Port multimodal transport path optimization method under uncertain condition

Abstract

The invention provides a port multimodal transport path optimization method under an uncertain condition, and relates to the technical field of port freight. The port multimodal transport path optimization method comprises the steps of model construction and model solving. Aiming at the random fluctuation phenomenon of port import and export cargo transportation volume caused by market environment change, the multimodal transport path optimization problem under the condition of uncertain transportation volume is researched, a multimodal transport model with the minimum transportation comprehensive cost upper limit as the target is established, and the model is solved by adopting a genetic algorithm improved based on fuzzy logic. Simulation results show that in the aspect of effectiveness, the transportation comprehensive cost of the constructed model is reduced by about 10% compared with that of a comparison model and is only about 7% higher than that of an optimal decision scheme, and it is proved that the constructed model can remarkably reduce the comprehensive cost of multimodal transportation under the condition that the transportation volume is uncertain.

Description

Technical Field

[0001] The present invention relates to the technical field of port freight transportation, and in particular to a method for optimizing port multimodal transport routes under uncertain conditions. Background Art

[0002] With the continuous advancement of global economic integration, port logistics transshipment, as a vital link connecting domestic and international trade, has become an increasingly prominent focus due to its high efficiency, low cost, and environmental friendliness. Intermodal transport, considered a solution that coordinates different modes of logistics and transportation to achieve resource sharing, cost reduction, and efficiency improvement, has become a key transportation organization model in modern logistics. Currently, research on intermodal transport is extensive, with route optimization under uncertain conditions being a hot topic, offering significant research implications for improving the cost-effectiveness and stability of transportation services.

[0003] In summary, existing research focuses on making assumptions about uncertain factors in the multimodal transport process when modeling. Therefore, the result obtained is a robust solution with the minimum expectation under a specific probability, which makes it difficult to ensure the optimality of the comprehensive cost upper limit in the scenario of uncertain transport volume.

[0004] Existing research focuses on assuming uncertain factors in the multimodal transport process when modeling. Therefore, the result obtained is a robust solution with the minimum expectation under a specific probability, which makes it difficult to ensure the optimality of the upper limit of the comprehensive cost in the scenario of uncertain transport volume. Chen Mili et al. Optimize multimodal transport paths under uncertain conditions. The "Highway Traffic Science and Technology" effectively dealt with the above problem by combining the Monte Carlo method with a genetic algorithm, but the solution process of the article requires a large number of iterative calculations, and the algorithm converges slowly, which makes it difficult to meet the needs of real-time decision-making. Therefore, from the perspective of the transshipment decision-making and scheduling party, the present invention considers the uncertainty of the import and export cargo volume at the port, and constructs a multimodal transport path optimization model with the minimization of the upper limit of the comprehensive cost of transportation, time and carbon emissions as the goal. The adaptive genetic algorithm improved based on fuzzy logic theory is used to solve the model, verify the effectiveness and convergence of the model, and obtain an ideal logistics path optimization solution. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention provides a method for optimizing multimodal transport paths at ports under uncertain conditions, which solves the problems that the current multimodal transport path planning methods require a large number of iterative calculations in the solution process, the algorithm converges slowly, and it is difficult to meet the needs of real-time decision-making.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a method for optimizing port multimodal transport routes under uncertain conditions, including model construction and model solution, wherein the model construction specifically includes the following types:

[0007] 1) Calculation of transportation costs

[0008] Transportation costs refer to the costs incurred while transporting goods by means of transport, and are related to the volume of goods transported, the distance transported, the unit transportation cost, and the mode of transport. The calculation formula is:

[0009]

[0010] Where A is the set of transportation routes, M is the set of transportation modes, h and i are the transportation nodes along the way, q is the cargo weight, k and j are the transportation modes, It represents the unit cost of transporting a unit of goods between nodes h and i by transport mode k. It represents the transportation distance of a unit of cargo between nodes h and i when transport mode k is selected. It takes 1 when transport mode k is selected between nodes (h, i), otherwise it takes 0;

