and decoding method of Step 3; initialize the pheromone matrix 1 according to population 1, and initialize the pheromone matrix 2 according to population 2.

[0035] Furthermore, the multi - rules include: the first rule and the second rule; the generation of population 1 and population 2 with the population size popsize respectively according to the multi - rules is specifically: generate population 1 with the population size popsize according to the first rule and the second rule, and generate population 2 with the population size popsize according to the first rule and the second rule; among them, the number of individuals generated according to the first rule in population 1 is equal to the number of individuals generated according to the first rule in population 2, the second rule for generating population 1 is established using the first optimization objective, and the second rule for generating population 2 is established using the second optimization objective.

[0036] Taking the generation of population 1 as an example for further illustration: initialize the first preset number of individuals using the first rule, and initialize the second preset number of individuals using the second rule to obtain the initial population of popsize individuals; where the sum of the first preset number and the second preset number is the population size popsize. Exemplarily, the first rule is the random rule, and the second rule is the greedy rule. Exemplarily, the first preset number is 29, and the second preset number is 1. After simulation, 1 is generated under the greedy rule, and 29 are generated under the random rule, with the following advantages: on the one hand, it can make the population diversity better, and on the other hand, it can make the interactive operation of the algorithm reach a better - quality area for search faster, improving the convergence of the algorithm.

[0037] Each multi - compartment vehicle starts from the collection center, serves each delivery customer and then returns to the collection center. Due to the limitation of the vehicle compartment capacity constraint, it must serve the customers and return to the collection center under the premise of meeting the constraints. Therefore, the driving route of each vehicle constitutes 1 sub - route, and the driving routes of multiple multi - compartment vehicles involved in the multi - compartment vehicle transportation process constitute an individual. For example, refer to Figure 2, the solution (individual) formed by 9 customers is π = [1, 2, 6, 3, 4, 1, 7, 5, 1, 8, 10, 9, 1], indicating that vehicles 1, 2, and 3 serve customers 2, 6, 3, 4, customers 7, 5, and customers 8, 10, 9 in sequence, and then return to the collection center.

[0038] Step5. Generate the first population based on pheromone matrix 1; evaluate the individuals in the first population according to the first optimization objective in the multi-compartment vehicle scheduling problem optimization model, and use the Pareto non-dominated relationship to determine the non-dominated solution set, which is stored in the external archive set 1; generate the second population based on pheromone matrix 2; evaluate the individuals in the second population according to the second optimization objective in the multi-compartment vehicle scheduling problem optimization model, and use the Pareto non-dominated relationship to determine the non-dominated solution set, which is stored in the external archive set 2;

[0039] Step6. Interact the external archive set 1 and the external archive set 2 to obtain the third population;

[0040] Step7. Evaluate the individuals in the third population according to the first and second optimization objectives in the multi-compartment vehicle scheduling problem optimization model, use the Pareto non-dominated relationship to determine the non-dominated solution set and perform local operations. In the local operations, evaluate according to the first and second optimization objectives in the multi-compartment vehicle scheduling problem optimization model to obtain the updated non-dominated solution set, which is stored in the external archive set 3; and use the individuals in the external archive set 3 to update the pheromone matrix 1 and the pheromone matrix 2 respectively;

[0041] Step8. Judge whether the termination condition is satisfied: If satisfied, output the current optimal non-dominated solution; otherwise, jump to Step5 to continue execution.

[0042] Furthermore, the multi-compartment vehicle scheduling problem optimization model is specifically: taking minimizing the total transportation cost and maximizing the total customer satisfaction as the optimization objectives, and constructing constraint conditions.

[0043] Furthermore, the optimization objectives are specifically:

[0044] The first optimization objective: Minimize the total transportation cost Z1, which is calculated by the following formula:

[0045]

[0046] In the formula, Z1 represents the total transportation cost, c1 represents the unit distance coefficient, and d ij represents the distance between customer i and customer j; x ijk is a decision variable. If vehicle k travels from customer i to customer j, its value is 1, otherwise it is 0.

[0047] The second optimization objective: Maximize the total customer satisfaction function, and the expression is as follows:

[0048]

[0049] In the formula, Z2 represents the total customer satisfaction, and μ i (t i ) represents the customer satisfaction corresponding to the time when the vehicle arrives at customer i at time t i ;

[0050] The customer satisfaction function is as follows:

[0051]

[0052] In the formula, t i represents the time when the vehicle arrives at customer i; e i represents the scheduled start visit time with customer i; l i represents the scheduled end visit time with customer i; E i represents the earliest time to extend the visit to customer i, that is, the lower limit of the time window representing customer i; L i represents the latest time to extend the visit to customer i, that is, the upper limit of the time window of customer i.

