WO2024032376A1

WO2024032376A1 - Vehicle path optimization method based on hybrid genetic algorithm, and application thereof

Info

Publication number: WO2024032376A1
Application number: PCT/CN2023/109506
Authority: WO
Inventors: 姚陈潇
Original assignee: 姚陈潇
Priority date: 2022-08-11
Filing date: 2023-07-27
Publication date: 2024-02-15
Also published as: CN115330051A; CN115330051B

Abstract

The present invention provides a vehicle path optimization method based on a hybrid genetic algorithm and an application thereof, which relate to the technical field of combination optimization of cargo loading and vehicle paths. The method comprises: setting a target fitness function as a target evaluation indicator of a vehicle path scheme for evaluating the fitness of chromosomes in a genetic algorithm; obtaining cargo information and vehicle information, and executing a step of establishing sequence information and a step of generating a feasible scheme, so as to determine a vehicle path scheme; and in the step of generating a feasible scheme, computing target fitness function values corresponding to primary chromosomes to select Pareto solutions, computing a target fitness function value of the chromosomes in each iteration to compare Pareto solutions, and screening out a final Pareto solution and a corresponding vehicle path scheme. According to the method, a target fitness function is set according to actual operation conditions to evaluate vehicle path schemes generated after genetic operations, so as to screen out an optimal vehicle path scheme.

Description

Vehicle route optimization method and application based on hybrid genetic algorithm

Technical field

The present invention relates to the technical field of combined optimization of cargo loading and vehicle routing.

Background technique

As a combined optimization problem of cargo loading and vehicle routing, the circular pickup problem of auto parts in-factory logistics has always been one of the mainstream issues studied in the logistics industry.

In the problem of cyclic pickup in the logistics of auto parts entering the factory, a multi-frequency, small-batch cyclic pickup model is adopted. In order to meet the requirements of flexible production and inventory management of automobiles for parts pickup and transportation, in actual operations, a supplier is visited by multiple vehicles every day, the demand can be split and transported separately, multiple types of vehicles perform transportation tasks, and unloading is limited. time window, etc.

That is to say, in reality, there are a certain number of suppliers distributed throughout the geographical space. The distribution center (or the logistics company responsible for transportation management) organizes appropriate driving routes according to the production plan of the automobile manufacturer every day, and delivers the goods to each supplier. Collect parts of different quantities, sizes, and unloading time limits, use vehicles of different models to perform transportation tasks, and send the parts to the distribution center for unloading, so that the parts inventory in the distribution center can continue to meet the production plan of the automobile manufacturer. . At the same time, it can achieve the purpose of minimizing the total transportation cost and minimizing the total transportation time under certain constraints, such as vehicle loading constraints, the divisible supply demand of a single supplier, and limited vehicle types and quantities.

In the problem of recycling goods in the inbound logistics of auto parts, taking into account the total transportation cost and total transportation time, restrictions are placed on cargo loading and vehicle paths. The following limiting rules are set for cargo loading and vehicle paths:

For cargo loading, four assumptions need to be met: 1) the carriage and the goods to be loaded are both rectangular parallelepipeds; 2) the goods placed must be completely contained in the carriage; 3) the goods can only be loaded with edges parallel or perpendicular to the carriage Place the goods in the direction of the edge; 4) It is required that the goods can only be rotated around the height edge and cannot be placed upside down.

Five assumptions need to be met for vehicle routes: 1) Each route must start from the distribution center and finally return to the distribution center; 2) Each supplier can be visited more than once by the route; 3) Each route loads The goods should meet the three-dimensional restrictions of the carriage; 4) Each path can only be served by one vehicle; 5) All goods have a path for pickup and delivery.

In addition to the above restricted rules, it is also necessary to ensure that in the process of solving the recycling problem of auto parts inbound logistics, the goal is to minimize the total inventory and transportation costs and achieve resource optimization in vehicle transportation, distribution and other links.

To this end, providing a vehicle route optimization method and application based on a hybrid genetic algorithm to obtain feasible, multiple vehicle route solutions that consider cargo loading is an urgent technical problem that needs to be solved.

Contents of the invention

The purpose of the present invention is to overcome the shortcomings of the existing technology and provide a vehicle path optimization method and application based on a hybrid genetic algorithm. The present invention can set a target fitness function according to actual operating conditions to evaluate the vehicle path scheme generated after genetic operations. , to finally select the optimal vehicle route plan.

In order to solve the existing technical problems, the present invention provides the following technical solutions:

A vehicle route optimization method based on hybrid genetic algorithm, including the following steps:

Set the target fitness function as the target evaluation index of the vehicle routing plan to evaluate the adaptability of chromosomes in the genetic algorithm;

Obtain cargo information and vehicle information, execute the steps of establishing sequence information and generating feasible solutions to determine the vehicle route solution; in the step of generating feasible solutions, calculate the target fitness function value corresponding to the first-generation chromosome to select the Pareto solution, and calculate each In an iterative process, the target fitness function value of the chromosome is used to compare the Pareto solutions, and the final Pareto solution and its corresponding vehicle route plan are determined after eliminating the bad and retaining the good; the vehicle route plan includes all paths and the corresponding loading plan.

