Cold storage type multi-temperature co-transport path optimization method considering carbon emission under time-varying road network

Abstract

The invention provides a cold storage type multi-temperature co-transportation path optimization method considering carbon emission under a time-varying road network, and the method comprises the steps: obtaining vehicle resources of a logistics center, and the product quantity and position information of a pickup place, and building a time dependence function of travel speed and a calculation function of carbon emission cost based on a current pickup transportation network; analyzing the satisfaction degree of the product quality based on the value loss generated in the process of driving from each pickup place to the logistics center; based on the calculation function of the carbon emission cost, a multi-objective optimization model with the purpose of minimizing the total transportation cost and maximizing the product satisfaction degree is constructed, and constraint conditions of a multi-temperature co-transportation path planning model are constructed based on vehicle resources and goods taking place information; and solving the multi-objective optimization model by using a two-stage optimization method combining K-means multi-dimensional clustering and a fast non-dominated sorting genetic algorithm to obtain the optimal departure time and transportation path. According to the method, the energy efficiency utilization rate in the cold chain transportation process is improved, and the distribution path planning is more scientific and reasonable.

Description

Technical Field

[0001] This application belongs to the field of cold chain logistics and intelligent transportation technology, and in particular relates to a method for optimizing multi-temperature co-transport routes with cold storage considering carbon emissions under time-varying road networks. Background Technology

[0002] Traditional cold chain logistics route optimization methods are mostly based on the assumption of a fixed road network, failing to fully consider the time-varying characteristics of urban traffic (such as congestion during morning and evening rush hours), resulting in a large deviation between route planning and actual travel time. In addition, most existing studies use mechanical refrigeration vehicles, which have problems such as high energy consumption, large carbon emissions, and poor temperature range flexibility.

[0003] Cold storage multi-temperature distribution technology combines pre-cooled cold storage boxes with ordinary trucks to dynamically adapt to different temperature zones. However, existing research lacks a comprehensive consideration of route optimization and carbon emissions under time-varying road networks. On the one hand, vehicle speeds in time-varying road networks fluctuate over time, affecting travel time, energy consumption, and cargo quality. On the other hand, carbon emission cost calculations are not deeply integrated with dynamic route optimization, making it difficult to achieve green logistics goals. Furthermore, traditional optimization algorithms are prone to getting trapped in local optima when solving large-scale, variable road network problems, resulting in slow convergence speeds.

[0004] Therefore, providing a route optimization method that can comprehensively handle time-varying road networks, cold storage multi-temperature distribution, carbon emission constraints, and cargo quality assurance, and designing an efficient solution algorithm to reduce total logistics costs and improve product satisfaction has become an urgent technical problem to be solved in this field. Summary of the Invention

[0005] In view of this, the purpose of this application is to provide a method for optimizing multi-temperature co-transport routes with carbon emissions in the context of time-varying road networks. This method can make the distribution route planning more scientific and reasonable, improve transportation efficiency and resource utilization while ensuring the quality of cold chain products, and promote the development of the cold chain logistics industry towards intelligence and greenness.

[0006] This application provides a method for optimizing multi-temperature co-transport routes with thermal storage considering carbon emissions under time-varying road networks, including: S1. Obtain vehicle resources from the logistics center, as well as product quantity and location information from the pickup location, and establish time-dependent functions for travel speed and carbon emission cost calculation functions based on the current pickup and delivery network. S2. Based on the value loss incurred during the journey of goods from various pick-up locations to the logistics center, the product quality grades are classified, and product quality satisfaction is analyzed. S3. Based on the calculation function of carbon emission cost, construct a multi-objective optimization model with the goal of minimizing total transportation cost and maximizing product satisfaction, and construct the constraints of the multi-temperature co-transportation route planning model based on vehicle resources and pickup location information. S4. Solve the multi-objective optimization model using a two-stage optimization method that combines K-means multidimensional clustering and fast non-dominated sorting genetic algorithm to obtain the optimal departure time and transportation route.

[0007] This application provides a time-varying road network-based cold storage multi-temperature co-transport route optimization method that considers carbon emissions, making delivery route planning more scientific and rational. This method can dynamically adjust vehicle departure times based on traffic conditions, effectively avoiding congested periods and significantly improving the transportation efficiency and resource utilization of cold chain logistics. By optimizing vehicle routes and departure time strategies, this application effectively reduces energy consumption and carbon emissions during transportation. Especially in complex cold chain transportation scenarios with multiple temperature zones and nodes, the optimization algorithm can intelligently balance transportation costs and product satisfaction, achieving a balance between economic efficiency and environmental protection. This application provides a comprehensive route optimization solution for fresh produce cold chain logistics, combining time-varying road network characteristics, cold storage multi-temperature co-transport technology, and carbon emission control. It not only improves transportation efficiency but also reduces operating costs and carbon emissions through precise route planning and departure time optimization. By implementing the technical solution of this application, enterprises can improve transportation efficiency and resource utilization while ensuring the quality of cold chain products, promoting the development of the cold chain logistics industry towards intelligence and green practices. Attached Figure Description

[0008] Figure 1 The flowchart of the cold storage multi-temperature co-transport path optimization method considering carbon emissions under time-varying road networks provided in this application embodiment is shown. Figure 2 The experimental results of the cold storage multi-temperature co-transport path optimization method considering carbon emissions under time-varying road networks provided in the embodiments of this application are shown in the figure. Figure 3 This paper presents a comparison chart of the total costs of electric vehicles and gasoline vehicles under optimized and non-optimized departure time scenarios, based on embodiments of this application. Detailed Implementation

[0009] To make the objectives, technical solutions, and advantages of this technical solution clearer, the following detailed description, in conjunction with specific embodiments, further illustrates this technical solution. It should be understood that these descriptions are merely exemplary and not intended to limit the scope of this technical solution.

[0010] Example 1:

[0011] Please see as follows Figure 1 The flowchart shown illustrates a method for optimizing multi-temperature co-transport routes with thermal storage, considering carbon emissions, under a time-varying road network. Figure 1 As shown, the method includes: S1. Obtain vehicle resources from the logistics center, as well as product quantity and location information from the pickup location. Based on the current pickup and delivery network, establish time-dependent functions for travel speed and calculation functions for carbon emission costs.

