Road train intelligent scheduling method based on improved ant colony algorithm

Abstract

The invention discloses a road train intelligent scheduling method based on an improved ant colony algorithm, and relates to the field of intelligent traffic and logistics optimization. The method comprises the following steps: firstly, constructing a four-stage transportation network model comprising a loading point, a transfer station and an unloading point, and establishing a dual-target scheduling model for simultaneously minimizing the comprehensive cost and the carbon emission; and converting the multi-target model into a single-target optimization model through a normalization and linear weighting method. Furthermore, an improved hybrid ant colony algorithm is designed for solving, the algorithm firstly utilizes a genetic algorithm with a dynamic crossover mutation rate to carry out global exploration, then converts a high-quality solution set into pheromones to initialize the ant colony algorithm, and finally carries out refined search by the ant colony algorithm to obtain an optimal facility site selection and transportation volume distribution scheme. According to the method, the problems that the economic and environmental protection targets are difficult to cooperate and the algorithm is easy to premature convergence in the traditional scheduling method are effectively solved, and comprehensive optimization of road train scheduling in the aspects of efficiency, cost and low carbon is realized.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation and logistics optimization technology, and in particular to an intelligent scheduling method for highway trains based on an improved ant colony algorithm. Background Technology

[0002] Road freight transport, as an efficient mode of logistics, occupies an important position in the modern logistics system. However, with rapid economic development and ever-increasing transportation demand, traditional road freight dispatching methods have gradually exposed numerous problems. Currently, the road freight transport industry is facing a significant need to transform towards intelligent, efficient, and environmentally friendly operations. Traditional dispatching methods mainly rely on manual experience, which, while meeting basic transportation needs to a certain extent, shows increasing limitations when facing complex transportation networks and changing transportation conditions. First, traditional methods struggle to comprehensively optimize various costs during the transportation process, resulting in persistently high transportation costs. Second, due to the lack of an effective real-time dispatching mechanism, vehicles often experience idleness and congestion during transportation, significantly reducing transportation efficiency. Furthermore, with increasingly stringent environmental regulations, carbon emissions during transportation have attracted widespread attention, a crucial factor that traditional dispatching methods often overlook.

[0003] Chinese patent application CN111967668A, entitled "A Cold Chain Logistics Route Optimization Method Based on Improved Ant Colony Algorithm," provides a method for optimizing cold chain logistics delivery routes using an ant colony algorithm. This method achieves cost optimization through pheromone updates and route selection strategies. While this method improves route search efficiency to some extent, it fails to incorporate environmental indicators such as carbon emissions into the objective function. Furthermore, traditional ant colony algorithms suffer from poor pheromone distribution uniformity in the initial stage, and the positive feedback mechanism easily leads to premature convergence. Therefore, when dealing with multi-objective, multi-constraint road train scheduling problems, it is difficult to achieve low-carbon optimization while ensuring economic efficiency, and its global search capability is limited.

[0004] Chinese patent CN114971317B, entitled "An Optimization Method for Equipment Scheduling and On-Site Logistics Based on an Improved Genetic Algorithm," provides an optimization method for equipment scheduling and on-site logistics using an improved genetic algorithm. This method searches for the optimal scheduling scheme through selection, crossover, and mutation operations. While genetic algorithms exhibit good population diversity in the initial stages, they lack the positive feedback mechanism of pheromone accumulation found in ant colony optimization. This leads to a significant decrease in convergence speed in the later stages of iteration, a tendency to generate redundant solutions, and insufficient consideration of the synergistic optimization of facility construction costs, operating costs, and carbon emissions. Consequently, their effectiveness in optimizing overall costs in complex multi-level transportation networks is limited, making it difficult to achieve the dual goals of economic efficiency and environmental protection.

[0005] To address these shortcomings, many researchers have proposed improvements to ant colony optimization (ACO) and genetic algorithms, such as refining the state transition probability formula and pheromone update strategy to enhance convergence performance. These improvements have enhanced the application effectiveness of ACO to some extent, but further optimization and refinement are still needed when facing the diverse demands and complex constraints of road rail transportation. Summary of the Invention

[0006] In view of the above problems, this invention provides a smart scheduling method for highway trains based on an improved ant colony algorithm. By constructing a low-carbon scheduling model for highway trains with the lowest overall cost under low-carbon constraints, and solving it using an improved hybrid ant colony algorithm, the method achieves high efficiency, economy, and environmental friendliness in highway train scheduling.