Further, the target fitness function includes the total transportation cost costR and the total transportation time timeR of the vehicle route plan.

Further, the chromosome is compiled based on integer coding. The chromosome includes the first half of the cargo segment and the second half of the vehicle segment. The cargo segments are based on parts packaging containers, and one box is regarded as one gene. When the goods are combined After forming into pallets, combine one pallet into one box.

Further, in the step of establishing sequence information, corresponding to the collected cargo information, vehicle model information and supplier node information, a cargo sequence I_list, a vehicle model sequence K_list and a supplier node sequence S_list are established. The supplier node sequence in the supplier node sequence S_list is The goods are arranged according to their position information in the goods sequence I_list.

Further, the steps of generating feasible solutions include: Step S201, generate an initial population of goods + vehicle models; Step S202, based on the chromosome information in the initial population, correspondingly obtain the vehicle route plan, and calculate the target fitness function value of each chromosome in the initial population, Select the Pareto solution and its corresponding vehicle routing scheme, and write the new Pareto solution and its corresponding vehicle routing scheme into the Pareto optimal solution set; step S203, perform genetic operations on each chromosome in the initial population of the aforementioned goods + vehicle models, Generate the cargo offspring population; based on the chromosome information in the cargo offspring population, correspondingly obtain the vehicle route plan, calculate the target fitness function value of each chromosome in the cargo offspring population, compare the Pareto solutions, and obtain the new Pareto solution and its corresponding The vehicle route plan is written into the Pareto optimal solution set; step S204, determine whether the termination condition is met, and if the termination condition is reached, continue to step S205; otherwise, return to step S202; step S205, output the Pareto optimal solution set and For the corresponding vehicle route plan, the output information includes the supplier sequence, loading plan, transportation time _ti , total transportation cost costR and total transportation time timeR information on the route.

Further, when the vehicle route plan is obtained, the following steps are included: Step S2021, establish the relationship between goods and supplier nodes, obtain the goods sequence I_list, vehicle model sequence K_list and supplier node sequence S_list, initialize iL=1, iK=1, where, iL represents the iL-th cargo, iK represents the iK-th vehicle model, and sets the cargo sequence L_list to be loaded; step S2022, if iK≤length(K_list), retrieve the iK-th vehicle model from the vehicle model sequence K_list, otherwise, retrieve the iK-th vehicle model from the vehicle model sequence K_list. length(K_list) vehicle types, and obtain the compartment volume CV; step S2023, take out the iL-th cargo from the cargo sequence I_list and add it to the cargo sequence L_list to be loaded. When the sum of the volumes of the cargo in the cargo sequence to be loaded is greater than the compartment volume CV, execute Step S2024, otherwise, execute step S2025; step S2024, use the cargo sequence to be loaded and the carriage information corresponding to the vehicle model as parameters, call the preset simulated annealing algorithm, generate a loading plan, and follow the position order of the loaded cargo in the cargo sequence I_list , arrange the corresponding supplier node S_list sequence, remove duplicate supplier nodes, add a distribution center at the start node and end node of the supplier node sequence S_list, generate a path, store the path in the vehicle routing scheme, and calculate The path has been loaded with the earliest delivery time window of the goods and whether each goods is late α _i . Finally, the loaded goods are deleted from the sequence of goods to be loaded, and iK=iK+1 is set; step S2025, when iL≤length(I_list) When , set iL=iL+1 and execute step S2023; if the sequence of goods to be loaded L_list is empty, go to step S2026, otherwise, return to step S2024; step S2026, output the vehicle route plan, for any i∈R in the aforementioned vehicle route plan , the information of each R _i path includes the supplier sequence, packing scheme, Transportation time t _i , whether the loaded goods are late α _i .

Further, the α _i is a 0-1 variable. For the cargo i, i∈I, α _i =1 indicates that it is loaded on the vehicle, and α _i =0 indicates that it is not loaded on the vehicle.

Further, the genetic operation includes selection, crossover and mutation operations on the cargo + vehicle model chromosomes; the genetic operation independently performs selection, crossover and mutation operations on the cargo fragments and vehicle vehicle fragments on the chromosomes; the crossover probability of each of the aforementioned chromosomes or The aforementioned mutation probabilities can be set to the same value.

Further, the aforementioned Pareto solution is decoded to generate a vehicle route plan. The decoding includes the steps: Step S301, segmenting vehicle vehicle segments and cargo segments: For segmenting vehicle segment segments, the order is from the beginning to the end, and only one segment is segmented at a time. Car model, obtain the cabin volume of the car model; for segmented cargo fragments, the sequence is divided one by one from beginning to end, until the sum of the divided cargo volumes exceeds the cabin volume of the vehicle model, or the cargo fragments have been segmented Completed; step S302, perform loading verification: use the preset vehicle loading algorithm to verify the segmented vehicle models and goods and generate a loading plan; determine whether there are unloaded goods after verification, and determine if yes , put the unloaded goods back into the cargo fragments, and loop through steps S301 and S302 until all the goods are loaded into the corresponding vehicles; step S303, generate a vehicle route plan: according to the loaded goods of each vehicle , generate a vehicle route plan, and calculate the earliest unloading window time, total transportation cost and total transportation time for each route.