[0012] Specifically, the time-dependent function for establishing the travel speed in S1 includes: S11. Divide the service time of the logistics center into congested periods or normal periods. S12. Set the minimum vehicle speed, the maximum vehicle speed, the total speed, and the number of time periods. Use a random vehicle speed generation function to generate the vehicle speed for each time period. S13. Based on the location coordinates of all nodes in the pickup and delivery network, calculate the distances from the logistics center to different pickup locations and the distances between each pickup location. The calculation formula is as follows:

[0013] In the formula, i and j are nodes in the pickup and delivery network (i.e., logistics centers and pickup locations), and their coordinates are represented by... To indicate, This represents the distance between nodes i and j (i.e., unit road segment ij); S14. Based on the current traffic conditions of the pickup and delivery network, and according to the relationship between the vehicle's speed, distance, and time in each time period, and based on the first-in-first-out principle, calculate the vehicle's travel time in different time periods using the time period division method to obtain the total travel time of the vehicle.

[0014] In practice, the vehicle's travel time at different times is calculated using the following methods: S141. Divide the delivery time of the pickup and delivery network into r time periods of length L, and set the following parameters: T= For a set of time points, where For the r-th time period, Let r be the distance traveled by vehicle k during time period r. Let k be the speed of vehicle k during time interval r. Let r be the travel time of vehicle k within time period r. Let k be the remaining distance traveled by vehicle k on road segment ij outside of time period r. Let k be the amount of time that vehicle k can travel during the first time period. ,in, This marks the end of the first time period. Let be the time when vehicle k departs from node i; S142. Assume that the vehicle speed remains constant within the same time period r, and vehicle k starts from node i within time period r, i.e. ; S143. Calculate the travel time of vehicle k in the first time period using the following method: Given that the speed of vehicle k in the first time period is Based on the time that vehicle k can travel in the first time period Estimate the distance traveled by vehicle k in the first time period. :

[0015] Determine the distance traveled by vehicle k in the first time period. Distance between nodes i and j Size; like Then based on the distance of road segment ij Calculate the travel time of vehicle k in the first time period. :

[0016] Otherwise, it is determined that vehicle k has traveled the remaining distance on road segment ij outside the first time period. Therefore, the time that vehicle k can travel in the first time period is taken as the travel time of vehicle k in the first time period:

[0017] S144. Calculate the travel time of vehicle k in other time periods using the following method: Based on the speed of vehicle k during time period r Given the length L of the time period, calculate the distance traveled by vehicle k within the time period r. :

[0018] Determine the distance traveled by vehicle k within time period r. The remaining distance on road segment ij outside the time period r-1 for vehicle k. Size; like Then the length L of that time period is taken as the travel time of vehicle k within the time period r. :

[0019] At the same time, the remaining distance of vehicle k on road segment ij outside the time period r is obtained. :

[0020] Otherwise, based on the remaining distance of vehicle k on external road segment ij beyond time period r-1. And the known speed of vehicle k during time period r The travel time of vehicle k within time period r is obtained. :

[0021] Furthermore, the calculation function for carbon emission costs established in S1 specifically includes: S15. Establish a function for calculating the carbon emission cost of gasoline-powered vehicles using the following method: S151, The formula for calculating engine speed N is as follows:

[0022] In the formula, N is the engine speed, and S is the engine's highest gear. The ratio of engine speed to vehicle speed, where R(L) is the ratio of vehicle gears. The ratio of the highest gears of the vehicle, where v is the vehicle speed; S152, The formula for calculating engine power P is as follows:

[0023] In the formula, This refers to the engine's traction power. For the vehicle's traction efficiency. This refers to the power loss during output. in, The calculation formula is as follows:

[0024] In the formula, M represents the total weight of the vehicle being transported. Let g be the acceleration due to motion, g be the gravitational constant, and θ be the tilt angle. These are the coefficients of aerodynamic drag and rolling friction, respectively; ρ is the density of air; and A is the frontal contact area of ​​the vehicle. S153. The formula for calculating the fuel consumption rate (FR) per unit time is as follows:

[0025] In the formula, the relevant parameters of the engine include the friction coefficient Φ, the speed N, the displacement V, and the power P. air mass ratio, For conversion efficiency, This refers to the calorific value of the fuel. Substituting the engine speed N and engine power P into the fuel consumption rate FR per unit time, we get:

[0026] In the formula, λ = ξ / κψ, where ψ is the conversion factor. ω represents the empty vehicle weight, and f represents the loaded cargo weight. It is a constant. It is a coefficient that combines gradient, acceleration, and rolling resistance. , It is the coefficient of air resistance. ; S154. Based on the fuel consumption rate per unit time To obtain fuel consumption as follows:

[0027] In the formula, ΦNVd / v represents the engine module. Indicates speed module, Indicates the weight module; Then based on the carbon dioxide emission coefficient The carbon emission function for vehicles traveling on road segment ij during time period r is obtained as follows:

[0028] In the formula, This represents the comprehensive road resistance coefficient of the vehicle on road segment ij. The load capacity of the vehicle on road segment ij; S16. Establish a function for calculating the carbon emission cost of new energy vehicles using the following method: S161. Carbon emissions generated by the consumption of electricity by new energy vehicles during operation The calculation formula is as follows:

[0029] In the formula, The electrical energy consumed by new energy vehicles. The carbon emission intensity of the power structure, For charging efficiency, For power grid transmission and distribution efficiency; in, The calculation formula is as follows:

[0030] In the formula, ij represents the unit road segment, and m represents the total number of road segment units. Let n represent the types of new energy vehicles, and n be the total number of new energy vehicle types. For road segment ij The electrical energy consumed by new energy vehicles in operation for Average energy consumption per 100 km for new energy vehicles For unit road segment ij Traffic flow of new energy vehicles; S162, will Substitute the carbon emissions generated by the electricity consumed during the operation of new energy vehicles In the middle, obtain the pickup and delivery network The formula for calculating the carbon emissions generated by a new energy vehicle consuming electricity on road segment unit ij is as follows:

[0031] S2. Based on the value loss incurred during the journey of goods from various pickup locations to the logistics center, the product quality grades are classified, and product quality satisfaction is analyzed.

