Cold-chain logistics distribution path optimization method

Abstract

The application discloses a cold-chain logistics distribution path optimization method and belongs to the technical field of logistics distribution path optimization. The method comprises the following steps: S1, collecting and preprocessing cold-chain distribution basic data: collecting distribution center basic information, distribution node operation data, cold-chain commodity data, city time-varying road network traffic data and cold storage distribution vehicle parameter data, and performing standardized preprocessing on the collected data; and S2, constructing a customer satisfaction comprehensive evaluation model. The cold-chain logistics distribution path optimization method constructs a double-target optimization model of minimizing distribution cost and maximizing customer satisfaction, breaks through the limitation of traditional single-cost target optimization, realizes the collaborative optimization of cold-chain distribution cost control and service quality improvement through the satisfaction index of quantified distribution time window and goods freshness, has strong universality, is suitable for cold-chain logistics full-category and multi-scenario distribution requirements, and is suitable for various cold-chain distribution scenarios such as chain convenience stores, fresh food e-commerce, medical cold chain, supermarket cold chain and the like.

Description

Technical Field

[0001] This invention relates to the field of logistics distribution route optimization technology, and in particular to a method for optimizing cold chain logistics distribution routes. Background Technology

[0002] With the rapid development of the cold chain logistics industry and the upgrading of residents' consumption, the market demand for fresh food, pharmaceutical cold chain, and refrigerated beverages continues to grow, placing extremely high demands on the timeliness control, temperature control, route planning, and cost control of cold chain logistics distribution. Cold chain logistics differs from ambient temperature logistics; it requires maintaining the necessary low-temperature environment for goods throughout the entire process. During distribution, it is necessary to consider not only the length of the transportation route but also multiple constraints such as temperature control energy consumption, freshness degradation, and matching delivery time windows, demanding a higher degree of precision in route planning.

[0003] Currently, cold chain delivery generally suffers from high costs and low customer satisfaction. Route planning relies on manual experience, static road network models are out of touch with time-varying traffic scenarios, and unreasonable vehicle configuration leads to low capacity utilization and serious energy waste. Existing optimization solutions mostly focus on cost as the sole objective, ignoring the impact of delivery timeliness and product freshness on customer satisfaction. Furthermore, single algorithms are prone to getting stuck in local optima in large-scale, multi-constraint scenarios, resulting in slow convergence, low accuracy, and difficulty in adapting to the diverse and high-frequency delivery needs of cold chain logistics. Summary of the Invention

[0004] To address the aforementioned shortcomings of existing technologies, the technical problems to be solved by this invention are: reliance on experience in cold chain delivery route planning, high delivery costs, low customer satisfaction, unreasonable vehicle configuration, disconnect between static road network models and reality, and insufficient accuracy and efficiency of algorithm solutions. This invention provides a cold chain logistics delivery route optimization method that can achieve multi-objective synergistic optimization of delivery costs and customer satisfaction, providing a scientific, efficient, and universal route planning solution for cold chain logistics delivery across all categories and scenarios.

[0005] To solve the above problems, the following technical solutions are provided: The design of cold chain logistics distribution route optimization methods includes the following steps: S1. Cold chain distribution basic data collection and preprocessing: Collect basic information of distribution centers, operational data of distribution nodes, cold chain commodity data, urban time-varying road network traffic data, and parameter data of refrigerated delivery vehicles, and perform standardized preprocessing on the collected data; S2. Construct a comprehensive customer satisfaction evaluation model: Select delivery time window matching degree and goods freshness as core evaluation indicators, construct delivery time satisfaction function and goods freshness satisfaction function respectively, and obtain a comprehensive customer satisfaction evaluation model by weighted fusion; S3. Construct a multi-objective cold chain delivery route optimization model: With the dual optimization objectives of minimizing total delivery cost and maximizing customer satisfaction, combine the basic data from step S1 and the comprehensive customer satisfaction evaluation model from step S2, set constraints, and construct a multi-objective optimization mathematical model. S4. Solve the model using a hybrid genetic particle swarm optimization algorithm: Combine the local optimization capability of the genetic algorithm with the global search capability of the particle swarm optimization algorithm to design a hybrid genetic particle swarm optimization algorithm to iteratively solve the multi-objective optimization model constructed in step S3. S5. Output the optimal delivery plan: Based on the algorithm's solution, output the optimal cold chain delivery plan, which includes the optimal delivery route, vehicle allocation plan, and delivery time sequence arrangement.