A vehicle route optimization system based on hybrid genetic algorithm, including:

The condition preset module sets vehicle loading constraints and vehicle path constraints, as well as corresponding target evaluation indicators in the vehicle path plan;

Information input module is used to input cargo information, vehicle model information and supplier node information;

The solution generation module is used to obtain cargo information and vehicle information, execute the steps of establishing sequence information and generating feasible solutions to determine the vehicle route solution; in the step of generating feasible solutions, calculate the target fitness function value corresponding to the first-generation chromosome to select Pareto solution, and calculate the target fitness function value of the chromosome in each iteration process to compare the Pareto solutions, and determine the final Pareto solution and its corresponding vehicle path plan after eliminating the bad and retaining the good; the vehicle path plan includes all paths and Corresponding loading plan.

Based on the above advantages and positive effects, the advantage of the present invention is that based on actual operational needs, a target fitness function is set to evaluate the vehicle route plan generated after the genetic operation, and the selection, crossover and mutation operations are independently performed in the genetic operation. , filter the descendants The better Pareto solution corresponding to the chromosome in the population is finally selected through multiple iteration processes to select the optimal vehicle route plan.

Furthermore, when setting the target fitness function, the loading constraints, path constraints, total transportation cost and total transportation time are also considered to facilitate the handling of actual transportation problems in real scenarios.

Description of drawings

Figure 1 is a flow chart provided by an embodiment of the present invention.

Figure 2 is an example diagram of a coding and decoding process provided by an embodiment of the present invention.

FIG. 3 is an example diagram of a vehicle route considering cargo loading constraints and multi-vehicle demand splitting provided by an embodiment of the present invention.

Figure 4 is an example diagram of a crossover operation in a genetic operation provided by an embodiment of the present invention.

Figure 5 is a diagram illustrating an example of mutation operation in genetic operation provided by the embodiment of the present invention.

Figure 6 is a schematic structural diagram of a system provided by an embodiment of the present invention.

Explanation of reference symbols:
System 100, condition presetting module 110, information input module 120, solution generation module 130.

Detailed ways

A vehicle route optimization method and application based on a hybrid genetic algorithm disclosed in the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that the technical features or combinations of technical features described in the following embodiments should not be considered isolated, and they can be combined with each other to achieve better technical effects. In the drawings of the following embodiments, the same reference numerals appearing in each drawing represent the same features or components and can be applied to different embodiments. Thus, once an item is defined in one figure, it does not need further discussion in subsequent figures.

It should be noted that the structures, proportions, sizes, etc. shown in the drawings attached to this specification are only used to match the content disclosed in the specification and are for the understanding and reading of people familiar with this technology. They are not used to limit the scope of the invention. Any structural modifications, changes in proportions, or adjustments in size, as long as they do not affect the effects that the invention can produce and the purposes that it can achieve, should fall within the scope of the technical content disclosed by the invention. within the range. The scope of the preferred embodiments of the present invention includes alternative implementations in which the order described or discussed may be different, including in a substantially simultaneous manner or in the reverse order depending on the functionality involved, To perform functions, this should be understood by those skilled in the art to which embodiments of the present invention belong.

Techniques, methods and devices known to those of ordinary skill in the relevant art may not be discussed in detail, but where appropriate, such techniques, methods and devices should be considered part of the authorized specification. In all examples shown and discussed herein, any specific values are to be construed as illustrative only and not as limiting. Accordingly, other examples of the exemplary embodiments may have different values.

In the present invention, the definitions of technical terms such as sets, parameters, and decision variables required for the solution method of the combined optimization problem of cargo loading and vehicle routing are expressed as follows:

1) Collection

I={1,…,n}: A collection of goods to be loaded, including n goods to be loaded.

S={0,1,…,m}: node set, where 0 is the distribution center node and 1,…,m are parts suppliers.

K＝{1,…,k}: car model set.

R={1,…,r}: Output result path unit set, in which there are r output result paths.

2) Parameter settings

The parameter settings of the mathematical model are shown in Table 1:

Table 1 Parameter settings for vehicle routing problem

3) Decision variables

For the combined optimization problem of cargo loading and vehicle routing, the decision variables of the mathematical model are v _ir , α _rl , w _rk and u _ijr , as defined in the following table:

Example

As shown in Figure 1, a vehicle route optimization method based on a hybrid genetic algorithm is provided, including the following implementation steps:

Step S1, set the target fitness function as the target evaluation index of the vehicle route plan. The standard is used to evaluate the adaptability of chromosomes in genetic algorithms.