[0032] Specifically, the value loss incurred during the process of picking up goods from various pickup locations and transporting them to the logistics center in S2 includes: S21. When the temperature inside the insulated box is stable within the insulated time range, product loss is related to the transportation time. Based on the reasonable description that freshness follows an exponential distribution, the value loss cost of fresh products is measured by the product spoilage rate and the product of price and quantity.

[0033] In the formula, This is the set of pickup nodes; when the node number is 0, it represents the logistics center. For vehicle assembly, For different temperature zones, , For different types of products, , Indicates the temperature range of pickup location i. Supply volume of similar products express Unit price of similar products This indicates the temperature zone supplied by the loading and unloading location i for vehicle k. The initial freshness of similar products Indicates temperature zone The rate of spoilage of such products, among which, Let k be the transportation time from pickup location i to pickup location j. The earliest acceptable time for pickup location i. This indicates the temperature zone supplied by the loading and unloading location i for vehicle k. Decision variables for product categories; S22. When the insulated box exceeds the insulated time limit, meaning the temperature inside the insulated box has changed and gradually increased, product loss is not only related to the transportation time, but also gradually increases with the rise in temperature inside the insulated box. Based on the fresh product quality decay model, calculate the temperature range supplied by vehicle k at pickup location i at time t. The value loss cost of this type of product is:

[0034] In the formula, Let i be the time when vehicle k arrives at pickup point i. It is a temperature zone The reaction rate constant for this type of product, where E is the activation energy and G is the gas molar constant. It is the temperature inside the incubator at time t; The initial freshness of different products in different temperature zones was obtained through the following methods: The initial freshness is defined as the freshness index of a product immediately after leaving the pickup location. However, the temperature range varies depending on the type of product and the specific testing requirements. The formula for calculating the overall initial freshness of this type of product is as follows:

[0035] In the formula, Weights set by humans Temperature zone The non-conformity rate of this type of product Temperature zone Freshness deviation of similar products.

[0036] In practice, product quality levels are classified and product quality satisfaction is analyzed using the following methods: S23. Based on the initial freshness index, fresh produce is divided into three quality grades. The correspondence between each quality grade and the initial freshness index is as follows: When the freshness index is When the above conditions are met, the product quality grade is Grade I (fresh); when the freshness index is within a certain range... When the freshness index is between [a certain value], the product quality grade is Level II (secondary fresh); when the freshness index is [a certain value]... The following conditions indicate that the product quality grade is Level III rot. S24. Based on the initial freshness and corresponding product quality grade of the fresh products in S23, calculate the value loss of the fresh products during transportation using the loss function in S21 or S22, thereby obtaining the freshness and corresponding product quality grade of the fresh products when they arrive at the logistics center. S25. The freshness of the product upon arrival at the logistics center With pre-defined quality requirements Compare and determine product quality satisfaction. ; If the product's freshness index upon delivery Meets quality requirements If the quality is good upon delivery, the satisfaction level is 1; if the quality is poor upon delivery, the satisfaction level decreases; if the freshness index upon delivery is high... exist If a product is considered to have reached the point of decay, and the satisfaction level is 0, the following is a detailed explanation:

[0037] S3. Based on the calculation function of carbon emission cost, construct a multi-objective optimization model with the goal of minimizing total transportation cost and maximizing product satisfaction, and construct the constraints of the multi-temperature co-transportation route planning model based on vehicle resources and pickup location information.

[0038] The construction of a multi-objective optimization model with the objectives of minimizing total transportation costs and maximizing product satisfaction includes: S31. Introducing a carbon tax The carbon emission cost of gasoline-powered vehicles is:

[0039] In the formula, A represents the road segment between nodes. , As decision variables,

[0040] And, the carbon emission cost of new energy vehicles is:

[0041] S32. The cost of loss of goods value is:

[0042] In the formula, The insulation duration of the cold storage box; S33, Vehicle cost is:

[0043] In the formula, Fixed costs for vehicle k; And, the refrigeration cost is:

[0044] In the formula, Let be the number of cold storage boxes and cold storage units required at pickup location i, rounded to the nearest integer. The unit costs of the cold storage box and the cold storage unit are respectively. The electricity cost per unit time for product type W. Let k be the time when vehicle k arrives at node i. Let K be the time when vehicle K leaves the logistics center. For decision variables, ;

[0045] S34. Loading and unloading costs are:

[0046] In the formula, The cost of loading and unloading a single cold storage box; And, the time window penalty cost is:

[0047] In the formula, These represent the unit-time penalty costs for arriving early and late at the pickup point, respectively. The latest acceptable time for pickup location i. The penalty cost per unit time for exceeding the insulation time of the cold storage box; S35, Product Satisfaction includes Quality Satisfaction (S25). and time window satisfaction :

[0048] In the formula, These are the upper and lower limits of the tolerance time window for pickup location i, respectively; S36. Construct the optimization model that minimizes the total transportation cost as follows: minFall=min(F1+F2+F3+F4+F5+F6); And the optimization model for maximizing product satisfaction is:

[0049] In the formula, For product satisfaction, a1 and a2 are weighting coefficients, which are set according to the different degrees of attention that logistics and distribution companies pay to timeliness and quality.

[0050] Furthermore, the constraints for constructing a multi-temperature shared transport route planning model based on vehicle resources and pickup location information specifically include: S36. Constraints on pickup location and transportation route: S361. Limit each pickup location to one service only:

[0051] In the formula, As decision variables, For the pickup location, N'' = N\{0}; S362. Each vehicle is restricted to departing from the logistics center and returning to the original logistics center after the service is completed:

[0052] S363. Vehicles are restricted to picking up all products from all pickup locations:

[0053] S364. Eliminate sub-loops in the computation path:

[0054] In the formula, S' represents a subset of the pickup locations N'', and |S'| is the number of pickup locations; S37. Constraints on the product: S371. Limit the total weight of products picked up at pickup location i:

[0055] In the formula, For pickup location i pair Product supply quantity, when i=0, =0; S372. The capacity of the product supplied to the pick-up location i must be less than the weight that the cold storage box can hold.