[0006] The above technical solution breaks through the limitations of traditional single-cost-objective optimization by constructing a dual-objective optimization model that minimizes delivery costs and maximizes customer satisfaction. By quantifying the satisfaction index of delivery time window and freshness of goods, it achieves synergistic optimization of cold chain delivery cost control and service quality improvement. It is highly versatile and adaptable to the delivery needs of all categories and multiple scenarios of cold chain logistics, and is applicable to various cold chain delivery scenarios such as chain convenience stores, fresh food e-commerce, pharmaceutical cold chain, and supermarket cold chain.

[0007] Furthermore, in step S1, the basic information of the distribution center includes the geographical location of the distribution center, the temperature control standards of the temperature zones, the organizational structure, and the work process; The operational data of the delivery nodes includes the geographical location of the delivery nodes, order volume, delivery time window, and loading and unloading operation time. The cold chain commodity data includes temperature control requirements for commodity temperature zones, shelf life, and packaging parameters; the urban time-varying road network traffic data includes real-time driving speed, road segment distance, and congestion time segment divisions for different time periods. The parameters of the refrigerated delivery vehicles include the vehicle's rated load capacity, refrigeration power, fuel consumption parameters, and fixed operating cost parameters.

[0008] The above technical solutions are compatible with the full range of cold chain logistics delivery needs across multiple scenarios. They refine temperature control constraints, freshness decay models, and cost structures for different temperature zones of goods, accurately matching vehicle models and order requirements, improving vehicle load utilization, and are widely applicable to various cold chain delivery scenarios such as chain convenience stores, fresh food e-commerce, pharmaceutical cold chain, and supermarket cold chain.

[0009] Furthermore, in step S1, the preprocessing of urban time-varying road network traffic data specifically involves: dividing a single day of 24 hours into several time periods according to a preset time interval, matching the average vehicle speed of the corresponding road segment to each time period, constructing a segmented time-varying speed function, and calculating the vehicle road segment travel time based on this function to replace the traditional static speed model.

[0010] Furthermore, in the step S2, the delivery time satisfaction function is constructed by using a trapezoidal mixed time window linear function, specifically as follows: Set the acceptable delivery time window of the delivery node as [ET', LT'], and the expected delivery time window as [ET, LT], where ET' is the earliest acceptable delivery time, ET is the earliest expected delivery time, LT is the latest expected delivery time, and LT' is the latest acceptable delivery time; When the actual arrival time t of the vehicle belongs to [ET, LT], the time satisfaction is 100%; When ET' < t < ET, the time satisfaction decreases linearly with the advance duration; When LT < t < LT', the time satisfaction decreases linearly with the delay duration; When t < ET' or t > LT', the time satisfaction is 0.

[0011] Furthermore, in the step S2, the goods freshness satisfaction function is constructed by using a continuous exponential decay function, which respectively quantifies the temperature control decay in the transportation link and the normal temperature exposure decay in the loading and unloading link. Based on the initial freshness of the commodity, the freshness decay degree of the entire delivery process is characterized by an exponential function; The specific customer satisfaction comprehensive evaluation model is: S = ω1×S1 + ω2×S2, where S is the comprehensive satisfaction, S1 is the delivery time satisfaction, S2 is the goods freshness satisfaction, ω1 and ω2 are the weight coefficients of the corresponding indicators respectively, and ω1 + ω2 = 1.

[0012] Furthermore, in the step S3, the total delivery cost includes vehicle fixed cost, fuel cost and refrigeration cost, where: The vehicle fixed cost includes the single - delivery sharing costs of vehicle depreciation, insurance annual inspection, and driver salary, which is positively correlated with the number of delivery trips; The fuel cost is calculated based on the vehicle driving speed, vehicle real - time load, and road section distance under the time - varying road network, and is positively correlated with the vehicle start - stop frequency and load level; The refrigeration cost includes the continuous refrigeration energy consumption cost in the transportation link and the supplementary refrigeration energy consumption cost for temperature fluctuation in the loading and unloading link, and is calculated based on the refrigeration equipment power, ambient temperature, carriage temperature control standard, and transportation duration.

[0013] Furthermore, in the step S3, the constraint conditions include vehicle load constraint, time window constraint, flow conservation constraint, temperature control constraint, and vehicle number constraint, where: Vehicle load constraint: The total weight of the goods loaded on a single delivery vehicle does not exceed its rated load; Time window constraint: The delivery arrival time of a single delivery node must meet the preset acceptable delivery time window requirements of that node; Traffic conservation constraint: A single delivery vehicle departs from the distribution center, completes its delivery task, and returns to the distribution center, and each delivery node is served by only one vehicle once; Temperature control constraints: The temperature of the vehicle compartment must remain stable within the preset temperature control range for the corresponding product category throughout the entire transportation process of cold chain goods, and the temperature fluctuation range shall not exceed the preset threshold. Vehicle quantity constraint: The total number of delivery vehicles in use shall not exceed the maximum number of vehicles that can be allocated to the distribution center.