In this embodiment, the target fitness function is the total transportation cost costR and the total transportation time timeR of the vehicle route plan.

Among them, the total transportation cost is expressed as:

The total transportation time is expressed as:

As a preferred implementation of this embodiment, taking into account vehicle vehicle allocation and demand splitting, the solution objectives of the combined optimization problem of vehicle loading and vehicle routing are set to the lowest total transportation cost costR and the lowest total transportation time timeR, as shown below:

and

Among them, the first goal is to minimize the total transportation cost costR. The first item of this goal is the sum of the transportation cost in transit and the cost of loading and unloading trucks for each route. The second item is the penalty cost caused by picking up the goods early; the second goal is transportation. The total time timeR is the least, and the transportation time is the sum of the transportation time in transit and the loading and unloading time at the node.

In view of the above solution goals, when solving the combined optimization problem of cargo loading and vehicle routing, the constraint conditions for vehicle loading and vehicle routing are set accordingly to satisfy the minimum total transportation cost costR and the total transportation time timeR under the condition of many constraints. least.

The above solution objectives need to satisfy both the vehicle loading constraints and the vehicle path constraints.

Vehicle loading constraints are divided into general constraints and special constraints; one type is general constraints: mainly volume constraints, constraints that the loaded goods are not embedded in each other, and constraints that the loaded goods and the carriage are not embedded; the other type is special constraints: There are mainly complete support constraints, single box load-bearing constraints, cargo stacking constraints, vehicle load-bearing constraints and center of gravity constraints.

The above vehicle loading constraints are existing technologies and will not be described in detail in this embodiment.

The mathematical expression corresponding to the vehicle path constraints is as follows:

a) A supplier node is visited by at least one pickup path, as shown in equation (1).

b) All goods from the supplier have a pickup path for transportation, as shown in equation (2).

c) All goods from the supplier need to be delivered to the distribution center for unloading, as shown in equation (3).

d) A pickup route can only be served by one vehicle, as shown in equation (4).

e) A supplier node is visited by at least one pickup path, as shown in equation (5).

f) All goods of the supplier have pickup paths for access, as shown in equation (6).

g) The difference in access time of two nodes on a path arc is equal to the transportation time between the two points, as shown in equation (7).

h) A pickup route can only be served by one vehicle, as shown in equation (8).

Specifically, the chromosome is compiled based on integer coding. The chromosome includes the first half of the cargo segment and the second half of the vehicle segment. The cargo segment is based on the parts packaging container, and one box is regarded as one gene. When the goods After combining into pallets, one combined pallet is used as one box.

As a preferred implementation of this embodiment, see Figure 2. In the cargo segment coding, numbers from 1 to 9 are used to indicate that there are 9 boxes of goods. Each box of goods has its corresponding supplier information, unloading time window information, and Basic loading information such as cargo size, cargo weight, cargo stacking rules, etc.

Among them, the 9th, 6th and 8th boxes of goods belong to supplier S1, the 7th, 5th, 4th and 1st boxes of goods belong to supplier S2, and the 3rd and 2nd boxes of goods belong to supplier S3; in the vehicle model fragment code, 11 to The number 13 indicates that there are 3 vehicles responsible for transportation tasks, corresponding to 8-meter, 12-meter and 8-meter models respectively.

Step S2, obtain cargo information and vehicle information, execute the steps of establishing sequence information and generating feasible solutions to determine the vehicle route solution; in the step of generating feasible solutions, calculate the target fitness function value corresponding to the first-generation chromosome to select the Pareto solution, and Calculate the target fitness function value of the chromosome in each iteration process to compare the Pareto solutions, and determine the final Pareto solution and its corresponding vehicle route plan after eliminating the bad and retaining the good; the vehicle route plan includes all paths and corresponding loads plan.

Specifically, in the step S100 of establishing sequence information, a cargo sequence I_list, a vehicle model sequence K_list, and a supplier node sequence S_list are established corresponding to the collected cargo information, vehicle model information, and supplier node information. In the supplier node sequence S_list The supplier node sequence is arranged correspondingly according to the position information of the goods in the goods sequence I_list.

Preferably, the generated feasible solution preferably uses a hybrid genetic algorithm to solve the combined optimization problem of cargo loading and vehicle routing, and adopts the chromosome encoding method of cargo + vehicle type to obtain the initial population of cargo + vehicle type.

Specifically, the step S200 of generating feasible solutions includes:

Step S201: Generate an initial population of goods + vehicle models.

It should be noted that the parameters of the genetic algorithm include: initial population, population size N, number of iterations T, crossover probability PC, mutation probability PM, etc., and the initial population generation process is as follows.

Assume that the population size N is 100, first randomly generate 100 sets of cargo sequences and 100 sets of vehicle model sequences, then add a starting point before each cargo in the cargo sequence, and add a starting point at the end of the cargo sequence, and use the same method to The vehicle vehicle sequence is processed, and finally a cargo sequence and vehicle vehicle sequence are spliced to form a chromosome in the population.

