A cold chain transportation method and system based on swarm intelligence

Abstract

The application provides a cold chain transportation method and system based on swarm intelligence, and relates to the field of swarm intelligence. The method comprises the following steps: obtaining a plurality of cold chain transportation orders in a target time period, and determining a minimum number of mergeable orders; determining a number range of target transportation order merging based on a similar historical time period transportation order merging number and the minimum number of mergeable orders; generating a plurality of target merged transportation orders according to the number range of target transportation order merging through a wolf swarm algorithm improved based on a particle swarm algorithm; and determining an optimal transportation path corresponding to the target merged transportation order based on the cold chain transportation orders included in the target merged transportation order for each target merged transportation order. The application has the advantage of improving the efficiency of cold chain transportation.

Description

Technical Field

[0001] This invention relates to the field of swarm intelligence, and in particular to a cold chain transportation method and system based on swarm intelligence. Background Technology

[0002] In the field of cold chain logistics, the cold chain transportation management of mobile cold storage facilities is crucial for ensuring the quality and safety of temperature-sensitive goods such as fresh food and medicine.

[0003] Currently, traditional cold chain transportation management methods face numerous challenges. In order consolidation, traditional methods often rely on manual experience for simple order combining, lacking scientific evaluation and decision-making mechanisms. On the one hand, it's difficult to fully consider the characteristics of different orders' goods, such as temperature requirements, leading to quality risks due to incompatibility during consolidation. For example, mixing fresh produce requiring refrigeration with room-temperature medicines can easily cause spoilage or drug inactivation. On the other hand, the limitations of transport vehicle load capacity and volume are not fully considered, often resulting in unreasonable vehicle loading. This either leads to significant space waste and increased transportation costs, or overloading, violating traffic regulations and affecting driving safety.

[0004] In terms of route planning, existing technologies mostly use fixed algorithms or simple planning based on historical data, lacking collaborative optimization of transportation tasks after multiple orders are merged. It is difficult to comprehensively consider factors such as delivery time and location of each order, and the planned route may not meet the timeliness requirements of all orders, affecting customer satisfaction.

[0005] Although some studies have attempted to solve the above problems using algorithms, they suffer from drawbacks such as high computational complexity, slow convergence speed, and susceptibility to getting trapped in local optima, making them difficult to apply efficiently in the complex and ever-changing cold chain transportation scenarios.

[0006] Therefore, there is a need to provide a cold chain transportation method and system based on swarm intelligence to improve the efficiency of cold chain transportation. Summary of the Invention

[0007] This invention provides a cold chain transportation method based on swarm intelligence, comprising: acquiring multiple cold chain transportation orders within a target time period and determining the minimum number of orders that can be merged; determining the range of target transportation order merging quantities based on the number of transportation orders merged in similar historical time periods and the minimum number of orders that can be merged; generating multiple target merged transportation orders according to the target range of merged transportation orders using a wolf pack algorithm based on an improved particle swarm algorithm; and for each target merged transportation order, determining the optimal transportation route corresponding to the target merged transportation order based on the cold chain transportation orders included in the target merged transportation order.

[0008] Furthermore, determining the minimum number of orders that can be merged includes: determining the minimum number of orders that can be merged based on the weight and volume of goods in multiple cold chain transportation orders; determining the range of target transportation order merging quantities based on the number of transportation orders merged in similar historical time periods and the minimum number of orders that can be merged, including: determining similar historical time periods based on the weight, volume, refrigeration environment requirements, and destinations of goods in multiple cold chain transportation orders and multiple historical time periods; determining the initial range of target transportation order merging quantities based on the number of transportation orders merged in similar historical time periods; and determining the target transportation order merging quantity range based on the initial range of target transportation order merging quantities and the minimum number of orders that can be merged.

[0009] Furthermore, using a wolf pack algorithm improved from particle swarm optimization, multiple target merged transportation orders are generated based on the range of target transportation order merging quantities. This includes: calculating the merging index of any two cold chain transportation orders; initializing the wolf pack based on the merging index of the two cold chain transportation orders, the range of target transportation order merging quantities, and the constraint set; constructing a first fitness function; and iteratively optimizing the wolf pack based on the first fitness function until a first termination condition is met, generating multiple target merged transportation orders. In each iteration, the scout wolves in the wolf pack are identified using particle swarm optimization.

[0010] Further, the merging index of the two cold chain transportation orders is calculated, including: S11, determining whether the two cold chain transportation orders can be merged based on the weight and volume of the goods in the two orders; if not, the merging index of the two cold chain transportation orders is 0; if yes, proceed to S12; S12, calculating the similarity of the refrigeration requirements of the two cold chain transportation orders based on their refrigeration environment requirements; S13, determining whether the two cold chain transportation orders can be merged based on the similarity of their refrigeration requirements; if not, the merging index of the two cold chain transportation orders is 0; if yes, proceed to S14; S14, based on the similarity of the refrigeration requirements of the two cold chain transportation orders... For each order, determine the shortest transportation route between the two cold chain transportation orders based on their destination and the required delivery time. Calculate the maximum time difference between the two cold chain transportation orders based on their required delivery time. Calculate the resource coordination ratio between the two cold chain transportation orders based on their shortest transportation route and maximum time difference. S15: Determine whether the two cold chain transportation orders can be merged based on their resource coordination ratio. If not, the merging index of the two cold chain transportation orders is 0; if yes, proceed to S16. S16: Calculate the merging index of the two cold chain transportation orders based on the similarity of their refrigeration requirements and their resource coordination ratio.