[0014] Furthermore, in step S4, the solution process of the hybrid genetic particle swarm optimization algorithm is as follows: S41. Encoding Processing: Real number encoding is adopted. The integer part of the encoding corresponds to the delivery vehicle type assigned to the delivery node, and the fractional part corresponds to the delivery ranking weight of the delivery nodes of the same vehicle type, thus completing particle encoding. S42. Initialize the population: Randomly generate an initial particle swarm of a preset size, and set the initial parameters for the maximum number of algorithm iterations, crossover probability, mutation probability, particle swarm learning factor, and inertia weight. S43. Fitness Calculation: Construct a fitness function based on the dual objective function, calculate the fitness value of each particle in the population, and record the global optimal solution and the individual optimal solution; S44. Particle Swarm Velocity and Position Update: Based on the particle swarm algorithm rules, the velocity and position of each particle are updated using adaptive inertia weights to complete the global search optimization of the swarm. S45. Genetic operations: Perform selection, crossover, and mutation operations on the updated population to enhance the population's local optimization ability and diversity; S46. Optimal Solution Update: Compare the updated fitness values ​​of the particles and simultaneously update the individual optimal solution and the global optimal solution. S47. Termination condition judgment: Determine whether the current iteration count has reached the preset maximum iteration count. If it has, terminate the iteration and output the global optimal solution; otherwise, return to step S43 to continue iterating.

[0015] Furthermore, in step S4, the hybrid genetic particle swarm algorithm introduces an adaptive parameter adjustment mechanism, specifically: the inertia weight is adaptively adjusted with the number of iterations and particle fitness; a larger inertia weight is maintained in the early stage of iteration to enhance global search; and the inertia weight is reduced in the later stage of iteration to enhance local optimization. The crossover and mutation probabilities are dynamically adjusted according to the fitness distribution of the population. When the population tends to converge, the mutation probability is increased to avoid premature convergence of the algorithm and getting stuck in local optima.

[0016] Furthermore, in step S5, the output optimal delivery plan also includes single vehicle delivery node sorting, single node estimated loading and unloading time, road segment travel time estimation, single vehicle delivery cost accounting, single node satisfaction estimation, vehicle load utilization rate data, and simultaneously outputs a delivery priority plan that matches the hierarchical management of delivery nodes.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This cold chain logistics distribution route optimization method constructs a dual-objective optimization model that minimizes distribution costs and maximizes customer satisfaction. It breaks through the limitations of traditional single-cost objective optimization. By quantifying the satisfaction index of delivery time window and freshness of goods, it achieves synergistic optimization of cold chain distribution cost control and service quality improvement. It is highly versatile and adaptable to the distribution needs of all categories and multiple scenarios of cold chain logistics. It is applicable to various cold chain distribution scenarios such as chain convenience stores, fresh food e-commerce, pharmaceutical cold chain, and supermarket cold chain. 2. This cold chain logistics delivery route optimization method constructs a segmented time-varying speed function based on traffic big data, replacing the traditional static speed model. It accurately matches the speed changes of vehicles under urban congestion conditions, making the route planning highly consistent with the actual operation scenario of cold chain delivery, greatly improving the feasibility of the solution and effectively reducing the risk of delivery delays. 3. This cold chain logistics distribution route optimization method refines temperature control constraints, freshness decay models, and cost structures for the characteristics of goods in different temperature zones. It can accurately match vehicle type and order requirements, improve vehicle load utilization, and is widely applicable to various cold chain distribution scenarios such as chain convenience stores, fresh food e-commerce, pharmaceutical cold chain, and supermarket cold chain. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart illustrating the overall process of the cold chain logistics distribution route optimization method of the present invention. Figure 2 This is a flowchart illustrating the solution process of the hybrid genetic particle swarm optimization algorithm of the present invention. Figure 3 This is a flowchart of the standard cold chain logistics distribution process of the present invention; Figure 4 This is a graph showing the percentage of factors influencing customer satisfaction with cold chain delivery, as presented in this invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] like Figure 1 - Figure 4 As shown, the cold chain logistics distribution route optimization method provided in this embodiment includes the following steps: S1. Cold chain distribution basic data collection and preprocessing: Collect basic information of distribution centers, operational data of distribution nodes, cold chain commodity data, urban time-varying road network traffic data, and parameter data of refrigerated delivery vehicles, and perform standardized preprocessing on the collected data; The basic information of the distribution center includes its geographical location, temperature control standards for different temperature zones, organizational structure, and operational procedures. The operational data of the delivery nodes includes the geographical location of the delivery nodes, order volume, delivery time window, and loading and unloading operation time. The cold chain commodity data includes temperature control requirements for commodity temperature zones, shelf life, and packaging parameters; the urban time-varying road network traffic data includes real-time driving speed, road segment distance, and congestion time segment divisions for different time periods. The refrigerated delivery vehicle parameter data includes the vehicle's rated load capacity, refrigeration power, fuel consumption parameters, and fixed operating cost parameters; The preprocessing of urban time-varying road network traffic data is as follows: a single day of 24 hours is divided into several time periods according to a preset time interval, the average vehicle speed of the corresponding road segment is matched for each time period, a piecewise time-varying speed function is constructed, and the vehicle travel time on the road segment is calculated based on the function, replacing the traditional static speed model.