Step S202: Based on the chromosome information in the initial population, correspondingly obtain the vehicle route plan, calculate the target fitness function value of each chromosome in the initial population, select the Pareto solution and its corresponding vehicle route plan, and combine the new Pareto solution and its corresponding The vehicle routing plan is written into the Pareto optimal solution set.

Step S203, perform genetic operations on each chromosome in the initial population of cargo + vehicle models to generate a population of cargo offspring; based on the chromosome information in the population of cargo offspring, corresponding Obtain the vehicle route plan, calculate the target fitness function value of each chromosome in the cargo offspring population, compare the Pareto solutions, obtain the new Pareto solution and its corresponding vehicle route plan, and write it into the Pareto optimal solution set.

Step S204: Determine whether the termination condition is met. If the termination condition is met, continue to execute step S205. Otherwise, return to step S202.

In this embodiment, the maximum number of iterations is preferably used as the termination condition. When the genetic algorithm runs to the specified maximum number of iterations, the algorithm program iteration is terminated and the Pareto optimal solution set is output.

Step S205, output the Pareto optimal solution set and the corresponding vehicle route plan. The output information includes the supplier sequence, loading plan, transportation time _ti , total transportation cost costR and total transportation time timeR information on the route.

It should be emphasized that the hybrid genetic algorithm optimizes the vehicle model allocation and demand splitting scheme through genetic operations, and uses the vehicle loading algorithm to verify and divide the generated paths, and finally outputs the multi-objective Pareto optimal solution set to obtain Feasible vehicle routing options.

It should also be noted that in the aforementioned steps of generating feasible solutions, corresponding to obtaining the vehicle route solution, the following steps are included:

S2021, establish the relationship between goods and supplier nodes, obtain the goods sequence I_list, vehicle model sequence K_list and supplier node sequence S_list, initialize iL=1, iK=1, where iL represents the iL-th goods and iK represents the iK-th vehicle model , set the sequence of goods to be loaded L_list.

Step S2022, if iK≤length(K_list), retrieve the iK-th vehicle model from the vehicle model sequence K_list, otherwise, retrieve the length(K_list)-th vehicle model from the vehicle model sequence K_list, and obtain the cabin volume CV.

Step S2023, remove the iL-th cargo from the cargo sequence I_list and add it to the cargo sequence L_list to be loaded. When the sum of the volumes of the cargo in the cargo sequence to be loaded is greater than the compartment volume CV, step S2024 is executed. Otherwise, step S2025 is executed.

Step S2024, use the cargo sequence to be loaded and the carriage information corresponding to the vehicle model as parameters, call the preset simulated annealing algorithm, generate a loading plan, and arrange the corresponding supplier nodes S_list according to the position order of the loaded goods in the cargo sequence I_list. sequence, remove duplicate supplier nodes, add a distribution center at the start node and end node of the supplier node sequence S_list, generate a path, store the path in the vehicle routing scheme, and calculate the earliest delivery of goods loaded on the path The cargo time window and whether each cargo is late α _i are determined. Finally, the loaded cargo is deleted from the sequence of cargo to be loaded, and iK=iK+1 is set.

The simulated annealing algorithm refers to an algorithm that randomly adjusts the loading of goods during the simulated annealing process, generates new loading sequences, and selects the optimal loading plan by comparing the loading rates according to different loading sequences corresponding to different loading rates. .

In the preferred implementation of this embodiment, the content of calling the simulated annealing algorithm to generate the cargo loading plan is an existing technology, so it will not be elaborated in the preferred implementation of this embodiment.

It should be noted that when generating routes, a variety of situations need to be considered, as examples rather than limitations, for example: each supplier is allowed multiple visits per day, the same supplier has multiple paths to visit to pick up goods, each The daily cumulative supply of each supplier may exceed the capacity of one vehicle. Each vehicle may transport part of the supplied parts but must complete the transportation of all supplied parts on the same day.

As an example, see Figure 3. Multiple transportation paths can be generated corresponding to the transportation tasks performed by different models. For example: Path 1 is: a 12-meter truck departs from the distribution center at 6 o'clock on the 1st and travels to supplier 1 first. Pick up the goods at Supplier 2, then drive to Supplier 2 to pick up the goods, and finally send them to the distribution center for unloading; Route 2 is: an 8-meter truck departs from the distribution center at 12 o'clock on the 1st, first drives to Supplier 2 to pick up the goods, and then Drive to supplier 3 to pick up the goods, and finally send them to the distribution center for unloading.

Step S2025, when iL≤length(I_list), set iL=iL+1 and execute step S2023; if the sequence of goods to be loaded L_list is empty, proceed to step S2026, otherwise, return to step S2024.

Step S2026, output the vehicle route plan. For any i∈R in the aforementioned vehicle route plan, the information of each R _i path includes the supplier sequence, packing plan, transportation time t _i on the path, and whether the loaded goods are Late α _i . .