[0053] Furthermore, the constraint conditions of the multi-compartment vehicle scheduling problem optimization model include: the first constraint condition: each customer can only be served by one vehicle and can only be served once (i.e., equations (4) and (5)); the second constraint condition: each vehicle departs from the center, serves all customers, and returns to the collection center (i.e., equation (6)); the third constraint condition: flow balance, after the vehicle serves the current customer, it needs to drive out to serve another customer (i.e., equation (7)); the fourth constraint condition: the current compartment load is equal to the sum of the product demand quantities corresponding to all customers served by the vehicle plus the product demand quantity corresponding to the current served customer (i.e., equation (8)); the fifth constraint condition: the load capacity of each vehicle compartment cannot exceed the rated maximum load capacity of each compartment (i.e., equation (9)); the sixth constraint condition: the sum of the waiting time and service time of each vehicle at the current customer plus the driving time to reach the next customer does not exceed the maximum opening time of the collection center (i.e., equation (10)); the seventh constraint condition: the service time and waiting time of the vehicle at the collection center are 0 (i.e., equation (11)); the eighth constraint condition: the time when the vehicle arrives at the next customer is equal to the time when the vehicle arrives at the current customer plus the waiting time and service time for the current customer and the driving time to reach the next customer (i.e., equation (12)); the ninth constraint condition: the service time period of the vehicle (i.e., equation (13)); the tenth constraint condition: the waiting time of the vehicle (i.e., equation (14)); the eleventh constraint condition: used to eliminate sub-circuits (i.e., equation (15)); the twelfth constraint condition: used for the value range of decision variables (i.e., equation (16)).

[0054] The above-mentioned formulas are as follows:

[0055]

[0056] where; x ijk is a decision variable, which is 1 if vehicle k travels from customer i to customer j, and 0 otherwise; K = {1, 2,..., m} represents the set of vehicles, and m represents the total number of vehicles; N = N c ∪{0} represents the set of the collection center and customers, 0 represents the collection center, and N c = {1, 2, 3,..., n} represents the set of customers, and n represents the total number of customers; Q ipk represents the load of product p after vehicle k serves customer i, and q jp represents the delivery volume of product p required by customer j. The product set p = {1, 2,..., P}, and P represents the maximum number of product types required by customers (i.e., the number of vehicle compartments); Q p represents the maximum load of the p-th compartment; w i represents the waiting time spent by the vehicle when serving customer i, and s i represents the service time required by customer i, and r ij represents the driving duration from customer i to customer j, and g k represents the maximum opening duration of the collection center; w0 and s0 represent the waiting time and service time spent by the vehicle at the collection center; t i represents the time when the vehicle arrives at customer i, and t j represents the time when the vehicle arrives at customer j; E i represents the lower limit of the time window of customer i, and L i represents the upper limit of the time window of customer i.

[0057] Furthermore, the greedy rule is initialized, and the specific steps are as follows:

[0058] 1), Construct the customers to be served into a candidate customer set, and each vehicle departs from the collection center;

[0059] 2), Select the next customer point served by the vehicle from the candidate customer set according to the current greedy goal;

[0060] 3), Judge whether the products required by the selected customer exceed the rated load of each compartment of the vehicle; if not, serve the customer and remove it from the candidate customer set; otherwise, do not serve the customer point and return to the collection center;

[0061] 4), Judge whether the set of customers to be served is empty. If it is empty, it means that all customers have been served; otherwise, return to 2) and continue to execute.

[0062] Further, the greedy objective is to minimize the total transportation cost or maximize the total customer satisfaction; in the embodiments of the present invention, minimizing the total transportation cost is used as the greedy objective to construct population 1, and maximizing the total customer satisfaction is used as the greedy objective to construct population 2. Applying the above technical solutions, it can be known that in the present invention, different greedy objectives are used to initialize an individual in population 1 and population 2, which can improve the overall quality and individual diversity of the initial population. In addition, these individuals are used to initialize the pheromone matrix, and at the beginning of the optimization, they can guide the search towards the region where the target values are low transportation costs and high satisfaction levels, that is, the region of high-quality non-dominated solutions.