[0021] S2. Construct a comprehensive customer satisfaction evaluation model: Select delivery time window matching degree and goods freshness as core evaluation indicators, construct delivery time satisfaction function and goods freshness satisfaction function respectively, and obtain a comprehensive customer satisfaction evaluation model by weighted fusion; The delivery time satisfaction function is constructed using a trapezoidal mixed time window linear function, specifically as follows: Set the acceptable delivery time window for the delivery node as [ET', LT'], and the expected delivery time window as [ET, LT], where ET' is the earliest acceptable delivery time, ET is the earliest expected delivery time, LT is the latest expected delivery time, and LT' is the latest acceptable delivery time. When the actual arrival time of the vehicle is t∈[ET, LT], the time satisfaction rate is 100%. When ET' < t < ET, the time satisfaction decreases linearly with the advance duration; When LT < t < LT', the time satisfaction decreases linearly with the delay duration; When t < ET' or t > LT', the time satisfaction is 0; The freshness satisfaction function of the goods is constructed by a continuous exponential decay function, which quantifies the temperature control decay in the transportation link and the ambient temperature exposure decay in the loading and unloading link respectively. Based on the initial freshness of the commodity, the freshness decay degree of the whole distribution process is characterized by an exponential function; The specific customer satisfaction comprehensive evaluation model is: S = ω1×S1 + ω2×S2, where S is the comprehensive satisfaction, S1 is the distribution time satisfaction, S2 is the goods freshness satisfaction, ω1 and ω2 are the weight coefficients of the corresponding indicators respectively, and ω1 + ω2 = 1.

[0022] S3. Construct a multi-objective cold chain distribution path optimization model: Taking the minimization of the total distribution cost and the maximization of customer satisfaction as the dual optimization objectives, combining the basic data in step S1 and the customer satisfaction comprehensive evaluation model in step S2, setting constraint conditions, and constructing a multi-objective optimization mathematical model; The total distribution cost includes vehicle fixed cost, fuel cost and refrigeration cost, where: The vehicle fixed cost includes the single-distribution sharing costs of vehicle depreciation, insurance annual inspection and driver salary, which is positively correlated with the number of distribution trips; The fuel cost is calculated based on the vehicle driving speed, vehicle real-time load and section distance under the time-varying road network, and is positively correlated with the vehicle start-stop frequency and load level; The refrigeration cost includes the continuous refrigeration energy consumption cost in the transportation link and the supplementary refrigeration energy consumption cost for temperature fluctuation in the loading and unloading link, and is calculated based on the refrigeration equipment power, ambient temperature, carriage temperature control standard and transportation duration; The constraint conditions include vehicle load constraint, time window constraint, flow conservation constraint, temperature control constraint and vehicle number constraint, where: Vehicle load constraint: The total weight of the loaded goods of a single distribution vehicle does not exceed its rated load; Time window constraint: The distribution arrival time of a single distribution node needs to meet the requirements of the preset acceptable distribution time window of this node; Flow conservation constraint: A single distribution vehicle departs from the distribution center, completes the distribution task and returns to the distribution center, and each distribution node is served by only one vehicle once; Temperature control constraint: The temperature in the carriage during the whole process of cold chain commodity transportation needs to be stable within the preset temperature control range corresponding to the category, and the temperature fluctuation range does not exceed the preset threshold; Vehicle number constraint: The total number of distribution vehicles enabled does not exceed the maximum number of vehicles that can be allocated by the distribution center.

[0023] S4. Solve the model using a hybrid genetic particle swarm optimization algorithm: Combine the local optimization capability of the genetic algorithm with the global search capability of the particle swarm optimization algorithm to design a hybrid genetic particle swarm optimization algorithm to iteratively solve the multi-objective optimization model constructed in step S3. S5. Output the optimal delivery plan: Based on the algorithm's solution, output the optimal cold chain delivery plan, which includes the optimal delivery route, vehicle allocation plan, and delivery time sequence arrangement. The optimal delivery plan output also includes single-vehicle delivery node ranking, single-node estimated loading and unloading time, road segment travel time estimation, single-vehicle delivery cost calculation, single-node satisfaction estimation, and vehicle load utilization data, and simultaneously outputs a delivery priority plan that matches the hierarchical management of delivery nodes.