Among them, the α _i is a 0-1 variable. For the cargo i, i∈I, α _i =1 indicates that it is loaded on the vehicle, and α _i =0 indicates that it is not loaded on the vehicle.

Preferably, the genetic operation includes selection, crossover and mutation operations on the cargo+vehicle chromosomes; the genetic operation independently performs selection, crossover and mutation operations on the cargo fragments and vehicle vehicle fragments on the chromosomes; the crossover probability of each of the aforementioned chromosomes Or the aforementioned mutation probability can be set to the same value.

Wherein, the selection operation preferably adopts an elite retention strategy. Assume there are 100 chromosomes. First, select two chromosomes with the lowest total transportation cost and lowest total transportation time and retain them in the next generation population; then randomly pair the remaining 98 chromosomes. 49 pairs of new chromosomes were obtained; and crossover and mutation operations were performed on the new chromosomes.

The crossover operation preferably adopts the Inver-over method, that is, among the above-mentioned 49 pairs of chromosomes, a probability value is randomly generated for each pair of chromosomes. If the probability value corresponding to a pair of chromosomes is less than the crossover probability initially defined by the algorithm, this pair of chromosomes is used as the parent chromosome, and the crossover operation is performed to generate two offspring chromosomes.

As an example but not a limitation, see Figure 4. In the crossover operation, two parent chromosomes are randomly selected: P1 and P2 chromosomes, gene G1 is randomly selected from chromosome P1, and the location G1 of gene G1 is searched in chromosome P2. ', select the first gene G2 after gene G1'. If G1' is the last gene in chromosome P2, select the first gene in chromosome P2 as G2, find the G2 position G2' in chromosome P1, and match G1 together with The part between G1 and G2' is flipped and transformed, but the position of G2' remains unchanged. In the same way, crossover processing is performed on chromosome P2. The intersection operation using the Inver-over method has the characteristics of fast convergence speed and high accuracy.

The mutation operation randomly generates a probability value for each pair of chromosomes for the 98 chromosomes obtained after the crossover operation. If the probability value of a chromosome is less than the mutation probability initially defined by the algorithm, then this chromosome will be used as the parent chromosome, and a mutation operation will be performed to generate a offspring chromosome. If the two target fitness function values of the offspring chromosome are better than those of the parent chromosome If there are two target fitness function values, use the offspring chromosome to replace the parent chromosome; if the two target fitness function values of the offspring chromosome are worse than the two target fitness function values of the parent chromosome, continue to retain the parent chromosome chromosome.

The mutation operation preferably adopts a single gene replacement method. The mutation process of the single gene replacement method can randomly select two genes in the chromosome for position exchange.

As shown in Figure 5, assuming a parent chromosome P, randomly select a gene position G1 from the chromosome P, then search for G2 in a different position from G1 in the chromosome P, and then exchange the genes at the G1 and G2 positions in the chromosome P.

In addition, regarding the operation of removing the bad and saving the good, it should be noted that the corresponding solution goals of this embodiment are to minimize the total transportation cost and minimize the total transportation time. When solving the aforementioned goals, there are multiple optimal solutions, namely Pareto solutions, and their corresponding solution sets are Pareto optimal solution sets. Each Pareto solution cannot be compared with each other for all chromosomes. , so the target fitness function is set to evaluate the quality of chromosomes, the fitness function value corresponding to each chromosome is calculated, a Pareto solution is formed and selected.

That is, assuming that the new solution generated by the current iteration is S, you need to combine the S solution with the Pareto optimal All solutions in the solution set are compared with the target fitness function value. If all solutions in the Pareto optimal solution set are better than S, then S is abandoned; if there is some solution in the Pareto optimal solution set that S is better than, S is added to the Pareto optimal solution. set, and at the same time delete the solutions that are exceeded by all S in the Pareto optimal solution set.

Preferably, the aforementioned Pareto solution is decoded to generate a vehicle route solution, and the decoding includes the steps:

Step S301, segment vehicle models and cargo segments: For segmenting vehicle segments, the order is from beginning to end, and only one model is segmented at a time to obtain the cabin volume of that model; for segmenting cargo segments, the order is from beginning to end. direction, segment one by one until the sum of the volumes of the cargo that has been segmented this time exceeds the compartment volume of the vehicle model, or the cargo segments have been segmented.

Step S302, perform loading verification: use the preset vehicle loading algorithm to verify the segmented vehicle models and goods and generate a loading plan; determine whether there are unloaded goods after verification. If it is determined to be yes, The unloaded goods are put back into the goods fragments, and steps S301 and S302 are processed in a loop until all the goods are loaded into the corresponding vehicles.

Step S303, generate a vehicle route plan: generate a vehicle route plan based on the loaded goods of each vehicle, and calculate the earliest unloading window time, total transportation cost, and total transportation time for each route.