[0011] 2) Calculation of transfer fees

[0012] Transshipment fees refer to the fees incurred when goods are transported at a certain transit point. They are related to the volume of goods transported and the unit conversion cost. The calculation formula is:

[0013]

[0014] Where N is the node city set, It represents the cost of converting a unit of cargo from transport mode k to transport mode j at node i. It means that the value is 1 when transport mode k is converted to l at node i, and 0 otherwise;

[0015] 3) Cargo transportation time

[0016] Cargo transportation time refers to the time it takes for goods to be transported by means of transport and the time required to switch modes of transport at transit nodes. The calculation formula is:

[0017]

[0018] In the formula It represents the transport speed of unit cargo between nodes h and i when transport mode k is selected. It represents the time required for a unit of cargo to be converted from transport mode k to transport mode j at node i;

[0019] 5) Carbon emissions

[0020] Carbon emissions refer to the greenhouse gas emissions generated by transportation vehicles during cargo transportation. They are related to the mode of transportation, cargo volume, and transportation distance. The calculation formula is:

[0021]

[0022] In the formula represents the carbon emissions per unit of cargo when transport mode k is selected between nodes h and i;

[0023] 5) Calculation of objective function and constraints

[0024] The objective function is:

[0025]

[0026] C3(x,y)=q{c before max[(T ob -T),0]+c after max[(TT ob ),0]} (6)

[0027] Formula (6) represents the sum of warehouse storage or penalty costs incurred when goods arrive outside the specified time;

[0028]

[0029] Formula (7) indicates that the carbon emissions generated during the transportation and transshipment of goods are greater than the penalty fees imposed by the national standards;

[0030] In formula (5), w1+w2+w3=1; in formula (6), c before It represents the storage cost of the unit cargo arriving at the destination too early, c after T represents the penalty cost of late arrival of goods per unit. od represents the agreed delivery time, T represents the actual delivery time; E in formula (7) allow Indicates the allowed carbon emissions, e p Indicates the price of carbon emissions per unit weight;

[0031] The constraints are:

[0032]

[0033] q min ≤q≤q max (12)

[0034]

[0035] Constraint (8) indicates that all nodes in the path comply with the conservation of cargo flow; Constraint (9) indicates that cargo can only adopt one mode of transportation on a transportation path; Constraint (10) indicates that cargo can change its mode of transportation at most once at a transfer node; Constraint (11) indicates that the transportation mode information on the transportation route before and after each transfer node is consistent with the node transportation mode conversion information; Constraint (12) indicates that the weight of the transported cargo is within a fixed range; Constraint (13) indicates that the transportation mode decision variable is a 01 variable; Constraint (14) indicates that the node transfer decision variable is a 01 variable.

[0036] Preferably, the model solving specifically includes the following steps:

[0037] Step 1: Chromosome encoding

[0038] It includes the transport path node code and transport mode code between the starting point and the end point. According to different contents, the path node is coded with a serial number, and the transport mode is coded with an alphabetical code, where H represents road transport, R represents rail transport, and W represents water transport.

[0039] Step 2: Population initialization

[0040] Generate T consecutive segments starting with the port number and ending with the destination number through random generation method, calculate the length of the path node segment, and randomly generate the corresponding transport mode chromosome segment based on the length, where T is the population size;

[0041] Step 3: Calculation of fitness and adaptive crossover and mutation probability

[0042] The opposite of the objective function is selected as the fitness of the chromosome, and the fitness set of the population is brought into the defined fuzzy inference system to calculate the crossover and mutation probabilities;

[0043] Step 4: Select

[0044] Using the replay sampling method of roulette wheel selection, the probability of each chromosome entering the next generation is set equal to the ratio of its fitness value to the sum of the fitness values ​​of individuals in the entire population, and random sampling is performed;

[0045] Step 5: Cross

[0046] The crossover operator uses a partial matching crossover operator, which first performs a crossover operation on the path node gene segments of the two chromosomes, and then crosses the gene segments corresponding to the transport mode;

[0047] Step 6: Mutation

[0048] The mutation operator uses the basic bit mutation operator. The transmission path node gene can be mutated to any node sequence number in the transmission network, and the transportation mode gene can be mutated to any symbol in the {M, R, W} set, thereby generating a new population.