[0007] This invention provides a smart scheduling method for highway trains based on an improved ant colony algorithm, comprising the following steps: A four-level highway freight transport network model is constructed, comprising a set of loading points, a set of candidate primary transfer stations, a set of candidate secondary transfer stations, and a set of unloading points. Based on this four-level transport network model, a low-carbon scheduling model is established with the objectives of minimizing overall cost and minimizing carbon emissions during transport. The overall cost includes fixed infrastructure construction costs, transport costs, and operating costs. The dual objective function of the low-carbon scheduling model is normalized and transformed into a single-objective optimization model using a linear weighting method. A hybrid ant colony algorithm is used to solve the single-objective optimization model. The hybrid ant colony algorithm includes: firstly, using a genetic algorithm for global search to obtain an initial high-quality solution set; then, transforming the initial high-quality solution set into an initial pheromone distribution to guide the ant colony algorithm for refined search to obtain the optimal scheduling scheme. The optimal scheduling scheme includes planning the transportation routes of highway trains and allocating traffic flow; The location results of the first-level transfer station, second-level transfer station and unloading point are selected based on the optimal scheduling scheme, as well as the cargo transportation volume allocation results between each facility node; the cargo loading volume of each train is determined based on the optimal transportation volume between each facility node in the optimal solution results, and the number of carriages is determined.

[0008] Optionally, the low-carbon scheduling model includes the following constraints: each loading point is served by only one primary transfer station, each open primary transfer station is served by only one secondary transfer station, and each open secondary transfer station is served by only one unloading point; the cargo handling volume of each level of node does not exceed its maximum capacity; and the total inflow and outflow of cargo at primary and secondary transfer stations remains balanced.

[0009] Optionally, the linear weighting method uses weighting coefficients to adjust the importance of the overall cost target and the carbon emission target in the single-objective optimization model.

[0010] Optionally, the genetic algorithm uses the reciprocal of the objective function as fitness, and dynamically adjusts the crossover rate and mutation rate according to the relationship between the fitness of individuals in the population and the average fitness.

[0011] Optionally, the dynamic adjustment of crossover rate and mutation rate specifically involves: when the fitness of the individual to be operated on is higher than the average fitness of the population, a relatively low crossover rate or mutation rate is assigned to it; when the fitness of the individual to be operated on is lower than the average fitness of the population, a preset maximum crossover rate or maximum mutation rate is assigned to it.

[0012] Optionally, the step of converting the initial high-quality solution set into the initial pheromone distribution specifically involves selecting individuals with a predetermined proportion of high fitness rankings from the results of the genetic algorithm and adding their corresponding scheduling scheme path information to the initial pheromone of the ant colony algorithm.

[0013] Optionally, the state transition probability of the ant colony algorithm is determined by both the pheromone concentration on the path and the heuristic information reflecting the distance between nodes.

[0014] Optionally, in the four-level transportation network model, it is stipulated that goods must pass through the loading point, the first-level transfer station, and the second-level transfer station in sequence before finally arriving at the unloading point, and skipping levels of transportation is not allowed.

[0015] Compared with the prior art, the present invention has at least the following beneficial effects: This invention achieves intelligent and efficient highway train scheduling by constructing a low-carbon scheduling model with minimum overall cost and solving it using an improved hybrid ant colony algorithm. In terms of model design, a four-level highway train transportation network optimization model is proposed, effectively integrating objective functions such as fixed infrastructure cost, transportation cost, operating cost, and carbon emissions, thus solving the problem of traditional methods neglecting environmental constraints. The multi-objective function is normalized using a linear weighted method, supporting flexible adjustment of objective function weights, reducing train carbon emissions while ensuring cost optimization. Regarding algorithm improvement, an innovative hybrid strategy combining genetic algorithm and ant colony algorithm addresses the shortcomings of existing technologies, protecting high-quality solutions while enhancing population diversity and improving algorithm convergence speed. Compared with traditional methods, this invention significantly reduces transportation costs, decreases carbon emissions, and improves scheduling efficiency, providing an effective solution for intelligent highway train scheduling. Attached Figure Description

[0016] Figure 1 This is a flowchart of the intelligent scheduling method for highway trains based on the improved ant colony algorithm of the present invention.

[0017] Figure 2 This is a schematic diagram of a low-carbon dispatching network for highway trains with the lowest overall cost under the low-carbon constraints of the present invention.