[0024] In this embodiment, the solution process of the hybrid genetic particle swarm optimization algorithm in step S4 is as follows: S41. Encoding Processing: Real number encoding is adopted. The integer part of the encoding corresponds to the delivery vehicle type assigned to the delivery node, and the fractional part corresponds to the delivery ranking weight of the delivery nodes of the same vehicle type, thus completing particle encoding. S42. Initialize the population: Randomly generate an initial particle swarm of a preset size, and set the initial parameters for the maximum number of algorithm iterations, crossover probability, mutation probability, particle swarm learning factor, and inertia weight. S43. Fitness Calculation: Construct a fitness function based on the dual objective function, calculate the fitness value of each particle in the population, and record the global optimal solution and the individual optimal solution; S44. Particle Swarm Velocity and Position Update: Based on the particle swarm algorithm rules, the velocity and position of each particle are updated using adaptive inertia weights to complete the global search optimization of the swarm. S45. Genetic operations: Perform selection, crossover, and mutation operations on the updated population to enhance the population's local optimization ability and diversity; S46. Optimal Solution Update: Compare the updated fitness values ​​of the particles and simultaneously update the individual optimal solution and the global optimal solution. S47. Termination Condition Judgment: Determine whether the current iteration count has reached the preset maximum iteration count. If it has, terminate the iteration and output the global optimal solution; otherwise, return to step S43 to continue iterating. The hybrid genetic particle swarm algorithm introduces an adaptive parameter adjustment mechanism, specifically: the inertia weight is adaptively adjusted with the number of iterations and particle fitness; a larger inertia weight is maintained in the early stage of iteration to enhance global search, and the inertia weight is reduced in the later stage of iteration to enhance local optimization. The crossover and mutation probabilities are dynamically adjusted according to the fitness distribution of the population. When the population tends to converge, the mutation probability is increased to avoid premature convergence of the algorithm and getting stuck in local optima.

[0025] In this embodiment, the application scenario is as follows: T Chain Convenience Stores entrusts three professional third-party logistics distribution centers to carry out distribution operations in the Beijing area. Among them, the cold chain distribution center is located in Mafang Town, Pinggu District, Beijing. The temperature control standard is 0-4℃, and a zero-inventory operation mode is adopted. It is responsible for the cold chain fresh food distribution tasks of 91 stores (distribution nodes) in the Beijing area. It is equipped with three models of refrigerated delivery vehicles of 1.5 tons, 2 tons, and 2.5 tons. The cold chain fresh food is delivered three times a day. The first delivery from 3:00 am to 8:30 am completes nearly 90% of the store's replenishment task. This embodiment optimizes the route of the delivery process during this period. The specific steps are as follows: S1 Cold Chain Distribution Basic Data Collection and Preprocessing The following basic data were collected through the T-Chain Convenience Store Operation System, Baidu Maps Transportation Big Data Platform, and Distribution Center Management System: (1) Distribution center data: geographical coordinates of the distribution center, temperature control standard of 0-4℃, total number of refrigerated vehicles available for deployment, vehicle parameters, and operational procedures of the departments; (2) Delivery node data: Geographic coordinates of 91 stores, order volume of refrigerated fresh food per store, acceptable delivery time window [2:00, 9:30], expected delivery time window [3:00, 8:30], average loading and unloading time per store 15-20 minutes; (3) Cold chain commodity data: refrigerated fresh food (bento boxes, rice balls, sandwiches, salads, etc.), temperature control requirements 0-4℃, shelf life, packaging weight parameters, temperature fluctuation threshold ≤1℃; (4) Time-varying road network data: Real-time traffic data of Beijing on November 10, 2025 was collected. The 24 hours of the day were divided into 288 time periods with 5-minute intervals. The average vehicle speed of the corresponding road segment was matched for each time period, and a segmented time-varying speed function was constructed. The average vehicle speed during the morning peak period was 15-20 km / h. (5) Vehicle parameters: Rated load capacity, refrigeration power, fuel consumption per 100 kilometers, and fixed cost allocation parameters per vehicle for 1.5-ton, 2-ton, and 2.5-ton refrigerated trucks.

[0026] The collected data is preprocessed by deduplication, outlier removal, coordinate standardization, and unit unification to form the basic database for model solving.