As an example but not a limitation, see Figure 2. For 9 boxes of cargo information and 3 vehicle model information, using the above decoding method, the decoding results are 096870541320 and 0130120110. Then the vehicle routing solution for this problem is 2 paths, that is, the One route uses an 8-meter vehicle, starting from the distribution center, driving to the supplier node S1 to pick up the 9th, 6th, and 8th goods, and then driving to the supplier node S2 to pick up the 7th goods, and then sent to the distribution center for unloading; The second route uses a 12-meter vehicle, starting from the distribution center, driving to the supplier node S2 to pick up the 5th, 4th, and 1st goods, and then driving to the supplier node S3 to pick up the 3rd and 2nd goods, and then sending them to distribution. Center unloading.

In this embodiment, improved encoding and decoding methods and hybrid genetic algorithms are used to realize and optimize vehicle vehicle allocation and demand splitting, and the vehicle loading algorithm is used for verification to generate a vehicle route plan.

For other technical features, please refer to the previous embodiments and will not be described again here.

In addition, as shown in Figure 6, the present invention also provides an embodiment, providing a vehicle route optimization system 100 based on a hybrid genetic algorithm. The system 100 includes conditions Preset module 110, information input module 120 and solution generation module 130.

The condition presetting module 110 sets vehicle loading constraint conditions and vehicle route constraint conditions, as well as corresponding target evaluation indicators in the vehicle route plan.

The information input module 120 is used to input cargo information, vehicle model information and supplier node information.

The solution generation module 130 is used to obtain cargo information and vehicle information, execute the step of establishing sequence information and the step of generating feasible solutions to determine the vehicle route solution; in the step of generating feasible solutions, calculate the target fitness function value corresponding to the first-generation chromosome to select Pareto solution, and calculate the target fitness function value of the chromosome in each iteration process to compare the Pareto solution, and determine the final Pareto solution and its corresponding vehicle path plan after eliminating the bad and retaining the good; the vehicle path plan includes all paths and the corresponding loading plan.

The system may also include a user interface module for collecting user input information and outputting information to the user. The user interface module includes a graphical user interface (GUI) for users to view and analyze results.

The system may also include other components typically found in computing systems, such as an operating system, a queue manager, a device driver, a database driver, or one or more network protocols stored in memory and executed by a processor.

In the above description, components may be selectively and operatively combined in any number within the intended scope of the present disclosure. In addition, terms like "includes," "includes," and "having" should be construed as inclusive or open by default, rather than exclusive or closed, unless expressly qualified to the contrary. All technical, scientific or other terms have the same meaning as understood by those skilled in the art unless limited to a contrary meaning. Common terms found in dictionaries should not be interpreted too ideally or too impractically in the context of the relevant technical documentation, unless the present disclosure explicitly limits them to that.

Although example aspects of the present disclosure have been described for illustrative purposes, those skilled in the art will appreciate that the foregoing description is merely a description of preferred embodiments of the present invention and is not intended to limit the scope of the present invention in any way. The scope of the embodiments includes additional implementations in which functions may be performed out of the order in which they appear or are discussed. Any changes or modifications made by those of ordinary skill in the field of the present invention based on the above disclosure shall fall within the protection scope of the claims.

Claims

A vehicle route optimization method based on a hybrid genetic algorithm, which is characterized by including the following steps:

Set the target fitness function as the target evaluation index of the vehicle routing plan to evaluate the adaptability of chromosomes in the genetic algorithm;

Obtain cargo information and vehicle information, execute the steps of establishing sequence information and generating feasible solutions to determine the vehicle route solution; in the step of generating feasible solutions, calculate the target fitness function value corresponding to the first-generation chromosome to select the Pareto solution, and calculate each In an iterative process, the target fitness function value of the chromosome is used to compare the Pareto solutions, and the final Pareto solution and its corresponding vehicle route plan are determined after eliminating the bad and retaining the good; the vehicle route plan includes all paths and the corresponding loading plan.
The method according to claim 1, characterized in that the target fitness function includes the total transportation cost costR and the total transportation time timeR of the vehicle route plan.
The method according to claim 1, characterized in that the chromosome is compiled based on integer coding, and the chromosome includes a first half cargo segment and a second half vehicle segment segment, wherein the cargo segment is in units of parts packaging containers. , treat one box as one gene, and when the goods are combined into pallets, one pallet will be combined into one box.
The method according to claim 1, characterized in that in the step of establishing sequence information, a cargo sequence I_list, a vehicle model sequence K_list and a supplier node sequence S_list are established corresponding to the collected cargo information, vehicle model information and supplier node information, so The order of the supplier nodes in the supplier node sequence S_list is correspondingly arranged according to the position information of the goods in the goods sequence I_list.
The method according to claim 1, characterized in that the step of generating feasible solutions includes:

Step S201, generate an initial population of goods + vehicle models;

Step S202: Based on the chromosome information in the initial population, correspondingly obtain the vehicle route plan, calculate the target fitness function value of each chromosome in the initial population, select the Pareto solution and its corresponding vehicle route plan, and combine the new Pareto solution and its corresponding The vehicle routing plan is written into the Pareto optimal solution set;