[0063] Further, the initialization is performed according to the following specific steps for the random rule:

[0064] S1. Construct the candidate customer set with the customers to be served, and each vehicle departs from the collection center;

[0065] S2. Randomly select a customer from the candidate customer set as the next customer to be served by the current vehicle;

[0066] S3. Determine whether the products required by the selected customer exceed the rated load of each compartment of the vehicle; if not, serve the customer and remove it from the set of customers to be served; otherwise, dispatch the next vehicle;

[0067] S4. Determine whether the set of customers to be served is empty. If it is empty, it means that all customers have been served; otherwise, return to S2 and continue to execute.

[0068] Further, the pheromone matrix 1 and the pheromone matrix 2 have the same structure, and the structure of the pheromone matrix is defined as follows:

[0069]

[0070] where τ gen,f (n + 1, n + 1) is an element in, that is, the element in the (n + 1)-th row and the (n + 1)-th column; represents the f-th pheromone matrix at the gen-th generation; represents the probability that customer j appears after customer i in the f-th pheromone matrix at the gen-th generation, and n represents the total number of customers.

[0071] The update of the pheromone matrix is specifically as follows:

[0072]

[0073] where represents the pheromone concentration accumulated on the edge where customer j is after customer i is served in the (gen + 1)-th generation; is the pheromone concentration; Denote the pheromone concentration increment, which is calculated by equations (19) and (20) respectively:

[0074]

[0075] Among them, denotes the customer satisfaction corresponding to and denotes the transportation cost between customers corresponding to ; denotes the non - dominated solution set of the gth generation of population 1, denotes the non - dominated solution set of the gth generation of population 2

[0076] Furthermore, to generate population 1 and population 2 based on the pheromone matrix, it is necessary to continuously calculate the transfer probability between customers according to the pheromone matrix, and continuously select the next service customer according to the transfer probability, repeating in a cycle until all customers are served, and then a new individual is generated. The transfer probability is calculated according to the following formula:

[0077]

[0078] In the above formula (21), N j is a set storing unvisited customers, and α and β respectively represent the importance degrees of the pheromone heuristic function and The expressions of are as shown in equations (22) and (23):

[0079]

[0080] Among them, k is a positive constant, which is taken as 1000 in the embodiment of the present invention.

[0081] In addition, to maintain the diversity of individuals, first, try to make the first visited customers selected by the vehicle different. Secondly, a random factor is added to the selection of ants. A constant q0 and a random variable q are introduced, which are distributed in the interval [0, 1]. When q > q0, the transfer probability is selected by roulette, and vice versa, the greedy rule is used for selection.

[0082] Furthermore, the interaction between the external archive set 1 and the external archive set 2 is specifically: merging the external archive set 1 and the external archive set 2 as the third population.

[0083] Furthermore, the local operation is specifically as follows: successively adopt the "Insert", "Swap", and "Reverse" neighborhood operations based on the inside / outside of the sub-path. During the execution process, adopt the variable neighborhood descent strategy, that is, if the current individual improves when performing any one of the neighborhood operations in the local operation, then the individual will re-execute from the "Insert" neighborhood operation; otherwise, continue to execute the next neighborhood operation until all neighborhood operations are completed. Specifically, the operations based on the inside / outside of the sub-path are determined randomly. For example, the local operation is specifically to successively adopt the "Insert" based on the inside of the sub-path, the "Swap" based on the inside of the sub-path, and the "Reverse" neighborhood operation based on the outside of the sub-path. In the above, when an individual improves when performing any one of the neighborhood operations, that is, evaluated according to the first and second optimization objectives in the multi-compartment vehicle scheduling problem optimization model, when both of the two optimization objectives corresponding to the individual before and after performing a certain neighborhood operation are optimized, it means improvement.

[0084] To verify the effectiveness of the method proposed in the present invention, it is experimentally compared with the classical mainstream multi-objective optimization algorithms MOEAD, SGAII, and SPEA2.

[0085] To ensure the fairness of the comparison, the above 4 algorithms all take the same running time of 5*n as the termination condition, and solve the RC series examples in the Solomon VRP with Time Windows (VRPTW) standard dataset. All algorithms use an Intel i5 processor (2.5 GHZ), 16 GB of memory, a Win10 operating system, and a matlab2022b programming environment. And when solving each example, they all run independently 20 times, and two indicators as shown in formulas (24) and (25) are used for evaluation:

[0086]

[0087] Among them, PF * is the set of points on the first Pareto front. PF represents the final non-dominated solution set obtained by an algorithm. d i represents the minimum inverted generational distance from the i-th point in PF * to the individuals in PF. v i represents the hypervolume formed by the reference point (the worst solution of the PF of all algorithms selected in the present invention) and the i-th individual in PF. Obviously, the smaller the value of IGD and the larger the value of HV, the better the performance of the algorithm. The approximate optimal Pareto front of the present invention is composed of the individuals on the first front selected by combining the Pareto front individuals obtained by each algorithm.