[0056] In the formula, For the cold storage box The capacity of this type of product; S373, The weight of products loaded between road sections ij is limited to less than the maximum load capacity of the vehicle:

[0057] In the formula, The maximum load capacity for each vehicle; S374. The number of cold storage boxes is less than the total number of vehicles that can accommodate them, where k is an integer.

[0058]

[0059] In the formula, Let k be the number of cold storage boxes that vehicle k can carry, rounded to the nearest integer. S38. Constraints on various vehicle time parameters: S381, Departure time of the vehicle:

[0060] In the formula, Service hours at the pickup location; S382. Vehicle arrival time at the next pickup location:

[0061] S383, the time a vehicle spends traveling on road segment ij:

[0062] S4. Solve the multi-objective optimization model using a two-stage optimization method that combines K-means multidimensional clustering and fast non-dominated sorting genetic algorithm to obtain the optimal departure time and transportation route.

[0063] The multi-objective optimization model is solved using a two-stage optimization method combining K-means multidimensional clustering and a fast non-dominated sorting genetic algorithm. Specifically, this includes: S41. Combine the cargo attributes from different pickup locations and apply the K-means algorithm to complete the clustering operation: S411. Define the multidimensional data for the pickup location as follows: Geographical location Looking forward to the time window Tolerance time window Product demand in different temperature zones Quality requirements And the geographical location of the logistics center, and randomly selected The pickup locations were used as the initial cluster centers; S412. Assign a weight to each dimension of data, calculate the distance from each pickup location to each cluster center using weighted distance, and assign the pickup location to the nearest cluster center; the weighted distance formula is as follows:

[0064] S413. Calculate the weighted average distance and update the cluster centers:

[0065] In the formula, For the number of features, The weight of the Fth feature is... This is the Fth eigenvalue; Repeat steps S412-S413 until the cluster centers no longer change, and then assign the cluster centers to the logistics centers. S42. Apply the fast non-dominated sorting genetic algorithm to continuously adjust the transport routes within cluster units to obtain the optimal solution: S421. For each individual, two parameters are preset to determine the non-dominance condition: the number of other individuals who dominate that individual. and the solution set dominated by that individual. ; S422. First, in the population Find all individuals with a dominance of 0, i.e., all individuals that cannot dominate individual u, and place them in the first non-dominated layer Z1; then, in Z1, find the solution set dominated by each individual e. Find each individual h in the hierarchy; finally, since the individual e that dominates individual h has already been placed in the non-dominated layer Z1, then... Decrease by 1 until... When the value is 0, individual h is placed in the second non-dominated layer Z2; this process is repeated continuously to classify the entire population. S423. To avoid local optima and ensure a more even distribution of solutions, calculate the winner based on the distribution of an individual within the same non-dominated layer and its neighbors: First, according to a certain objective function value... Sort by size and record the maximum and minimum values ​​of the individual objective functions. Then, the crowding level of the boundary individuals is set to infinity, and the crowding level of the remaining individuals is set to... To indicate:

[0066] Finally, the total crowding degree of each individual is the result of accumulating the crowding degrees under both objective functions; At this point, all individuals in the population have obtained a non-dominated order rank and a crowding degree crowd. Individual u is better than individual e if and only if u_rank > e_rank or u_rank = e_rank and u_crowd > e_crowd. The above two conditions for judging the optimality of an individual are the crowding degree comparison operator. S424. Individuals selected from the parent population are genetically processed to generate offspring populations. The offspring and parent populations, both of which are of size NP, are merged to generate a new population of size 2NP. Based on the results of the non-dominated sorting, different levels of non-dominated layers Zg are generated. If Z1 and Z2 do not fill the new population to size NP, the crowding degree among individuals in Z3 is calculated. The optimal individual is selected from the crowding degree comparison operator until the new population size reaches NP. S425. Use natural numbers to represent all pickup locations in the example. Make judgments based on the constraints of the model and add the pickup locations to the current delivery vehicle's route in turn. If the condition cannot be met, arrange another vehicle and continue the previous operation until all pickup locations are put into the delivery route. The specific encoding operation is as follows: Vehicles pick up goods from pickup locations 1, 2, 3, 4, 5, 6, 7, and 8. The beginning and end points are both represented by 0 to indicate logistics centers. The encoding result is 0-2-3-0-5-6-0-4-7-8-1-0, indicating that a total of 3 vehicles are needed to complete the pickup service for all pickup locations. This process continues to generate the initial path of the model. By encoding and decoding the chromosomes and traversing the transportation path of each vehicle, the key quantities for calculating the objective function can be obtained, thus determining the decision variables and objective function values ​​corresponding to each solution. The selection operation is performed under the binary tournament method. First, the number of individuals to be selected from the population each time is determined. Then, based on the objective function value stored in the second half of the chromosome, a fast non-dominated sorting method is used to obtain high-ranking individuals with priority selection rights. If the individuals are of the same rank, the crowding degree is used for comparison, and individuals with higher crowding degree are selected from those of the same rank. Since each pickup location can only accept one service, the crossover operation uses partial mapping to randomly select a pair of parent individuals and find the gene segments at the same position; the positions of the two sets of gene segments are exchanged to produce offspring individuals; 1 and 3, 9 and 4, 2 and 5 in the two sets of gene segments are mapped to each other, and gene conflicts on chromosomes are eliminated according to the mapping relationship after the exchange; the mutation operation uses interval reversal, randomly finds an interval on the gene segment of the parent individual, and reverses the gene in this interval to obtain offspring individuals.

[0067] Example 2, Verification Experiment: The population size NP is set to 80, the maximum number of iterations is 500, and the crossover and mutation probabilities are Pc=0.8 and Pm=0.2, respectively.

[0068] The RC202 dataset was modified based on the Solomon-100 dataset, where number 0 represents the logistics center, and 30 pickup points are represented by natural numbers from 1 to 30. The product weight of each temperature zone at each pickup point must not exceed 400 kg. First-level or second-level quality requirements and tolerance time windows of 10 to 60 minutes are randomly generated, and a1 and a2 are set to 0.4 and 0.6, respectively.