[0049] Step 7: Verify the inner layer termination condition. Set the maximum number of inner layer iterations. If the optimal individual generated does not improve compared to the previous iteration or reaches the maximum number of times, end the inner layer loop and proceed to step 8. If there is improvement, return to step 3.

[0050] Step 8: Verify outer layer convergence

[0051] Set the maximum number of outer iterations and calculate the current cost upper limit. If the upper limit has not decreased significantly compared to the previous iteration or has reached the maximum number of times, end the outer loop and output the solution set; if there is a significant decrease, return to step 2.

[0052] Preferably, the fuzzy inference system in step 3 specifically includes the following steps:

[0053] Step 3-1: Calculate fitness variance and extreme values

[0054] First, find the corresponding objective function value C for the individuals in the population generated in the current round. i (x, y), and its reciprocal is the fitness l of the individual in the current situation i , the formula is as follows:

[0055]

[0056] According to l i Find the fitness of the current center individual And determine the individual fitness extreme difference R, the formula is as follows:

[0057]

[0058] Based on Calculate the fitness distance d between each individual in the population and the central individual i And find the population fitness variance S, the formula is as follows

[0059]

[0060] Finally, the fitness variance S and extreme value difference R are normalized;

[0061] Step 3-2: For ease of calculation, the membership function in the fuzzy inference system is triangular membership function;

[0062] Step 3-3: Map it to different fuzzy indices through appropriate fuzzification processing, and then use the prescribed fuzzy inference rules to transform it into fuzzy indices related to mutation and crossover probability;

[0063] Step 3-4: Finally, defuzzification is performed to obtain the adaptive crossover and mutation probabilities of the current population, and these are substituted into the genetic algorithm to solve the optimization problem.

[0064] The present invention provides a method for optimizing port multimodal transport routes under uncertain conditions. It has the following beneficial effects:

[0065] 1. The present invention studies the problem of multimodal transport path optimization under the scenario of uncertain transport volume by targeting the phenomenon of random fluctuations in the volume of import and export cargo at ports caused by changes in the market environment, establishes a multimodal transport model with the goal of minimizing the upper limit of the comprehensive transport cost, and solves the model using a genetic algorithm improved based on fuzzy logic. The simulation results show that in terms of effectiveness, the comprehensive transport cost of the constructed model is about 10% lower than that of the comparison model, and is only about 7% higher than the optimal decision-making solution, proving that the constructed model can significantly reduce the comprehensive cost of multimodal transport under the scenario of uncertain transport volume. In terms of convergence, the average number of iterations of the constructed model is about 28 times, which is about 62.5% less than the average of 75 times of the comparison model, indicating that the proposed model achieves a more efficient optimization strategy through adaptive probability changes and can reach convergence in a shorter time. The research results can provide a reference for the subsequent solution of multimodal transport path optimization problems under uncertain conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 Schematic diagram of the fuzzy reasoning workflow of the present invention;

[0067] Figure 2 Schematic diagram of the variance and extreme value membership function of the present invention;

[0068] Figure 3 Schematic diagram of the crossover and mutation probability membership function of the present invention;

[0069] Figure 4 This is a schematic diagram of the model solving process of the present invention;

[0070] Figure 5 Schematic diagram of the coding structure of individual chromosomes of the present invention;

[0071] Figure 6 Schematic diagram of the transportation network structure of the present invention;

[0072] Figure 7 This is a schematic diagram of the multimodal transport route decision of the present invention;