[0088] The statistical results are shown in Table 1. The bold values in the table represent the optimal values of each example. For the HV index, the larger the better, and for the IGD index, the smaller the better. It can be seen from Table 1 that the test results of the algorithm of the present invention are better than those of the other three comparison algorithms in most examples, thus verifying the effectiveness of the present invention.

[0089] Table 1

[0090]

[0091] In addition, to verify the effectiveness of the key interaction mechanism in the method proposed in the present invention, it was compared with the variant IACOA-V1 that does not utilize this co-evolution mechanism (that is, the external archive set 1 and external archive set 2 obtained in Step 5 are used to update the pheromone matrices 1 and 2 respectively). That is, the rest of IACOA-V1 is the same as that of the present invention, but it only has two populations evolving independently without interaction. The comparison on different examples is shown in Table 2. It can be seen from Table 2 that the ant colony algorithm with the interactive mechanism used in the present invention has significant advantages, verifying the effectiveness of the key operations of the present invention.

[0092] Table 2

[0093]

[0094] Embodiment 2: An interactive ant colony optimization system for multi-compartment vehicle scheduling, including the module of the interactive ant colony optimization method for multi-compartment vehicle scheduling described in any one of the above, specifically including: The first module is used to execute Step1: According to the multi-compartment vehicle transportation process, construct an optimization model for the multi-compartment vehicle scheduling problem; the multi-compartment vehicle transportation process is specifically as follows: Each multi-compartment vehicle starts from the cargo collection center, returns to the cargo collection center after serving each delivery customer, and at the same time the load of each vehicle compartment meets the capacity constraint; The second module is used to execute Step2: Set the main parameters; The third module is used to execute Step3: Encode the numbers of the cargo collection center and the delivery customers; The fourth module is used to execute Step4: Respectively generate population 1 and population 2 with a population size of popsize according to multiple rules using the encoding and decoding methods of Step3; Initialize pheromone matrix 1 according to population 1, and initialize pheromone matrix 2 according to population 2; The fifth module is used to execute Step5: Based on pheromone matrix 1, generate the first population; Evaluate the individuals in the first population according to the first optimization objective in the multi-compartment vehicle scheduling problem optimization model, and use the Pareto non-dominated relationship to determine the non-dominated solution set, which is stored in the external archive set 1; Based on pheromone matrix 2, generate the second population; Evaluate the individuals in the second population according to the second optimization objective in the multi-compartment vehicle scheduling problem optimization model, and use the Pareto non-dominated relationship to determine the non-dominated solution set, which is stored in the external archive set 2; The sixth module is used to execute Step6: Interact the external archive set 1 and the external archive set 2 to obtain the third population; The seventh module is used to execute Step7: Evaluate the individuals in the third population according to the first and second optimization objectives in the multi-compartment vehicle scheduling problem optimization model, use the Pareto non-dominated relationship to determine the non-dominated solution set and perform local operations, and evaluate according to the first and second optimization objectives in the multi-compartment vehicle scheduling problem optimization model during the local operations to obtain the updated non-dominated solution set and store it in the external archive set 3; Use the individuals in the external archive set 3 to update pheromone matrix 1 and pheromone matrix 2 respectively; The eighth module is used to execute Step8: Determine whether the termination condition is met: If it is met, output the current optimal non-dominated solution; Otherwise, jump to Step5 and continue to execute. For the parts not detailed in the above modules, reference can be made to the relevant descriptions of other embodiments.

[0095] Embodiment 3: A processor, the processor is used to run a program, wherein, when the program runs, it executes the interactive ant colony optimization method for multi-compartment vehicle scheduling described in any one of the above.

[0096] Embodiment 4: A computer-readable storage medium, the computer-readable storage medium includes a stored program, wherein, when the program runs, it controls the device where the computer-readable storage medium is located to execute the interactive ant colony optimization method for multi-compartment vehicle scheduling described in any one of the above.

[0097] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to: various media such as USB flash drives, read-only memory (ROM), random access memory (RAM), mobile hard disks, magnetic disks, or optical discs that can store computer programs.

[0098] The specific embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings. However, the present invention is not limited to the above embodiments, and various changes can be made without departing from the spirit of the present invention within the scope of knowledge possessed by those of ordinary skill in the art.