[0027] S2. Construct a comprehensive customer satisfaction evaluation model Delivery time window matching degree and goods freshness are selected as core evaluation indicators. Based on the operational needs of this scenario, the weighting coefficients are set to ω1=0.5 and ω2=0.5 respectively, and a comprehensive evaluation model is constructed: (1) Construction of delivery time satisfaction function: Set the acceptable delivery time window for the store as [ET' = 2:00, LT' = 9:30], and the expected delivery time window as [ET = 3:00, LT = 8:30]. When the vehicle arrival time t is between 3:00 and 8:30, the time satisfaction S1 = 100%; when 2:00 < t < 3:00, the satisfaction linearly decreases by 1 minute in advance, and S1 = 0 when it is more than 60 minutes in advance; when 8:30 < t < 9:30, the satisfaction linearly decreases by 1 minute in delay, and S1 = 0 when it is more than 60 minutes in delay; when t < 2:00 or t > 9:30, S1 = 0; (2) Construction of goods freshness satisfaction function: Based on the temperature-sensitive characteristics of refrigerated fresh food, construct a continuous exponential decay function. When the temperature in the carriage is stable at 0 - 4°C during transportation, the freshness decays exponentially with the transportation duration; when exposed to normal temperature during loading and unloading, the decay rate of freshness increases by 3 times. By combining the decay amounts in the transportation and loading / unloading links, the goods freshness satisfaction S2 is obtained; (3) Comprehensive satisfaction model: S = 0.5×S1 + 0.5×S2, and the construction of the comprehensive customer satisfaction evaluation model is completed.

[0028] S3. Construct a multi-objective cold chain distribution path optimization model (1) Determine the double-objective function: Objective function 1: minC = Cfix + Cfuel + Ccool Among them, Cfix is the vehicle fixed cost, the single-trip fixed cost is 150 - 250 yuan, which is positively correlated with the number of vehicles in use; Cfuel is the fuel cost, calculated based on the time-varying speed, vehicle real-time load, and road section distance; Ccool is the refrigeration cost, calculated based on the transportation duration, ambient temperature, and refrigeration power; Objective function 2: maxS, where S is the customer comprehensive satisfaction constructed in step S2; (2) Set the constraint conditions: Vehicle load constraint: The total weight of goods loaded on a single vehicle ≤ the rated load of the corresponding vehicle type; Time window constraint: The delivery arrival time of all stores needs to meet the requirements of the acceptable delivery time window of 2:00 - 9:30; Flow conservation constraint: All delivery vehicles start from the Pinggu District cold chain distribution center, return to the distribution center after completing the delivery, and each store is served by only one vehicle once; Temperature control constraint: The temperature in the carriage is stable at 0 - 4°C throughout the transportation, and the temperature fluctuation ≤ 1°C; Vehicle quantity constraint: The total number of vehicles in use ≤ the maximum number of refrigerated trucks that can be allocated by the distribution center.

[0029] S4. Use the hybrid genetic particle swarm optimization algorithm to solve the model The algorithm is programmed using Python software, and the solution process is as follows: (1) Encoding process: Real number encoding is used. The integer part of the encoding corresponds to the delivery vehicle type assigned to the store (1 represents 1.5 tons, 2 represents 2 tons, and 3 represents 2.5 tons), and the decimal part corresponds to the delivery ranking weight of the store with the same vehicle type, thus completing particle encoding; (2) Initialization parameters: Population size is set to 200, maximum number of iterations is set to 500, initial crossover probability is 0.8, initial mutation probability is 0.05, particle swarm learning factor c1=c2=2, initial inertia weight is 0.7; (3) Fitness calculation: Based on the dual objective function, a fitness function is constructed. The principle is that the lower the cost and the higher the satisfaction, the higher the fitness value. The fitness of each particle is calculated, and the global optimal solution and the individual optimal solution are recorded. (4) Particle swarm update: Based on the particle swarm velocity and position update formula, combined with adaptive inertial weights, update the velocity and position of each particle to complete the global search; (5) Genetic operations: Perform roulette wheel selection, sequential crossover, and exchange mutation operations on the updated population to enhance local optimization ability and population diversity; (6) 2-opt local search: Randomly select two nodes in the path, swap the node positions and recalculate the path length and fitness to retain the better path and avoid getting trapped in local optima; (7) Adaptive parameter adjustment: Based on the number of iterations and the fitness distribution of the population, the inertia weight, crossover probability and mutation probability are dynamically adjusted. In the early stage of iteration, the inertia weight is maintained to strengthen the global search, and in the later stage of iteration, the inertia weight is reduced to strengthen the local optimization. When the population tends to converge, the mutation probability is increased to avoid premature convergence. (8) Optimal solution update and termination judgment: Compare the fitness values ​​of the updated particles, update the individual optimal solution and the global optimal solution simultaneously. When the number of iterations reaches 500, terminate the iteration and output the global optimal solution.