[0073] Figure 8 This is the comprehensive cost diagram of multimodal transport under different transport volumes of the present invention. DETAILED DESCRIPTION

[0074] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0075] Example:

[0076] like Figure 1-8 As shown, an embodiment of the present invention provides a method for optimizing port multimodal transport routes under uncertain conditions, including model construction and model solution. The model construction specifically includes the following types:

[0077] 1) Calculation of transportation costs

[0078] Transportation costs refer to the costs incurred while transporting goods by means of transport, and are related to the volume of goods transported, the distance transported, the unit transportation cost, and the mode of transport. The calculation formula is:

[0079]

[0080] Where A is the set of transportation routes, M is the set of transportation modes, h and i are the transportation nodes along the way, q is the cargo weight, k and j are the transportation modes, It represents the unit cost of transporting a unit of goods between nodes h and i by transport mode k. It represents the transportation distance of a unit of cargo between nodes h and i when transport mode k is selected. It takes 1 when transport mode k is selected between nodes (h, i), otherwise it takes 0;

[0081] 2) Calculation of transfer fees

[0082] Transshipment fees refer to the fees incurred when goods are transported at a certain transit point. They are related to the volume of goods transported and the unit conversion cost. The calculation formula is:

[0083]

[0084] Where N is the node city set, It represents the cost of converting a unit of cargo from transport mode k to transport mode j at node i. It means that the value is 1 when transport mode k is converted to l at node i, and 0 otherwise;

[0085] 3) Cargo transportation time

[0086] Cargo transportation time refers to the time it takes for goods to be transported by means of transport and the time required to switch modes of transport at transit nodes. The calculation formula is:

[0087]

[0088] In the formula It represents the transport speed of unit cargo between nodes h and i when transport mode k is selected. It represents the time required for a unit of cargo to be converted from transport mode k to transport mode j at node i;

[0089] 4) Carbon emissions

[0090] Carbon emissions refer to the greenhouse gas emissions generated by transportation vehicles during cargo transportation. They are related to the mode of transportation, cargo volume, and transportation distance. The calculation formula is:

[0091]

[0092] In the formula represents the carbon emissions per unit of cargo when transport mode k is selected between nodes h and i;

[0093] 5) Calculation of objective function and constraints

[0094] The objective function is:

[0095]

[0096] C3(x,y)=q{c before max[(T ob -T),0]+c after max[(TT ob ),0]} (6)

[0097] Formula (6) represents the sum of warehouse storage or penalty costs incurred when goods arrive outside the specified time;

[0098]

[0099] Formula (7) indicates that the carbon emissions generated during the transportation and transshipment of goods are greater than the penalty fees imposed by the national standards;

[0100] In formula (5), w1+w2+w3=1; in formula (6), c before It represents the storage cost of the unit cargo arriving at the destination too early, c after T represents the penalty cost of late arrival of goods per unit. od represents the agreed delivery time, T represents the actual delivery time; E in formula (7) allow Indicates the allowed carbon emissions, e p Indicates the price of carbon emissions per unit weight;

[0101] The constraints are:

[0102]

[0103] q min ≤q≤q max (12)

[0104]

[0105] Constraint (8) indicates that all nodes in the path comply with the conservation of cargo flow; Constraint (9) indicates that cargo can only adopt one mode of transportation on a transportation path; Constraint (10) indicates that cargo can change its mode of transportation at most once at a transfer node; Constraint (11) indicates that the transportation mode information on the transportation route before and after each transfer node is consistent with the node transportation mode conversion information; Constraint (12) indicates that the weight of the transported cargo is within a fixed range; Constraint (13) indicates that the transportation mode decision variable is a 01 variable; Constraint (14) indicates that the node transfer decision variable is a 01 variable.