[0018] Figure 3 This is a flowchart illustrating the use of an improved hybrid ant colony algorithm to solve a low-carbon scheduling model for highway trains, as presented in this invention. Detailed Implementation

[0019] To better understand the above-described objectives, features, and advantages of the present invention, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments of the present invention and the features thereof can be combined with each other. Furthermore, the present invention can be implemented in other ways different from those described herein; therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0020] A specific embodiment of the present invention, such as Figure 1-3 A method for intelligent scheduling of highway trains based on an improved ant colony algorithm is disclosed, comprising the following steps: Step S1: Determine the vehicle type, functional composition, and operating rules of the highway train.

[0021] Step S2: Conduct a comprehensive analysis of the joint scheduling problem of highway trains, establish a four-level highway train freight transport network model, and make reasonable assumptions based on the actual situation; Specifically, (1) during the transportation process, the highway train departs from the loading point and passes through the first-level transfer station and the second-level transfer station in sequence, and finally travels to the unloading point for unloading and processing of goods; Furthermore, all goods are first processed at a primary transit station before being sent to a secondary transit station. Specifically, the primary transit station classifies and stores the goods, and once a certain quantity is reached, they are transported to the secondary transit station. The secondary transit station inspects goods of different categories and quality levels, and then takes further processing measures based on the condition of each product.

[0022] (2) The alternative locations of loading point, unloading point, primary transfer station and secondary transfer station are known, and the primary transfer station and secondary transfer station are divided into different levels. The construction cost and processing capacity of each facility node (i.e. loading point, unloading point, primary transfer station and secondary transfer station) are known.

[0023] (3) The impact of road conditions on vehicle movement is not considered during the transportation of goods at each facility node, and the vehicles move at a constant speed.

[0024] (4) Goods shipped from the loading point can only be transported to one primary transfer station. Each primary transfer station can accept goods from multiple loading points. Cross-level transportation is not allowed. (5) Goods processed and classified by a primary transit station can only be transported to one secondary transit station, and each secondary transit station can accept goods from multiple primary transit stations; (6) The distance between each facility node in the scheduling network is known, and the transportation cost is proportional to the transportation distance and the transportation quantity; (7) All the carriages of the highway trains in the same fleet are of the same model and the load capacity of each carriage is the same. The number of carriages to be attached is selected according to the amount of cargo loaded on each train.

[0025] Step S3: Considering the characteristics of road train transportation and environmental issues, and taking into account the various costs and constraints of road trains in the scheduling and transportation process, construct a low-carbon scheduling model for road trains with the lowest overall cost under low-carbon constraints.

[0026] Among them, the objective function of the low-carbon scheduling model for highway trains, which minimizes the overall cost under low-carbon constraints, considers two aspects: minimizing the overall cost in the scheduling network and minimizing the environmental impact caused by waste recycling. The expression is as follows: (1) Min = + + (2) (3) (4) (5) Where Z1 represents the objective function for carbon emissions generated during the transportation of goods between loading points, unloading points, primary transfer stations, and secondary transfer stations, and Z2 represents the carbon emissions generated by fixed infrastructure construction costs. Transportation costs and operating costs The objective function represents the overall cost in the highway train dispatching network; r represents the r-th loading point, R represents the set of loading points; i represents the i-th primary transfer station, I represents the set of primary transfer stations; j represents the j-th secondary transfer station, J represents the set of secondary transfer stations; w represents the w-th unloading point, W represents the set of unloading points; This represents the distance from loading point r to primary transfer station i. This represents the distance from primary transfer station i to secondary transfer station j. This represents the distance from secondary transfer station j to unloading point w; This represents the fixed construction cost of primary transfer station i. This represents the fixed construction cost of secondary transfer station j; This represents the unit transportation cost of goods from loading point r to primary transfer station i. This represents the unit transportation cost of goods from primary transit station i to secondary transit station j. This represents the unit transportation cost of goods from secondary transit station j to unloading point w; , , These represent the unit operating costs when handling goods at primary transfer station i, secondary transfer station j, and unloading point w, respectively. This represents the carbon emission coefficient of road trains, i.e., the carbon emissions per unit of fuel. Quantity (kg / L) This represents the energy consumption coefficient of a train during transport, which is the amount of fuel consumed per unit distance (L / km). , , , , , , , All are decision variables, among which, , , For 0-1 variables, Indicates whether the primary transfer station i serves the loading point r. This indicates whether secondary transfer station j serves primary transfer station i. Indicates whether the unloading point w serves the secondary relay station j; , These are 0-1 variables, representing whether a primary transfer station i is established at the node, and whether a secondary transfer station j is established at the node, respectively; , , These are continuous variables, representing the weight of goods transported from loading point r to primary transfer station i, the weight of goods transported from primary transfer station i to secondary transfer station j, and the weight of goods transported from secondary transfer station j to unloading point w, respectively.