[0030] Meanwhile, the same model was solved using basic genetic algorithms and particle swarm optimization algorithms, and compared and verified with the hybrid algorithm of this invention.

[0031] S5, Output the optimal delivery plan Based on the algorithm's solution, the final optimized solution is output, including: optimal delivery route planning, number of vehicles of each type, order of delivery stores per vehicle, delivery time arrangement per store, total cost calculation, estimated overall satisfaction, estimated loading and unloading time per store, estimated travel time per road segment, and vehicle load utilization rate data.

[0032] In the description of this invention, it should be understood that the terms "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. In the description of this invention, unless otherwise specified and limited, it should be noted that the terms "installed," "connected," and "linked" should be interpreted broadly. For example, they can refer to mechanical or electrical connections, or internal connections between two elements; they can be direct connections or indirect connections through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms according to the specific circumstances.

[0033] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for optimizing cold chain logistics distribution routes, characterized in that... , including the following steps: S1. Collection and preprocessing of basic cold-chain distribution data: Collect the basic information of the distribution center, the operation data of distribution nodes, the cold-chain commodity data, the time-varying road network traffic data of the city, and the parameter data of refrigerated distribution vehicles, and perform standardized preprocessing on the collected data; S2. Construction of a comprehensive customer satisfaction evaluation model: Select the matching degree of the delivery time window and the freshness of goods as the core evaluation indicators, construct the delivery time satisfaction function and the freshness satisfaction function of goods respectively, and obtain the comprehensive customer satisfaction evaluation model through weighted fusion; S3. Construction of a multi-objective cold-chain distribution route optimization model: Take the minimization of the total distribution cost and the maximization of customer satisfaction as the dual optimization objectives, combine the basic data in step S1 and the comprehensive customer satisfaction evaluation model in step S2, set the constraint conditions, and construct a multi-objective optimization mathematical model; S4. Solve the model using a hybrid genetic particle swarm optimization algorithm: Integrate the local optimization ability of the genetic algorithm and the global search ability of the particle swarm optimization algorithm, design a hybrid genetic particle swarm optimization algorithm, and perform iterative solution on the multi-objective optimization model constructed in step S3; S5. Output the optimal distribution plan: According to the algorithm solution result, output a cold-chain distribution optimization plan including the optimal distribution route, vehicle allocation plan, and distribution time sequence arrangement.

2. The cold chain logistics distribution route optimization method according to claim 1, characterized in that: In the step S1, the basic information of the distribution center includes the geographical location of the distribution center, the temperature control standard of the temperature-controlled area, the organizational structure, and the operation process; The operation data of the distribution node includes the geographical location of the distribution node, the order quantity, the delivery time window, and the loading and unloading operation duration; The cold-chain commodity data includes the temperature control requirements of the commodity in different temperature-controlled areas, the freshness period, and the packaging parameters; the time-varying road network traffic data of the city includes the real-time driving speed, road section distance, and congestion period division of different road sections at different times; The parameter data of the refrigerated distribution vehicle includes the rated load capacity of the vehicle, the refrigeration power, the fuel consumption parameter, and the fixed operation cost parameter.

3. The cold chain logistics distribution route optimization method according to claim 1, characterized in that: In the step S1, the preprocessing of the time-varying road network traffic data of the city is specifically as follows: Divide the 24 hours of a single day into several time periods at a preset time interval, match the average driving speed of the corresponding road section for each time period, construct a piecewise time-varying speed function, and calculate the driving duration of the vehicle on the road section based on this function to replace the traditional static speed model.

4. The cold chain logistics distribution route optimization method according to claim 1, characterized in that: In the step S2, the delivery time satisfaction function is constructed using a trapezoidal hybrid time window linear function, specifically as follows: Set the acceptable delivery time window of the distribution node as [ET', LT'], and the expected delivery time window as [ET, LT], where ET' is the earliest acceptable delivery time, ET is the expected earliest delivery time, LT is the expected latest delivery time, and LT' is the latest acceptable delivery time; When the actual arrival time t of the vehicle belongs to [ET, LT], the time satisfaction is 100%; When ET' < t < ET, the time satisfaction decreases linearly with the advance duration; When LT < t < LT', the time satisfaction decreases linearly with the delay duration; When t < ET' or t > LT', the time satisfaction is 0.