[0106] The improved genetic algorithm has better global search capability and faster optimization capability. According to the characteristics of the model, the designed solution algorithm includes 8 steps. The process is as follows: Figure 4 As shown in Figure 2, the model solution specifically includes the following steps:

[0107] Step 1: Chromosome encoding

[0108] It includes the transport path node coding and transport mode coding between the starting point and the end point. According to different contents, the path node selects the serial number coding and the transport mode selects the letter coding, where H represents road transport, R represents rail transport, and W represents water transport. The individual chromosome coding structure is as follows Figure 5 As shown;

[0109] Step 2: Population initialization

[0110] Generate T consecutive segments starting with the port number and ending with the destination number through random generation method, calculate the length of the path node segment, and randomly generate the corresponding transport mode chromosome segment based on the length, where T is the population size;

[0111] Step 3: Calculation of fitness and adaptive crossover and mutation probability

[0112] The opposite of the objective function is selected as the fitness of the chromosome, and the fitness set of the population is brought into the defined fuzzy inference system to calculate the crossover and mutation probabilities;

[0113] Step 4: Select

[0114] Using the replay sampling method of roulette wheel selection, the probability of each chromosome entering the next generation is set equal to the ratio of its fitness value to the sum of the fitness values ​​of individuals in the entire population, and random sampling is performed;

[0115] Step 5: Cross

[0116] The crossover operator uses a partial matching crossover operator, which first performs a crossover operation on the path node gene segments of the two chromosomes, and then crosses the gene segments corresponding to the transport mode;

[0117] Step 6: Mutation

[0118] The mutation operator uses the basic bit mutation operator. The transmission path node gene can be mutated to any node sequence number in the transmission network, and the transportation mode gene can be mutated to any symbol in the {M, R, W} set, thereby generating a new population.

[0119] Step 7: Verify the inner layer termination condition. Set the maximum number of inner layer iterations. If the optimal individual generated does not improve compared to the previous iteration or reaches the maximum number of times, end the inner layer loop and proceed to step 8. If there is improvement, return to step 3.

[0120] Step 8: Verify outer layer convergence

[0121] Set the maximum number of outer iterations and calculate the current cost upper limit. If the upper limit has not decreased significantly compared to the previous iteration or has reached the maximum number of times, end the outer loop and output the solution set; if there is a significant decrease, return to step 2.

[0122] An improved genetic algorithm based on fuzzy logic adaptive adjustment of population crossover and mutation probability is proposed. Taking into account population diversity and individual fitness, a fuzzy inference system is used to adaptively adjust population crossover and mutation probability to improve algorithm exploration and accelerate algorithm convergence. The fuzzy inference system workflow is as follows: Figure 1 As shown. The fuzzy inference system in step 3 specifically includes the following steps:

[0123] Step 3-1: Calculate fitness variance and extreme values

[0124] First, find the corresponding objective function value C for the individuals in the population generated in the current round. i (x, y), and its reciprocal is the fitness l of the individual in the current situation i , the formula is as follows:

[0125]

[0126] According to l i Find the fitness of the current center individual And determine the individual fitness extreme difference R, the formula is as follows:

[0127]

[0128] Based on Calculate the fitness distance d between each individual in the population and the central individual i And find the population fitness variance S, the formula is as follows

[0129]

[0130] Finally, the fitness variance S and extreme value difference R are normalized;

[0131] Step 3-2: For ease of calculation, the membership function in the fuzzy inference system is triangular membership function, normalized variance S, normalized extreme value difference R, crossover probability P c and mutation probability P m The membership function is as follows Figure 2 、 Figure 3 As shown, the fuzzy inference rules are set as follows. Rule n: If S is W n1 , R is O n1 , then P c It's X n1 , P m It's X n2 , where n=1,2,...,N represents the nth rule, W n1 ,Q n1 ,X n1 ,X n2 are all fuzzy representations of each variable. Each fuzzy variable is divided into three levels: large, medium, and small, from which the following fuzzy rules can be determined:

[0132] If S is small, R is small, then P c It's small, P m It is big;

[0133] If S is small and R is medium, then P c It's small, P m It is big;

[0134] If S is small and R is large, then P c Yes, P m It is in the middle;

[0135] If S is medium and R is small, then P c Yes, P m It is in the middle;

[0136] If S is medium and R is medium, then P c Yes, P m It is in the middle;

[0137] If S is medium and R is large, then P c It's big, P m It is small;

[0138] If S is large and R is small, then P c Yes, P m It is in the middle;

[0139] If S is large and R is medium, then P c It's big, P m It is small;

[0140] If S is large, R is large, then P c It's big, P m It is small;

[0141] Step 3-3: Map it to different fuzzy indices through appropriate fuzzification processing, and then use the prescribed fuzzy inference rules to transform it into fuzzy indices related to mutation and crossover probability;

[0142] Step 3-4: Finally, defuzzification is performed to obtain the adaptive crossover and mutation probabilities of the current population, and these are substituted into the genetic algorithm to solve the optimization problem.

[0143] A simulation experiment was conducted on the above model. The experimental example was derived from the actual road-rail-water combined transportation network of Shenzhen Port to verify the optimization effect of the proposed minimax model and its solution algorithm based on fuzzy logic. Assume that there is a batch of goods with uncertain transportation volume that needs to be transferred from the port to a specific destination. There are 12 intermediate node cities to choose from, and each node can carry the transportation and loading and unloading tasks covering the entire cargo volume. The transportation mode between two nodes in the network can be divided into single transportation and multimodal transportation. The specific network structure is as follows: Figure 6 shown.

[0144] The port's transshipment scheduling decision-making department must transport goods from the port node (numbered 1) to the destination (numbered 14), and the goods must arrive at the destination within the specified delivery time window. If the goods arrive at the destination early, a unit storage fee of 8 yuan / (h*kg) will be incurred; if they arrive late, a unit penalty cost of 20 yuan / (h*kg) will be incurred. The transmission distances corresponding to different transportation modes for each transmission path in the network are shown in Appendix 1. The transportation speed, unit distance transportation cost, and unit distance carbon emissions of different transportation modes are shown in Appendix 2. Based on the specific conditions of the nodes in the network, transit parameters such as unit transit cost and unit transit time can be obtained, as shown in Appendix 3.

[0145] Due to the advanced nature of transportation plans, port transit dispatch departments are unable to determine the actual weight of transported cargo. Based on the characteristics of real-world port cargo transportation and relevant literature, we designed three different scenarios: low cargo volume, medium cargo volume, and high cargo volume, with cargo weights limited to [100kg, 150kg], [150kg, 200kg], and [200kg, 250kg], respectively.

[0146] The proposed model and the comparative model are used to conduct simulation tests on multimodal transport path optimization under the condition of uncertain transport volume. Under the set simulation parameters, the path decision obtained by the proposed model and the comparative model is as follows: Figure 7 As shown, Figure 7 a is the model of the present invention, Figure 7 b is the comparison model. Figure 7 It can be seen that according to the simulation setting parameters, the transportation route solved by the model constructed by the present invention is "1-2-7-9-13-14", the transportation mode is "road-rail-road-rail-road", and the comprehensive cost is 221,300 yuan; the transportation route of the comparison model is "1-2-7-9-13-14", the transportation mode is "road-rail-road-road-road-road", and the comprehensive cost is 241,500 yuan. Compared with the comparison model, the transportation cost of the constructed model is reduced by about 10%, indicating that the constructed model can reduce the comprehensive cost of multimodal transport under the scenario of uncertain transportation volume. The gap between the comprehensive cost generated by the path decisions obtained by different models and the minimum comprehensive cost is as follows. Figure 8 shown.