[0027] It is understandable that transportation costs refer to the total costs incurred by road trains when transporting goods between various facility nodes, mainly including vehicle rental costs, maintenance and repair costs, and fuel consumption costs; operating costs mainly include the expenses incurred in sorting, inspecting and other goods handling activities at each facility node, as well as the personnel labor costs and equipment maintenance costs incurred to ensure the normal operation of the recycling work at each node.

[0028] Furthermore, the constraints of the low-carbon scheduling model for highway trains are as follows: (6) (7) (8) 1 (9) 1 (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (twenty one) in, , , These represent the maximum storage capacity of primary transfer station i, secondary transfer station j, and unloading point w, respectively.

[0029] It can be understood that constraints (6)-(8) are uniqueness constraints: constraint (6) means that each loading point can only be served by one primary transfer station; constraint (7) means that each open primary transfer station can only be served by one secondary transfer station; constraint (8) means that each open secondary transfer station can only be served by one unloading point. Constraints (9)-(10) are node quantity constraints: constraints (10) and (11) mean that at least one of the candidate primary transfer stations and candidate secondary transfer stations must be selected for construction. Constraints (11)-(13) represent capacity constraints, meaning that the flow into the primary transfer station, secondary transfer station, and unloading point cannot exceed their maximum processing capacity. Constraints (14)-(15) represent flow direction constraints: constraint (14) means that the loading point can only be served by an open primary transfer station; constraint (15) means that the primary transfer station can only be served by an open secondary transfer station. Constraints (16)-(17) represent flow balance constraints: Constraint (16) represents flow balance at the primary transfer station; Constraint (17) represents flow balance at the secondary transfer station. Constraints (18)-(21) are variable value constraints, representing the decision variables respectively. , , Non-negative decision variables , , , , It is a 0-1 variable.

[0030] Step S4: The constructed multi-objective hybrid low-carbon constraint road train low-carbon scheduling model with the minimum overall cost is transformed into a single-objective optimization model that is easier to solve using the linear weighting method. The specific steps are as follows: Step S41: Dimensionless model processing.

[0031] The two objectives of the low-carbon scheduling model for highway trains, which minimizes overall cost under low-carbon constraints, are normalized to the same dimension. The range method is then used to make the objectives dimensionless, and the expression is: = (twenty two) = (twenty three) In the formula, , The objective function Z1 is the carbon emissions generated during the transportation of goods between loading points, unloading points, primary transfer stations, and secondary transfer stations, derived from the fixed infrastructure construction costs. Transportation costs and operating costs The normalized value of the objective function Z2; , , , The objective function Z1 represents the carbon emissions generated during the transportation of goods between loading points, unloading points, primary transfer stations, and secondary transfer stations, respectively, and is derived from the fixed infrastructure construction costs. Transportation costs and operating costs The objective function Z2 is found to be the minimum and maximum value in the current set of network optimization schemes. Through variable normalization, each objective function can be transformed into a value in the interval [0, 1].

[0032] Step S42: Problem Transformation. Using the linear weighted method, the road train low-carbon scheduling model with the minimum overall cost under the low-carbon constraints is transformed into a single-objective programming model for solution. The transformed single-objective function is: = + (twenty four) In the formula, , These represent the two normalized values ​​respectively. , The weighting coefficients.

[0033] Furthermore, the weighting coefficients satisfy: + =1, and ≥0, ≥0, the above weight coefficients reflect the degree of preference of the total benefit for each sub-objective. By combining different weight coefficients, the optimal travel route can be selected under different objective conditions, and the optimal transportation volume between each node can be reasonably arranged.