[0069] The main parameters are set as follows: temperature of each temperature zone is set to [10-16, 0-9, -18], product spoilage rate is [0.06, 0.08, 0.06], unit price of product in each temperature zone is [8, 12, 15] yuan / kg, reaction rate is [0.005, 0.08, 0.01], required number of cold storage units is [2, 5, 8], cold storage unit cost is [0.2, 0.26, 0.65], heat preservation time is 10h, and temperature changes will occur after the heat preservation time is exceeded. The speed is 0.02, the load capacity of the transport vehicle is 2000kg, the fixed cost is 70 yuan / vehicle, the number of cold storage boxes that can be loaded is 20, and the transportation cost per unit distance is 0.65 yuan / km; the unit cost of the cold storage box is 0.5 yuan / box, and the loading and unloading cost is 0.25 yuan / box; the waiting cost when arriving at the pickup point early is 0.02 yuan / min, the waiting cost when arriving at the pickup point late is 0.04 yuan / min, and the cost exceeding the cold storage and insulation time is 0.1 yuan / min.

[0070] Since relevant information about transport vehicles is needed when calculating carbon emissions, the specific parameters are shown in Table 1.

[0071] Table 1. Relevant parameters of transport vehicles

[0072] To verify the effectiveness of the two-stage optimization method proposed in this invention, which combines K-means multidimensional clustering and a fast non-dominated sorting genetic algorithm (hereinafter referred to as the NSGA-II algorithm), in the optimization method of cold storage multi-temperature co-transportation routes (i.e., cold chain logistics route planning), systematic experimental tests and analyses were conducted. The experimental results are as follows: Figure 2 As shown in the figure, the optimized vehicle travel route and the pickup order at each pickup point are clearly illustrated. The results show that all pickup points were rationally allocated and successfully completed the pickup service, indicating that the two-stage algorithm has good feasibility and completeness in route planning. Further analysis shows that this method, while ensuring the quality of fresh produce transportation, reduces the dimensionality of the problem by grouping pickup points using K-means clustering and combines this with the improved NSGA-II algorithm for multi-objective optimization, significantly improving the economy of the route plan. Specifically, Figure 2 The optimized path shown effectively reduces vehicle travel distance and time costs, while the reasonable arrangement of pickup order reduces the risk of default within the time window. These calculation results fully demonstrate the superiority of the algorithm in solving the multi-temperature co-distribution path optimization problem under time-varying road networks, providing reliable technical support for cost reduction and efficiency improvement in cold chain logistics.

[0073] To demonstrate that this application fully considers carbon emissions in the optimization of multi-temperature co-transport cold chain logistics routes using cold storage, it makes carbon emissions one of the key optimization objectives. Data analysis in Table 2 shows that electric vehicles exhibit significant environmental advantages compared to traditional fuel vehicles. Their carbon emissions are lower than those of fuel vehicles under all three algorithms. In particular, the method combining K-means clustering and the improved NSGA-II algorithm results in electric vehicle carbon emissions of only 87.94 kg, a 22.4% reduction compared to fuel vehicles. More importantly, the algorithm combination proposed in this application demonstrates excellent performance in carbon emission control. Compared to the single K-means clustering algorithm, electric vehicle carbon emissions are further reduced by 14.05%, and compared to the NSGA-II algorithm, by 31.11%, while the total time is shortened to 9.87 minutes, and the number of vehicles remains at a reasonable level of 5. This fully demonstrates that the algorithm combination not only effectively solves the route optimization problem under time-varying road networks but also minimizes carbon emissions through intelligent grouping and multi-objective optimization, providing a reliable technical solution for green cold chain logistics.

[0074] Table 2. Vehicle carbon emissions from the optimized route

[0075] To demonstrate the advantages of optimizing departure time in time-varying road networks, the total costs of electric vehicles and gasoline vehicles are compared under optimized and non-optimized departure time scenarios. The results are as follows: Figure 3 As shown. From Figure 3 The results show that optimizing departure time has a significant positive impact on the total cost of both vehicle types. In the electric vehicle scenario, the optimized total cost decreased from RMB 2,189 to RMB 1,968, a reduction of 10.1%; in the gasoline vehicle scenario, the total cost decreased from RMB 2,315 to RMB 2,160, a reduction of 6.7%. This significant cost reduction is mainly due to the departure time optimization strategy effectively avoiding traffic congestion periods, allowing vehicles to maintain higher speeds during transportation, thereby reducing high energy consumption and time losses caused by low-speed driving and traffic jams, optimizing overall transportation efficiency, and avoiding additional penalty costs due to time window violations. This fully demonstrates the practical value of dynamic path planning under time-varying road networks.

[0076] To verify the superiority and performance of the proposed algorithm combining K-means clustering and the improved NSGA-II algorithm in model solving, it was compared with the single K-means clustering algorithm and the NSGA-II algorithm. The comparative analysis was conducted under the condition that the product quality satisfaction threshold was set at 90%, and the results are shown in Table 3. It can be seen that the algorithm of this invention shows significant advantages in both trolleybus and gasoline vehicle scenarios: in trolleybus applications, the total cost was reduced to 1968, a decrease of 8.9% compared to the K-means clustering algorithm and a decrease of 15.7% compared to the NSGA-II algorithm; at the same time, the product quality satisfaction was increased to 0.946, which is 3.1% and 4.5% better than the comparative algorithms, respectively. Similar optimization effects were also observed in gasoline vehicle applications. This result fully demonstrates that this application, through algorithmic fusion innovation, effectively optimizes the total transportation cost while ensuring high product quality satisfaction, showcasing its comprehensive performance advantages in solving multi-objective path optimization problems.

[0077] Table 3. Comparison of algorithm results when the product quality satisfaction threshold is 90%.

[0078] In summary, the cold storage multi-temperature co-transport route optimization method considering carbon emissions under time-varying road networks provided in this application makes delivery route planning more scientific and rational. This method can dynamically adjust vehicle departure times according to traffic conditions, effectively avoiding congested periods and significantly improving the transportation efficiency and resource utilization of cold chain logistics. By optimizing vehicle routes and departure time strategies, this application effectively reduces energy consumption and carbon emissions during transportation. Especially in complex cold chain transportation scenarios with multiple temperature zones and nodes, the optimization algorithm can intelligently balance transportation costs and product satisfaction, achieving a balance between economic efficiency and environmental protection. This application provides a comprehensive route optimization solution for fresh produce cold chain logistics, combining time-varying road network characteristics, cold storage multi-temperature co-transport technology, and carbon emission control. It not only improves transportation efficiency but also reduces operating costs and carbon emissions through precise route planning and departure time optimization. By implementing the technical solution of this application, enterprises can improve transportation efficiency and resource utilization while ensuring the quality of cold chain products, promoting the development of the cold chain logistics industry towards intelligence and greening.