[0147] Depend on Figure 8 It can be seen that the comprehensive costs of the three routing schemes vary significantly in different freight volume scenarios. When the freight volume is [100, 150 kg], the comprehensive cost of the optimal decision-making scheme is 205,600 yuan, the comprehensive cost of the proposed model routing scheme is 221,300 yuan, and the comprehensive cost of the comparison model routing scheme is 241,500 yuan. The cost of the proposed model routing scheme is 7.6% higher than the optimal decision-making scheme and approximately 8.3% lower than the comparison model routing scheme. When the freight volume is [150, 200 kg], the comprehensive cost of the optimal decision-making scheme is 261,400 yuan, the comprehensive cost of the proposed model routing scheme is 289,800 yuan, and the comprehensive cost of the comparison model routing scheme is 329,700 yuan. In this case, the cost of the proposed model routing scheme is 10.8% higher than the optimal decision-making scheme and approximately 12.1% lower than the comparison model routing scheme. When the freight volume is [200, 250] kg, the overall cost of the optimal decision-making solution is 386,300 yuan, the overall cost of the proposed model routing solution is 427,700 yuan, and the overall cost of the comparison model routing solution is 484,200 yuan. The cost of the proposed model routing solution is 10.7% higher than the optimal decision-making solution, but approximately 11.7% lower than the comparison model routing solution. Overall, the cost of the proposed model routing solution is significantly lower than that of the comparison model routing solution in various transportation volume scenarios, approaching the optimal decision-making solution. The proposed model demonstrates significant advantages in controlling the impact of uncertainties on transportation costs, effectively reducing the adverse effects of uncertainties and lowering overall transportation costs.

[0148] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A port multimodal transport path optimization method under uncertain conditions, including model construction and model solving, characterized in that: The model building specifically includes the following types: 1) Calculation of transportation costs Transportation costs refer to the costs incurred by goods during transportation by means of transportation, which are related to the volume of goods, transportation distance, unit transportation cost and transportation method. The calculation formula is: Where A is the set of transport routes, M is the set of transport modes, h and i are transport nodes on the way, q is the cargo weight, k and j are transport modes, It represents the unit cost of transporting a unit of goods between nodes h and i by transport mode k. It represents the transportation distance of a unit of cargo between nodes h and i when transport mode k is selected. It takes 1 when transport mode k is selected between nodes (h, i), otherwise it takes 0; 2) Calculation of transfer fees Transit costs refer to the costs incurred when goods are converted to other modes of transport at a certain transit node. They are related to the volume of goods transported and the unit conversion cost. The calculation formula is: Where N is the node city set, It represents the cost of converting a unit of cargo from transport mode k to transport mode j at node i. It means that it takes 1 when the transport mode k is converted to l at node i, otherwise it takes 0; 3) Cargo transportation time Cargo transportation time refers to the time it takes for cargo to be transported by means of transportation and the time required to convert the mode of transportation at the transit node. The calculation formula is: In the formula It represents the transport speed of the unit cargo when the transport mode k is selected between nodes h and i. It represents the time required for a unit of cargo to be converted from transport mode k to transport mode j at node i; 4) Carbon emissions Carbon emissions refer to the greenhouse gas emissions generated by transportation tools during cargo transportation, which is related to the mode of transportation, cargo volume and transportation distance. The calculation formula is: In the formula It represents the carbon emission per unit distance of a unit of cargo between nodes h and i when transport mode k is selected; 5) Calculation of objective function and constraints The objective function is: C3(x,y)=q{c before max[(T ob -T),0]+c after max[(T-T ob ),0]} (6) Formula (6) represents the sum of warehouse storage or penalty costs incurred when goods arrive outside the specified time; Formula (7) indicates that the carbon emissions generated during the transportation and transshipment of goods are greater than the penalty fee imposed by the national standards; In formula (5), w1+w2+w3=1; in formula (6), c before It represents the storage cost of the unit cargo arriving at the destination too early, c after It represents the penalty cost of late arrival of a unit of goods, T od represents the agreed delivery time, T represents the actual delivery time; E in formula (7) allow Indicates the allowable carbon emissions, e p Indicates the price of carbon emissions per unit weight; The constraints are: q min ≤q≤q max (12) Constraint (8) indicates that all nodes in the path comply with the conservation of cargo flow; Constraint (9) indicates that cargo can only adopt one mode of transportation on a transportation path; Constraint (10) indicates that cargo can change its mode of transportation at most once at a transfer node; Constraint (11) indicates that the transportation mode information on the transportation route before and after each transfer node is consistent with the node transportation mode conversion information; Constraint (12) indicates that the weight of the transported cargo is within a fixed range; Constraint (13) indicates that the transportation mode decision variable is a 01 variable; Constraint (14) indicates that the node transfer decision variable is a 01 variable.