Step S203: Perform genetic operations on each chromosome in the initial population of cargo + vehicle models to generate a population of cargo offspring; based on the chromosome information in the population of cargo offspring, obtain the corresponding vehicle route plan and calculate each chromosome in the population of cargo offspring. Target fitness function value, compare the Pareto solution, obtain the new Pareto solution and its corresponding vehicle route plan, and write it into the Pareto optimal solution set;

Step S204, determine whether the termination condition is met. If the termination condition is met, continue to execute step S205. Otherwise, return to step S202;

Step S205, output the Pareto optimal solution set and the corresponding vehicle route plan. The output information includes the supplier sequence, loading plan, transportation time ti , total transportation cost costR and total transportation time timeR information on the route.
The method according to claim 5, characterized in that when correspondingly obtaining the vehicle route plan, it includes the steps:

Step S2021, establish the relationship between goods and supplier nodes, obtain the goods sequence I_list, vehicle model sequence K_list and supplier node sequence S_list, initialize iL=1, iK=1, where iL represents the iL-th goods and iK represents the iK-th goods Car model, set the sequence of goods to be loaded L_list; step S2022, if iK ≤ length (K_list), take out the iK-th car model from the car model sequence K_list, otherwise, take out the length (K_list)-th car model from the car model sequence K_list, and obtain the cabin volume CV ;

Step S2023, remove the iL-th cargo from the cargo sequence I_list and add it to the cargo sequence L_list to be loaded. When the sum of the volumes of the cargo in the cargo sequence to be loaded is greater than the compartment volume CV, step S2024 is executed. Otherwise, step S2025 is executed;

Step S2024, use the cargo sequence to be loaded and the carriage information corresponding to the vehicle model as parameters, call the preset simulated annealing algorithm, generate a loading plan, and arrange the corresponding supplier nodes S_list according to the position order of the loaded goods in the cargo sequence I_list. sequence, remove duplicate supplier nodes, add a distribution center at the start node and end node of the supplier node sequence S_list, generate a path, store the path in the vehicle routing scheme, and calculate the earliest delivery of goods loaded on the path The cargo time window and whether each cargo is late α i , finally delete the loaded cargo from the sequence of cargo to be loaded, and set iK=iK+1;

Step S2025, when iL≤length(I_list), set iL=iL+1 and execute step S2023; if the sequence of goods to be loaded L_list is empty, enter step S2026, otherwise, return to step S2024;

Step S2026, output the vehicle route plan. For any i∈R in the aforementioned vehicle route plan, the information of each R i path includes the supplier sequence, packing plan, transportation time t i on the path, and whether the loaded goods are Late α i .
The method according to claim 6, characterized in that the α i is a 0-1 variable. For the cargo i, i∈I, α i =1 indicates that it has been loaded on the vehicle, and α i =0 indicates that it has not been loaded on the vehicle. in the car.
The method of claim 5, wherein the genetic manipulation includes The selection, crossover and mutation operations of object + vehicle vehicle chromosomes; the genetic operation independently performs selection, crossover and mutation operations on the cargo fragments and vehicle vehicle fragments on the chromosomes; the respective crossover probabilities or the aforementioned mutation probabilities of the aforementioned chromosomes can be set to the same value.
The method according to claim 5, characterized in that the aforementioned Pareto solution is decoded to generate a vehicle route solution, and the decoding includes the steps:

Step S301, segment vehicle models and cargo segments: For segmenting vehicle segments, the order is from beginning to end, and only one model is segmented at a time to obtain the cabin volume of that model; for segmenting cargo segments, the order is from beginning to end. direction, segment one by one until the sum of the volume of the cargo that has been segmented this time exceeds the compartment volume of the vehicle model, or the cargo segments have been segmented;

Step S302, perform loading verification: use the preset vehicle loading algorithm to verify the segmented vehicle models and goods and generate a loading plan; determine whether there are unloaded goods after verification. If it is determined to be yes, The unloaded goods are put back into the cargo fragments, and steps S301 and S302 are processed in a loop until all the goods are loaded into the corresponding vehicles;

Step S303, generate a vehicle route plan: generate a vehicle route plan based on the loaded goods of each vehicle, and calculate the earliest unloading window time, total transportation cost, and total transportation time for each route.
A vehicle path optimization system based on a hybrid genetic algorithm for implementing any one of claims 1-9, characterized in that it includes:

The condition preset module sets vehicle loading constraints and vehicle path constraints, as well as corresponding target evaluation indicators in the vehicle path plan;

The information input module is used to input cargo information, vehicle model information and supplier node information; the solution generation module is used to obtain cargo information and vehicle information, and execute the steps of establishing sequence information and generating feasible solutions to determine the vehicle routing solution; after generating In the feasible solution step, the target fitness function value corresponding to the first-generation chromosome is calculated to select the Pareto solution, and the target fitness function value of the chromosome in each iteration is calculated to compare the Pareto solution, and the final Pareto solution is determined after eliminating the bad and retaining the good. and its corresponding vehicle route plan; the vehicle route plan includes all routes and corresponding loading plans.