[0034] Step S5: Based on the improved hybrid ant colony algorithm, solve the single-objective optimization model to obtain the corresponding highway train scheduling. The specific steps are as follows: Step S51: Initialize genetic algorithm parameters: Each chromosome represents a train; the order of each chromosome represents the route of each train; the population size is N; the initial number of iterations is 0; the number of iterations is g, and the maximum number of iterations is g. max ; Step S52: Set the reciprocal of the objective function in the transformed single-objective optimization model as the population fitness, expressed as: F( )= ; Step S53: Dynamically calculate the crossover rate P in the genetic algorithm based on the population fitness. c and the rate of variation P m The expression is: P c = (25) P m = (26) In the formula, The maximum fitness value. This represents the average fitness value. The larger fitness value among the two chromosomes involved in the crossover. This represents the fitness value of the variant chromosome. To achieve the maximum crossover rate, To minimize the crossover rate, For the maximum variation rate, Minimum rate of variation; Used to determine population diversity Used to determine the degree of dispersion of chromosomes from a population. Step S54: Generate a random number u (0,1), randomly select two chromosomes, and determine the crossover rate of the two chromosomes. With random numbers u Based on the size relationship, cross over all two chromosomes that meet the crossover condition to obtain the crossover chromosome; Furthermore, the crossover condition is: If cross rate less than or equal to random numberu Discard both chromosomes; if both have a crossover rate... Greater than random number u Then, these two chromosomes are retained for crossover, expressed as: (27) In the formula, , Indicates the first n and the One chromosome; X′( ′), X′( + 1) indicates the first ′ and the The new chromosomes produced after crossing over; u A random number generated between (0,1).

[0035] Step S55: Compare each crossed chromosome X′ obtained in step S54. The rate of variation P of ′) m ( The relationship between P' and the random number u, if P' m ( (′) less than or equal to random number u Discard the crossed chromosome; If P is satisfied m ( ) If u is mutated, then a mutation operation is performed on it: the mutation rate is multiplied by the character length of each crossover chromosome to obtain the number of characters in each crossover chromosome that needs to be mutated, expressed as: gene ( ) =dP m ( (28) In the formula, gene ( ) is the first The number of characters that need to be mutated in a crossover chromosome; d is the length of the crossover chromosome. P m ( ) is the first The variation rate of a cross chromosome.

[0036] Furthermore, based on the number of characters in each chromosome that needs to be mutated and crossed over. gene ( The mutation starts randomly from the last character of the crossing chromosome, and the expression is: Y( ) = Y1 + b + rand(a - b) Among them, Y( Y1 represents the final individual generated by the mutation, Y2 represents the individual that does not participate in the mutation, a and b represent the maximum and minimum values ​​of each character in the solution, and rand() is a random function between (0,1).

[0037] After completing one round of crossover and mutation, all the solutions obtained are compared. A roulette wheel method is used to retain the same number of final individuals as the initial population. The best final individual is selected as the final offspring individual, which can retain most of the better solutions and increase the diversity of solutions.

[0038] This invention obtains all solutions after crossover and mutation, and uses a roulette wheel method to retain individuals with the same number as the initial population. The roulette wheel method is as follows: First, the fitness value of each individual is calculated. Since the objective function is to minimize the total cost Z*, the fitness is defined as the reciprocal of the total cost F(Z*) = 1 / Z*, so the lower the total cost, the higher the fitness value of the individual. Then, the probability of each individual being selected is determined according to the proportion of each individual's fitness value to the total fitness. The higher the fitness, the larger the area occupied by the mutated individual on the "roulette wheel". Then, the mutated individual is repeatedly selected by generating random numbers and determining their positions on the roulette wheel until the number of selected individuals is the same as the initial population size. This mechanism ensures that excellent individuals have a higher probability of being retained, maintains the diversity of the population, and avoids the algorithm from converging to a local optimum too early.

[0039] Step S56: Compare the number of iterations; if g > g max If the genetic algorithm ends, an initial high-quality solution set is obtained, and the process proceeds to step S57; otherwise, let g = g + 1 and return to step S52 to step S56. Furthermore, in each new iteration, the following parameters are recalculated based on the new generation of the population: 1. Population Fitness (F(Z*)): Due to crossover and mutation, the individuals (chromosomes) in the population have changed. Therefore, in step S52, the objective function value Z is recalculated for all individuals of the new generation. ， Then the fitness of each individual is recalculated. F(Z ) = 1 / Z*.

[0040] 2. Population statistical characteristics ( and For): Based on the recalculated fitness of all individuals, step S53 will recalculate the maximum fitness value of the new generation population. and average fitness value These two values ​​are the basis for dynamically adjusting Pc and Pm.