2. The method for optimizing port multimodal transport routes under uncertain conditions according to claim 1, characterized in that: The model solution specifically includes the following steps: Step 1: Chromosome encoding It includes the transportation path node code and transportation mode code between the starting point and the end point. According to different contents, the path node selects the serial number code, and the transportation mode selects the letter code, where H represents road transportation, R represents rail transportation, and W represents water transportation; Step 2: Population initialization Generate T consecutive segments starting with the port number and ending with the destination number through random generation method, calculate the length of the path node segment, and randomly generate the corresponding transport mode chromosome segment according to the length, T is the population size; Step 3: Calculation of fitness and adaptive crossover and mutation probability The opposite number of the objective function is selected as the fitness of the chromosome, and the fitness set of the population is brought into the defined fuzzy inference system to calculate the crossover and mutation probabilities; Step 4: Select Using the replay sampling method of roulette wheel selection, the probability of each chromosome entering the next generation is set equal to the ratio of its fitness value to the sum of the fitness values ​​of individuals in the entire population, and random sampling is performed; Step 5: Cross The crossover operator uses a partial matching crossover operator, which first performs a crossover operation on the gene segments of the path nodes of the two chromosomes, and then crosses the gene segments corresponding to the transport mode; Step 6: Mutation The mutation operator uses the basic bit mutation operator. The transmission path node gene can be mutated to any node number in the transmission network, and the transportation mode gene can be mutated to any symbol in the {M, R, W} set, thereby generating a new population; Step 7: Verify the inner termination condition. Set the maximum number of inner iterations. If the optimal individual generated is not improved compared with the previous iteration or reaches the maximum number, end the inner loop and go to step 8; if there is an improvement, return to step 3; Step 8: Verify outer layer convergence Set the maximum number of outer iterations and calculate the current cost upper limit. If the upper limit has not decreased significantly compared with the previous iteration or has reached the maximum number of times, end the outer loop and output the solution set; if there is a significant decrease, return to step 2.

3. The method for optimizing port multimodal transport routes under uncertain conditions according to claim 2, characterized in that: The fuzzy inference system in step 3 specifically includes the following steps: Step 3-1: Calculate the fitness variance and extreme value First, find the corresponding objective function value C for the population individuals generated in the current round. i (x, y), and its reciprocal is taken as the fitness l of the individual in the current situation i , the formula is as follows: According to l i Find the fitness of the current centered individual And determine the individual fitness extreme difference R, the formula is as follows: According to Calculate the fitness distance d between each individual in the population and the central individual i And find the population fitness variance S, the formula is as follows Finally, the fitness variance S and extreme value difference R are normalized; Step 3-2: For the convenience of calculation, the membership function in the fuzzy inference system takes the triangular membership function; Step 3-3: Map it to different fuzzy indicators through appropriate fuzzification processing, and then use the prescribed fuzzy inference rules to transform it into fuzzy indicators related to mutation and crossover probability; Step 3-4: Finally, defuzzification is performed to obtain the adaptive crossover and mutation probabilities of the current population, and these are substituted into the genetic algorithm to solve the optimization problem.

Images