[0079] The above content is only a preferred embodiment of the present invention. For those skilled in the art, many changes can be made in the specific implementation and application scope based on the ideas of the present invention. As long as these changes do not depart from the concept of the present invention, they all fall within the protection scope of the present invention.

Claims

1. A method for optimizing multi-temperature co-transport routes with thermal storage considering carbon emissions under time-varying road networks, characterized in that, The method includes: S1. Obtain vehicle resources from the logistics center, as well as product quantity and location information from the pickup location, and establish time-dependent functions for travel speed and carbon emission cost calculation functions based on the current pickup and delivery network. S2. Based on the value loss incurred during the journey of goods from various pick-up locations to the logistics center, the product quality grades are classified, and product quality satisfaction is analyzed. S3. Based on the calculation function of carbon emission cost, construct a multi-objective optimization model with the goal of minimizing total transportation cost and maximizing product satisfaction, and construct the constraints of the multi-temperature co-transportation route planning model based on vehicle resources and pickup location information. S4. Solve the multi-objective optimization model using a two-stage optimization method that combines K-means multidimensional clustering and fast non-dominated sorting genetic algorithm to obtain the optimal departure time and transportation route.

2. The method as described in claim 1, characterized in that, The time-dependent function for establishing the travel speed in S1 specifically includes: S11. Divide the service time of the logistics center into congested periods or normal periods. S12. Set the minimum vehicle speed, the maximum vehicle speed, the total speed, and the number of time periods. Use a random vehicle speed generation function to generate the vehicle speed for each time period. S13. Based on the location coordinates of all nodes in the pickup and delivery network, calculate the distances from the logistics center to different pickup locations and the distances between each pickup location. The calculation formula is as follows: In the formula, i and j are nodes in the pickup and delivery network (i.e., logistics centers and pickup locations), and their coordinates are represented by... To indicate, This represents the distance between nodes i and j (i.e., unit road segment ij); S14. Based on the current traffic conditions of the pickup and delivery network, and according to the relationship between the vehicle's speed, distance, and time in each time period, and based on the first-in-first-out principle, calculate the vehicle's travel time in different time periods using the time period division method to obtain the total travel time of the vehicle.

3. The method as described in claim 2, characterized in that, The vehicle's travel time at different times is calculated using the following method: S141. Divide the delivery time of the pickup and delivery network into r time periods of length L, and set the following parameters: T= For a set of time points, where For the r-th time period, Let r be the distance traveled by vehicle k during time period r. Let k be the speed of vehicle k during time interval r. Let r be the travel time of vehicle k within time period r. Let k be the remaining distance traveled by vehicle k on road segment ij outside of time period r. Let k be the amount of time that vehicle k can travel during the first time period. ,in, This marks the end of the first time period. Let be the time when vehicle k departs from node i; S142. Assume that the vehicle speed remains constant within the same time period r, and vehicle k starts from node i within time period r, i.e. ; S143. Calculate the travel time of vehicle k in the first time period using the following method: Given that the speed of vehicle k in the first time period is Based on the time that vehicle k can travel in the first time period Estimate the distance traveled by vehicle k in the first time period. : Determine the distance traveled by vehicle k in the first time period. Distance between nodes i and j Size; like Then based on the distance of road segment ij Calculate the travel time of vehicle k in the first time period. : Otherwise, it is determined that vehicle k has traveled the remaining distance on road segment ij outside the first time period. Therefore, the time that vehicle k can travel in the first time period is taken as the travel time of vehicle k in the first time period: S144. Calculate the travel time of vehicle k in other time periods using the following method: Based on the speed of vehicle k during time period r Given the length L of the time period, calculate the distance traveled by vehicle k within the time period r. : Determine the distance traveled by vehicle k within time period r. The remaining distance on road segment ij outside the time period r-1 for vehicle k. Size; like Then the length L of that time period is taken as the travel time of vehicle k within the time period r. : At the same time, the remaining distance of vehicle k on road segment ij outside the time period r is obtained. : Otherwise, based on the remaining distance of vehicle k on external road segment ij beyond time period r-1. And the known speed of vehicle k during time period r The travel time of vehicle k within time period r is obtained. :

4. The method as described in claim 1, characterized in that, The calculation function for carbon emission costs established in S1 specifically includes: S15. Establish a function for calculating the carbon emission cost of gasoline-powered vehicles using the following method: S151, The formula for calculating engine speed N is as follows: In the formula, N is the engine speed, and S is the engine's highest gear. The ratio of engine speed to driving speed, This represents the ratio of the vehicle's gears. The ratio of the highest gears of the vehicle, where v is the vehicle speed; S152, The formula for calculating engine power P is as follows: In the formula, This refers to the engine's traction power. For the vehicle's traction efficiency. This refers to the power loss during output. in, The calculation formula is as follows: In the formula, M represents the total weight of the vehicle being transported. Let g be the acceleration due to motion, g be the gravitational constant, and θ be the tilt angle. These are the coefficients of aerodynamic drag and rolling friction, respectively; ρ is the density of air; and A is the frontal contact area of ​​the vehicle. S153. The formula for calculating the fuel consumption rate (FR) per unit time is as follows: In the formula, the relevant parameters of the engine include the friction coefficient Φ, the speed N, the displacement V, and the power P. air mass ratio, For conversion efficiency, This refers to the calorific value of the fuel. Substituting the engine speed N and engine power P into the fuel consumption rate FR per unit time, we get: In the formula, λ = ξ / κψ, where ψ is the conversion factor. ω represents the empty vehicle weight, and f represents the loaded cargo weight. It is a constant. It is a coefficient that combines gradient, acceleration, and rolling resistance. , It is the coefficient of air resistance. ; S154. Based on the fuel consumption rate per unit time To obtain fuel consumption as follows: In the formula, ΦNVd / v represents the engine module. Indicates speed module, Indicates the weight module; Then based on the carbon dioxide emission coefficient The carbon emission function for vehicles traveling on road segment ij during time period r is obtained as follows: In the formula, This represents the comprehensive road resistance coefficient of the vehicle on road segment ij. The load capacity of the vehicle on road segment ij; S16. Establish a function for calculating the carbon emission cost of new energy vehicles using the following method: S161. Carbon emissions generated by the consumption of electricity by new energy vehicles during operation The calculation formula is as follows: In the formula, The electrical energy consumed by new energy vehicles. The carbon emission intensity of the power structure, For charging efficiency, For power grid transmission and distribution efficiency; in, The calculation formula is as follows: In the formula, ij represents the unit road segment, and m represents the total number of road segment units. Let n represent the types of new energy vehicles, and n be the total number of new energy vehicle types. For road segment ij The electrical energy consumed by new energy vehicles in operation for Average energy consumption per 100 km for new energy vehicles For unit road segment ij Traffic flow of new energy vehicles; S162, will Substitute the carbon emissions generated by the electricity consumed during the operation of new energy vehicles In the middle, obtain the pickup and delivery network The formula for calculating the carbon emissions generated by a new energy vehicle consuming electricity on road segment unit ij is as follows:

5. The method as described in claim 1, characterized in that, The method, specifically including the value loss incurred during the process of picking up goods from various pickup locations and transporting them to the logistics center in step S2, includes: S21. When the temperature inside the insulated box is stable within the insulated time range, product loss is related to the transportation time. Based on the reasonable description that freshness follows an exponential distribution, the value loss cost of fresh products is measured by the product spoilage rate and the product of price and quantity. In the formula, This is the set of pickup nodes; when the node number is 0, it represents the logistics center. For vehicle assembly, For different temperature zones, , For different types of products , Indicates the temperature range of pickup location i. Supply volume of similar products express Unit price of similar products This indicates the temperature zone supplied by the loading and unloading location i for vehicle k. The initial freshness of similar products Indicates temperature zone The rate of spoilage of such products, among which, Let k be the transportation time from pickup location i to pickup location j. The earliest acceptable time for pickup location i. This indicates the temperature zone supplied by the loading and unloading location i for vehicle k. Decision variables for product categories; S22. When the insulated box exceeds the insulated time limit, meaning the temperature inside the insulated box has changed and gradually increased, product loss is not only related to the transportation time, but also gradually increases with the rise in temperature inside the insulated box. Based on the fresh product quality decay model, calculate the temperature range supplied by vehicle k at pickup location i at time t. The value loss cost of this type of product is: In the formula, Let i be the time when vehicle k arrives at pickup point i. It is a temperature zone The reaction rate constant for this type of product, where E is the activation energy and G is the gas molar constant. It is the temperature inside the incubator at time t; The initial freshness of different products in different temperature zones was obtained through the following methods: The initial freshness is defined as the freshness index of a product immediately after leaving the pickup location. However, the temperature range varies depending on the type of product and the specific testing requirements. The formula for calculating the overall initial freshness of this type of product is as follows: In the formula, Weights set by humans Temperature zone The non-conformity rate of this type of product Temperature zone Freshness deviation of similar products.

6. The method as described in claim 1, characterized in that, In step S2, product quality levels are classified and product quality satisfaction is analyzed in the following ways: S23. Based on the initial freshness index, fresh produce is divided into three quality grades. The correspondence between each quality grade and the initial freshness index is as follows: When the freshness index is When the above conditions are met, the product quality grade is Grade I (fresh); when the freshness index is within a certain range... When the freshness index is between [a certain value], the product quality grade is Level II (secondary fresh); when the freshness index is [a certain value]... The following conditions indicate that the product quality grade is Level III rot. S24. Based on the initial freshness and corresponding product quality grade of the fresh products in S23, calculate the value loss of the fresh products during transportation using the loss function in S21 or S22, thereby obtaining the freshness and corresponding product quality grade of the fresh products when they arrive at the logistics center. S25. The freshness of the product upon arrival at the logistics center With pre-defined quality requirements Compare and determine product quality satisfaction. ; If the product's freshness index upon delivery Meets quality requirements If the quality is good upon delivery, the satisfaction level is 1; if the quality is poor upon delivery, the satisfaction level decreases; if the freshness index upon delivery is high... exist If a product is considered to have reached the point of decay, and the satisfaction level is 0, the following is a detailed explanation:

7. The method as described in claim 1, characterized in that, The S3 section constructs a multi-objective optimization model with the objectives of minimizing total transportation costs and maximizing product satisfaction, specifically including: S31. Introducing a carbon tax The carbon emission cost of gasoline-powered vehicles is: In the formula, A represents the road segment between nodes, A={(i,j), i≠j, i, j∈N'}. As decision variables, ; And, the carbon emission cost of new energy vehicles is: S32. The cost of loss of goods value is: In the formula, The insulation duration of the cold storage box; S33, Vehicle cost is: In the formula, Fixed costs for vehicle k; And, the refrigeration cost is: In the formula, Let be the number of cold storage boxes and cold storage units required at pickup location i, rounded to the nearest integer. The unit costs of the cold storage box and the cold storage unit are respectively. The electricity cost per unit time for product type W. Let k be the time when vehicle k arrives at node i. Let K be the time when vehicle K leaves the logistics center. As decision variables, S34. Loading and unloading costs are: In the formula, The cost of loading and unloading a single cold storage box; And, the time window penalty cost is: In the formula, These represent the unit-time penalty costs for arriving early and late at the pickup point, respectively. The latest acceptable time for pickup location i. The penalty cost per unit time for exceeding the insulation time of the cold storage box; S35, Product Satisfaction includes Quality Satisfaction (S25). and time window satisfaction : In the formula, These are the upper and lower limits of the tolerance time window for pickup location i, respectively; S36. Construct the optimization model that minimizes the total transportation cost as follows: minFall=min(F1+F2+F3+F4+F5+F6); And the optimization model for maximizing product satisfaction is: In the formula, For product satisfaction, This is a weighting coefficient, set according to the different levels of emphasis that logistics and distribution companies place on timeliness and quality.