[0041] 3. Dynamic crossover rate (Pc) and mutation rate (Pm): In step S53, the newly calculated maximum fitness value is used. and average fitness value And for each specific chromosome (the individual involved in crossover or mutation), the fitness (f' or f) is used to recalculate the probabilities Pc and Pm for each crossover and mutation operation in this iteration.

[0042] Step S57: Initialize ant colony algorithm parameters: Select the top 20% of the final offspring individuals from the initial high-quality solution set as target ants in the ant colony algorithm, with the number of target ants being m. ’ The initial number of iterations is set to 0, and the number of iterations g ’ The maximum number of iterations is set to g. max ’ Each ant simulates the transportation route planning process of a highway train between loading and unloading points, and constructs a complete scheduling scheme by moving between various facility nodes.

[0043] Specifically, the initial values ​​of the ant algorithm parameters refer to the top 20% of the scheduling schemes with the highest fitness and lowest cost Z* selected from the final generation population after the genetic algorithm has run. This represents a complete road train scheduling scheme (i.e., it specifies the complete path selection and flow allocation of goods from the loading point, through which primary and secondary transfer stations, to the final unloading point).

[0044] Step S58: Place m ’ Target ants are randomly placed at m loading points, and the state transition probability P of each target ant is determined. ij k Move according to the rules of roulette to find the optimal path.

[0045] Furthermore, the state transition probability P of each target ant ij k The expression is: P ij k = (29) Among them, P ij k Let be the state transition probability of the k-th target ant moving from the i-th primary relay station to the j-th secondary relay station. Let represent the residual pheromone concentration of the k-th target ant between the i-th primary relay station and the j-th secondary relay station. Let represent the residual pheromone concentration of the k-th target ant between the i-th primary relay station and the s-th secondary relay station. This is the heuristic information from the i-th primary transfer station to the j-th secondary transfer station, and =1 / , This represents the distance from the i-th primary transfer station to the j-th secondary transfer station; This provides heuristic information for the journey from the i-th primary transfer station to the s-th secondary transfer station, where α is the pheromone factor. To inspire information factors, Let be the set of the next facility nodes that the k-th target ant can reach from the current facility node; In this system, the movement of ants represents the selection of facility nodes at each level by the highway train during the scheduling process. When choosing the next facility node, the ant does not select randomly, but rather based on the state transition probability. This probability comprehensively considers the pheromone concentration and heuristic information along the path; paths with higher probabilities are more likely to be selected. Then, a roulette wheel mechanism is used to randomly select the movement direction according to the probability distribution of each path.

[0046] For example, the heuristic information for the distance from the i-th primary transfer station to the j-th secondary transfer station refers to the reciprocal of the distance from the i-th primary transfer station to the j-th secondary transfer station.

[0047] Step S59: After an ant completes its path optimization, it is placed in the tabu list. After all ants have built their own optimized paths, a global pheromone update is performed on all paths traversed by any ant in this iteration.

[0048] Specifically, g ’ In the next iteration, update the pheromone (pheromone concentration) on the optimization path (i,j). (g ’ The expression for +1) is: (g ’ +1) = (1-ρ)· (g ’ ) +Δ (g ’ (30) Where ρ represents the pheromone evaporation factor, a coefficient ranging from (0,1); 1-ρ represents the pheromone residue coefficient; Δ (g ’ ) represents the g-th ’ The total amount of pheromone left by all ants on the optimal path (i,j) at the next iteration; (g ’ ) represents the g-th digit on the optimal path (i,j). ’ Pheromones during the next iteration.

[0049] Furthermore, the gth ’In the next iteration, the expression for the total pheromone amount left by all ants on the path is: Δ (31) in, The amount of pheromone left by the k-th target ant on the edge (i,j) is expressed as: = (32) Where Q represents the total amount of pheromone released by each ant after completing a full path. It is a constant and will affect the convergence speed of the algorithm to some extent. This represents the total cost of the path taken by the k-th ant in this cycle. The greater the total cost of the path chosen by the ant, the smaller the amount of pheromone it releases per unit length of the path.

[0050] Step S510: Compare the number of iterations; if g ’ >g max ’ If the ant colony algorithm succeeds, the algorithm ends, and the optimal solution of the improved hybrid ant colony algorithm is obtained (i.e., a complete highway train scheduling scheme, for example, specifying the complete path selection of goods from the loading point, through which primary and secondary transfer stations, to the final unloading point, and the distribution of transport flow between the two nodes), proceeding to step S6; otherwise, let g... ’ = g ’ +1, return to step S58.