Claims

1. An interactive ant colony optimization method for multi-compartment vehicle scheduling, characterized in that It includes the following steps: Step 1. Based on the multi-compartment vehicle transportation process, construct an optimization model for the multi-compartment vehicle scheduling problem. The multi-compartment vehicle transportation process is specifically as follows: Each multi-compartment vehicle starts from the cargo collection center, serves each delivery customer, and then returns to the cargo collection center. At the same time, the load of each vehicle compartment meets the capacity constraint. Step 2. Set the main parameters. Step 3. Encode the cargo collection center and the delivery customers. Step 4. Generate population 1 and population 2 with a population size of popsize respectively using the encoding and decoding method in Step 3 according to multiple rules. Initialize pheromone matrix 1 based on population 1, and initialize pheromone matrix 2 based on population 2. Step 5. Generate the first population based on pheromone matrix 1. Evaluate the individuals in the first population according to the first optimization objective in the optimization model for the multi-compartment vehicle scheduling problem, and use the Pareto non-dominated relationship to determine the non-dominated solution set, which is stored in the external archive set 1. Generate the second population based on pheromone matrix 2. Evaluate the individuals in the second population according to the second optimization objective in the optimization model for the multi-compartment vehicle scheduling problem, and use the Pareto non-dominated relationship to determine the non-dominated solution set, which is stored in the external archive set 2. Step 6. Interact external archive set 1 and external archive set 2 to obtain the third population. Step 7. Evaluate the individuals in the third population according to the first and second optimization objectives in the optimization model for the multi-compartment vehicle scheduling problem, use the Pareto non-dominated relationship to determine the non-dominated solution set and perform local operations. In the local operations, evaluate according to the first and second optimization objectives in the optimization model for the multi-compartment vehicle scheduling problem to obtain the updated non-dominated solution set, which is stored in the external archive set 3. Use the individuals in the external archive set 3 to update pheromone matrix 1 and pheromone matrix 2 respectively. Step 8. Judge whether the termination condition is met: If it is met, output the current optimal non-dominated solution. Otherwise, jump to Step 5 and continue to execute.

2. The interactive ant colony optimization method for multi-compartment vehicle scheduling according to claim 1, characterized in that The multiple rules include: the first rule and the second rule. The generation of population 1 and population 2 with a population size of popsize respectively according to multiple rules is specifically as follows: Generate population 1 with a population size of popsize according to the first rule and the second rule, and generate population 2 with a population size of popsize according to the first rule and the second rule. Among them, the number of individuals generated according to the first rule in population 1 is equal to the number of individuals generated according to the first rule in population 2. The second rule for generating population 1 is established using the first optimization objective, and the second rule for generating population 2 is established using the second optimization objective.

3. The interactive ant colony optimization method for multi-compartment vehicle scheduling according to claim 2, characterized in that, The optimization model for the multi-compartment vehicle scheduling problem is specifically as follows: Take minimizing the total transportation cost as the first optimization objective and maximizing the total customer satisfaction as the second optimization objective, and construct the constraint conditions.

4. The interactive ant colony optimization method for multi-compartment vehicle scheduling according to claim 3, wherein, The first rule is the random rule, and the second rule is the greedy rule. The greedy objective used in the greedy rule for generating population 1 is the first optimization objective, and the greedy objective used in the greedy rule for generating population 2 is the second optimization objective. The number of individuals generated according to the greedy rule in population 1 and population 2 is 1.

5. The interactive ant colony optimization method for multi-compartment vehicle scheduling according to claim 1, characterized in that The interaction between external archive set 1 and external archive set 2 is specifically as follows: Combine external archive set 1 and external archive set 2 as the third population.

6. The interactive ant colony optimization method for multi-compartment vehicle scheduling according to claim 1, characterized in that, The specific local operation is as follows: successively adopt the "Insert", "Swap" and "Reverse" neighborhood operations based on the inside / outside of the sub-path. During the execution process, adopt the variable neighborhood descent strategy, that is, if the current individual is improved when performing any one of the neighborhood operations in the local operation, then the individual will re-execute from the "Insert" neighborhood operation; otherwise, continue to execute the next neighborhood operation until all neighborhood operations are completed.

7. An interactive ant colony optimization system for multi-compartment vehicle scheduling, characterized in that, A module including the interactive ant colony optimization method for multi-compartment vehicle scheduling according to any one of claims 1-6.

8. A processor, characterized in that, The processor is used to run a program, wherein when the program runs, it executes the interactive ant colony optimization method for multi-compartment vehicle scheduling according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein when the program runs, it controls the device where the computer-readable storage medium is located to execute the interactive ant colony optimization method for multi-compartment vehicle scheduling according to any one of claims 1-6.

Images