5. The cold chain logistics distribution route optimization method according to any one of claims 1 or 4, characterized in that: In step S2, the goods freshness satisfaction function is constructed using a continuous exponential decay function, which quantifies the temperature control decay during transportation and the ambient temperature exposure decay during loading and unloading. Based on the initial freshness of the goods, the exponential function characterizes the degree of freshness decay throughout the entire delivery process. The comprehensive customer satisfaction evaluation model is as follows: S = ω1 × S1 + ω2 × S2, where S is the overall satisfaction, S1 is the satisfaction with delivery time, S2 is the satisfaction with the freshness of goods, ω1 and ω2 are the weight coefficients of the corresponding indicators, and ω1 + ω2 = 1.

6. The cold chain logistics distribution route optimization method according to claim 1, characterized in that: In step S3, the total delivery cost includes vehicle fixed costs, fuel costs, and refrigeration costs, wherein: Vehicle fixed costs include vehicle depreciation, insurance and annual inspection, and the cost allocated to drivers' salaries per delivery, which are positively correlated with the number of delivery trips. Fuel costs are calculated based on vehicle speed, real-time vehicle load, and road distance under time-varying road networks, and are positively correlated with vehicle start-stop frequency and load level. Refrigeration costs include the energy consumption costs of continuous refrigeration during transportation and the energy consumption costs of supplemental refrigeration due to temperature fluctuations during loading and unloading. These costs are calculated based on the power of the refrigeration equipment, ambient temperature, temperature control standards for the vehicle compartment, and transportation time.

7. The cold chain logistics distribution route optimization method according to claim 1, characterized in that: In step S3, the constraints include vehicle load constraints, time window constraints, flow conservation constraints, temperature control constraints, and vehicle quantity constraints, wherein: Vehicle load limit: The total weight of goods loaded on a single delivery vehicle shall not exceed its rated load capacity. Time window constraint: The delivery arrival time of a single delivery node must meet the preset acceptable delivery time window requirements of that node; Traffic conservation constraint: A single delivery vehicle departs from the distribution center, completes its delivery task, and returns to the distribution center, and each delivery node is served by only one vehicle once; Temperature control constraints: The temperature of the vehicle compartment must remain stable within the preset temperature control range for the corresponding product category throughout the entire transportation process of cold chain goods, and the temperature fluctuation range shall not exceed the preset threshold. Vehicle quantity constraint: The total number of delivery vehicles in use shall not exceed the maximum number of vehicles that can be allocated to the distribution center.

8. The cold chain logistics distribution route optimization method according to claim 1, characterized in that: In step S4, the solution process of the hybrid genetic particle swarm optimization algorithm is as follows: S41. Encoding Processing: Real number encoding is adopted. The integer part of the encoding corresponds to the delivery vehicle type assigned to the delivery node, and the fractional part corresponds to the delivery ranking weight of the delivery nodes of the same vehicle type, thus completing particle encoding. S42. Initialize the population: Randomly generate an initial particle swarm of a preset size, and set the initial parameters for the maximum number of algorithm iterations, crossover probability, mutation probability, particle swarm learning factor, and inertia weight. S43. Fitness Calculation: Construct a fitness function based on the dual objective function, calculate the fitness value of each particle in the population, and record the global optimal solution and the individual optimal solution; S44. Particle Swarm Velocity and Position Update: Based on the particle swarm algorithm rules, the velocity and position of each particle are updated using adaptive inertia weights to complete the global search optimization of the swarm. S45. Genetic operations: Perform selection, crossover, and mutation operations on the updated population to enhance the population's local optimization ability and diversity; S46. Optimal Solution Update: Compare the updated fitness values ​​of the particles and simultaneously update the individual optimal solution and the global optimal solution. S47. Termination condition judgment: Determine whether the current iteration count has reached the preset maximum iteration count. If it has, terminate the iteration and output the global optimal solution; otherwise, return to step S43 to continue iterating.

9. The cold chain logistics distribution route optimization method according to claim 8, characterized in that: In step S4, the hybrid genetic particle swarm algorithm introduces an adaptive parameter adjustment mechanism, specifically: the inertia weight is adaptively adjusted with the number of iterations and particle fitness; a larger inertia weight is maintained in the early stage of iteration to enhance global search; and the inertia weight is reduced in the later stage of iteration to enhance local optimization. The crossover and mutation probabilities are dynamically adjusted according to the fitness distribution of the population. When the population tends to converge, the mutation probability is increased to avoid premature convergence of the algorithm and getting stuck in local optima.

10. The cold chain logistics distribution route optimization method according to claim 1, characterized in that: In step S5, the output optimal delivery plan also includes single vehicle delivery node sorting, single node estimated loading and unloading time, road segment travel time estimation, single vehicle delivery cost accounting, single node satisfaction estimation, vehicle load utilization rate data, and simultaneously outputs a delivery priority plan that matches the hierarchical management of delivery nodes.

Images