[0051] Furthermore, in each new iteration, the following parameters are recalculated based on the new generation of ants: 1. Taboo list for all ants: In step S58, each ant will rebuild its path again from a random loading point. Therefore, their respective taboo lists will be cleared; 2. Pheromone concentration τ along the path: The pheromone concentration will not be "reset" to its initial value, but will be globally updated according to formula (30). This is the core of the iteration; 3. The state transition probability P of an ant ij k Since the pheromone has been updated, according to formula (29), the probability of choosing all paths P ij k This information is automatically updated. In the next iteration, the ants will choose their path based on this new probability distribution.

[0052] Step S6: Select the optimal locations of the primary transfer station, secondary transfer station and unloading point in the low-carbon dispatching network for highway trains based on the optimal solution results of the improved hybrid ant colony algorithm; determine the cargo load of each train based on the optimal transport volume between each facility node in the optimal solution results, and thus select the number of carriages to be attached to it.

[0053] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for intelligent scheduling of highway trains based on an improved ant colony algorithm, characterized in that, Includes the following steps: A four-level highway freight transport network model is constructed, comprising a set of loading points, a set of candidate primary transfer stations, a set of candidate secondary transfer stations, and a set of unloading points. Based on this four-level transport network model, a low-carbon scheduling model is established with the objectives of minimizing overall cost and minimizing carbon emissions during transport. The overall cost includes fixed infrastructure construction costs, transport costs, and operating costs. The dual objective function of the low-carbon scheduling model is normalized and transformed into a single-objective optimization model using a linear weighting method. A hybrid ant colony algorithm is used to solve the single-objective optimization model. The hybrid ant colony algorithm includes: firstly, using a genetic algorithm for global search to obtain an initial high-quality solution set; then, transforming the initial high-quality solution set into an initial pheromone distribution to guide the ant colony algorithm for refined search to obtain the optimal scheduling scheme. The optimal scheduling scheme includes planning the transportation routes of highway trains and allocating traffic flow; The location results of the first-level transfer station, second-level transfer station and unloading point are selected based on the optimal scheduling scheme, as well as the cargo transportation volume allocation results between each facility node; the cargo loading volume of each train is determined based on the optimal transportation volume between each facility node in the optimal solution results, and the number of carriages is determined.

2. The method according to claim 1, characterized in that, The low-carbon scheduling model includes the following constraints: each loading point is served by only one primary transfer station, each open primary transfer station is served by only one secondary transfer station, and each open secondary transfer station is served by only one unloading point; the cargo handling volume of each level of node does not exceed its maximum capacity; and the total inflow and outflow of cargo at primary and secondary transfer stations remains balanced.

3. The method according to claim 1, characterized in that, The linear weighting method uses weighting coefficients to adjust the importance of the overall cost target and the carbon emission target in the single-objective optimization model.

4. The method according to claim 1, characterized in that, The genetic algorithm uses the reciprocal of the objective function as fitness, and dynamically adjusts the crossover rate and mutation rate based on the relationship between the fitness of individuals in the population and the average fitness.

5. The method according to claim 4, characterized in that, The dynamic adjustment of crossover rate and mutation rate specifically means that when the fitness of the individual to be operated on is higher than the average fitness of the population, a relatively low crossover rate or mutation rate is assigned to it; when the fitness of the individual to be operated on is lower than the average fitness of the population, a preset maximum crossover rate or maximum mutation rate is assigned to it.

6. The method according to claim 1, characterized in that, The specific steps of transforming the initial high-quality solution set into the initial pheromone distribution are as follows: select individuals with a predetermined proportion of the highest fitness ranking from the results of the genetic algorithm, and enhance the corresponding scheduling scheme path information into the initial pheromone of the ant colony algorithm.

7. The method according to claim 1, characterized in that, The state transition probability of the ant colony algorithm is determined by both the pheromone concentration on the path and the heuristic information reflecting the distance between nodes.

8. The method according to any one of claims 1-7, characterized in that, In the four-level transportation network model, it is stipulated that goods must pass through the loading point, the first-level transfer station, and the second-level transfer station in sequence before finally arriving at the unloading point, and skipping levels of transportation is not allowed.

Images