8. The method as described in claim 1, characterized in that, The constraints in S3 for constructing the multi-temperature shared transport route planning model based on vehicle resources and pickup location information specifically include: S36. Constraints on pickup location and transportation route: S361. Limit each pickup location to one service only: In the formula, As decision variables, For pickup location, ; S362. Each vehicle is restricted to departing from the logistics center and returning to the original logistics center after the service is completed: S363. Vehicles are restricted to picking up all products from all pickup locations: S364. Eliminate sub-loops in the computation path: In the formula, S' represents a subset of the pickup locations N'', and |S'| is the number of pickup locations; S37. Constraints on the product: S371. Limit the total weight of products picked up at pickup location i: In the formula, For pickup location i pair Product supply quantity, when i=0, =0; S372, Restriction of pickup location i The capacity of this type of product supplied must be less than the weight that the cold storage box can hold. In the formula, For the cold storage box The capacity of this type of product; S373, The weight of products loaded between road sections ij is limited to less than the maximum load capacity of the vehicle: In the formula, The maximum load capacity for each vehicle; S374. The number of cold storage boxes is less than the total capacity of all vehicles for them, where k is an integer. In the formula, Let k be the number of cold storage boxes that vehicle k can carry, rounded to the nearest integer. S38. Constraints on various vehicle time parameters: S381, Departure time of the vehicle: In the formula, Service hours at the pickup location; S382. Vehicle arrival time at the next pickup location: S383, the time a vehicle spends traveling on road segment ij:

9. The method as described in claim 1, characterized in that, The S4 section utilizes a two-stage optimization method combining K-means multidimensional clustering and a fast non-dominated sorting genetic algorithm to solve the multi-objective optimization model, specifically including: S41. Combine the cargo attributes from different pickup locations and apply the K-means algorithm to complete the clustering operation: S411. Define the multidimensional data for the pickup location as follows: Geographical location Looking forward to the time window Tolerance time window Product demand in different temperature zones Quality requirements And the geographical location of the logistics center, and randomly selected The pickup locations were used as the initial cluster centers; S412. Assign a weight to each dimension of data, calculate the distance from each pickup location to each cluster center using weighted distance, and assign the pickup location to the nearest cluster center; the weighted distance formula is as follows: S413. Calculate the weighted average distance and update the cluster centers: In the formula, For the number of features, The weight of the Fth feature is... This is the Fth eigenvalue; Repeat steps S412-S413 until the cluster centers no longer change, and then assign the cluster centers to the logistics centers. S42. Apply the fast non-dominated sorting genetic algorithm to continuously adjust the transport routes within cluster units to obtain the optimal solution: S421. For each individual, two parameters are preset to determine the non-dominance condition: the number of other individuals who dominate that individual. and the solution set dominated by that individual. ; S422. First, in the population Find all individuals with a dominance of 0, i.e., all individuals that cannot dominate individual u, and place them in the first non-dominated layer Z1; then, in Z1, find the solution set dominated by each individual e. Find each individual h in the hierarchy; finally, since the individual e that dominates individual h has already been placed in the non-dominated layer Z1, then... Decrease by 1 until... When the value is 0, individual h is placed in the second non-dominated layer Z2; this process is repeated continuously to classify the entire population. S423. To avoid local optima and ensure a more even distribution of solutions, calculate the winner based on the distribution of an individual within the same non-dominated layer and its neighbors: First, according to a certain objective function value... Sort by size and record the maximum and minimum values ​​of the individual objective functions. Then, the crowding level of the boundary individuals is set to infinity, and the crowding level of the remaining individuals is set to... To indicate: Finally, the total crowding degree of each individual is the result of accumulating the crowding degrees under both objective functions; At this point, all individuals in the population have obtained a non-dominated order rank and a crowding degree crowd. Individual u is better than individual e if and only if u_rank > e_rank or u_rank = e_rank and u_crowd > e_crowd. The above two conditions for judging the optimality of an individual are the crowding degree comparison operator. S424. Individuals selected from the parent population are genetically processed to generate offspring populations. The offspring and parent populations, both of which are of size NP, are merged to generate a new population of size 2NP. Based on the results of the non-dominated sorting, different levels of non-dominated layers Zg are generated. If Z1 and Z2 do not fill the new population to size NP, the crowding degree among individuals in Z3 is calculated. The optimal individual is selected from the crowding degree comparison operator until the new population size reaches NP. S425. Use natural numbers to represent all pickup locations in the example. Make judgments based on the constraints of the model and add the pickup locations to the current delivery vehicle's route in turn. If the condition cannot be met, arrange another vehicle and continue the previous operation until all pickup locations are put into the delivery route. The specific encoding operation is as follows: Vehicles pick up goods from pickup locations 1, 2, 3, 4, 5, 6, 7, and 8. The beginning and end points are both represented by 0 to indicate logistics centers. The encoding result is 0-2-3-0-5-6-0-4-7-8-1-0, indicating that a total of 3 vehicles are needed to complete the pickup service for all pickup locations. This process continues to generate the initial path of the model. By encoding and decoding the chromosomes and traversing the transportation path of each vehicle, the key quantities for calculating the objective function can be obtained, thus determining the decision variables and objective function values ​​corresponding to each solution. The selection operation is performed under the binary tournament method. First, the number of individuals to be selected from the population each time is determined. Then, based on the objective function value stored in the second half of the chromosome, a fast non-dominated sorting method is used to obtain high-ranking individuals with priority selection rights. If the individuals are of the same rank, the crowding degree is used for comparison, and individuals with higher crowding degree are selected from those of the same rank. Since each pickup location can only accept one service, the crossover operation uses partial mapping to randomly select a pair of parent individuals and find the gene segments at the same position; the positions of the two sets of gene segments are exchanged to produce offspring individuals; 1 and 3, 9 and 4, 2 and 5 in the two sets of gene segments are mapped to each other, and gene conflicts on chromosomes are eliminated according to the mapping relationship after the exchange; the mutation operation uses interval reversal, randomly finds an interval on the gene segment of the parent individual, and reverses the gene in this interval to obtain offspring individuals.

Images