[0011] Furthermore, based on the merging index of two cold chain transportation orders, the range of the target transportation order merging quantity, and the constraint set, the wolf pack is initialized, including: for each cold chain transportation order, determining the sampling probability of the cold chain transportation order based on the merging index of any two cold chain transportation orders; when generating each wolf, sampling the transportation order merging quantity corresponding to the wolf from the range of the target transportation order merging quantity; based on the sampling probability of each cold chain transportation order and the transportation order merging quantity corresponding to the wolf, sampling cold chain transportation orders from multiple centers from multiple cold chain transportation orders; and deduplicating the wolves based on the cold chain transportation orders from multiple centers, the merging index of any two cold chain transportation orders, and the constraint set, and with the already generated wolves.

[0012] Furthermore, the scout wolves in the wolf pack are determined using the particle swarm optimization algorithm, including: constructing a second fitness function; determining multiple candidate wolves from the wolf pack based on the first fitness value of each wolf in the pack; initializing the population based on the multiple candidate wolves; and iteratively optimizing the population according to the second fitness function until a second termination condition is met to determine the scout wolves in the wolf pack. The second fitness function is related to the merged similarity of any two scout wolves and the global merged similarity of each scout wolf.

[0013] Furthermore, the first fitness function is correlated with the mean of the merging index for each target merged transportation order.

[0014] Furthermore, based on the cold chain transportation orders included in the target combined transportation order, the optimal transportation route corresponding to the target combined transportation order is determined, including: for each target combined transportation order, a global reachable path map corresponding to the target combined transportation order is constructed based on an electronic map, wherein the global reachable path map consists of multiple reachable road segments; and the optimal transportation route corresponding to the target combined transportation order is determined based on the global reachable path map corresponding to the target combined transportation order using a particle swarm optimization algorithm.

[0015] Furthermore, using the particle swarm optimization algorithm, the optimal transportation path for the target merged transportation order is determined based on the global reachability path graph corresponding to the target merged transportation order. This includes: for each reachable segment, calculating the weight of the reachable segment based on historical transportation records; constructing a third fitness function, where the third fitness function is related to the weight of the reachable segments included in the transportation path corresponding to the particle; initializing the population based on the global reachability path graph corresponding to the target merged transportation order; and iteratively optimizing the population based on the third fitness function until the third termination condition is met, thereby determining the optimal transportation path for the target merged transportation order.

[0016] This invention provides a cold chain transportation system based on swarm intelligence, comprising: an order acquisition module for acquiring multiple cold chain transportation orders within a target time period and determining the minimum number of orders that can be merged; a quantity constraint module for determining the range of target transportation order merging quantities based on the number of transportation orders merged in similar historical time periods and the minimum number of orders that can be merged; an order merging module for generating multiple target merged transportation orders according to the target range of merged quantities using a wolf pack algorithm based on an improved particle swarm algorithm; and a route generation module for determining the optimal transportation route corresponding to each target merged transportation order based on the cold chain transportation orders included in the target merged transportation order.

[0017] Compared with existing technologies, the cold chain transportation method and system based on swarm intelligence provided by this invention has at least the following beneficial effects: 1. By comprehensively considering factors such as cargo weight, volume, refrigeration environment requirements, and destination of cold chain transportation orders, the range of possible merged transportation orders is first determined. Then, a wolf pack algorithm based on an improved particle swarm optimization algorithm is used to generate multiple target merged transportation orders. When calculating the merging index, multiple dimensions are considered to determine whether orders can be merged, ensuring that merged orders will not encounter problems such as inadequate refrigeration or time conflicts during transportation. This scientific order merging method can fully integrate transportation resources, reduce vehicle empty running rates, improve the transportation efficiency of mobile cold storage, and reduce transportation costs.

[0018] 2. For each combined transport order, a global reachability map is constructed based on an electronic map, and the optimal transport route is determined using a particle swarm optimization algorithm. When calculating the weights of reachable road segments, historical transport records are referenced to ensure the weights accurately reflect the traffic conditions of the road segments. The third fitness function is related to the weights of the road segments included in the particle transport path. Through iterative optimization of the population, the path with the highest fitness value, i.e., the optimal transport path, can be found. This ensures that combined transport orders can reach their destination in the shortest time and along the smoothest route, guaranteeing the quality of cold chain goods, reducing the risk of goods spoiling due to excessive transport time, and improving customer satisfaction.

[0019] 3. This algorithm integrates the advantages of particle swarm optimization (PSO) and wolf pack algorithms. It uses PSO to identify scout wolves within the pack, leverages the wolf pack algorithm to generate merged transport orders, and then combines PSO to determine the optimal transport route. It exhibits self-organizing, adaptive, and self-learning characteristics, enabling it to automatically adjust its search strategy in complex cold chain transport environments to find better solutions. Furthermore, this algorithm demonstrates good adaptability to cold chain transport orders of different sizes and types, flexibly adjusting parameters and strategies according to actual conditions, providing strong technical support for cold chain transport management in mobile cold storage facilities. Attached Figure Description

[0020] This specification will be further described by way of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting; in these embodiments, the same reference numerals denote the same structures, wherein: Figure 1 This is a schematic flowchart of a cold chain transportation method based on swarm intelligence, as shown in some embodiments of this specification. Figure 2 This is a flowchart illustrating the calculation of the combined index of two cold chain transport orders according to some embodiments of this specification; Figure 3 This is a schematic diagram of a cold chain transportation system based on swarm intelligence, as shown in some embodiments of this specification. Detailed Implementation

[0021] To more clearly illustrate the technical solutions of the embodiments in this specification, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are merely some examples or embodiments of this specification. For those skilled in the art, these drawings can be applied to other similar scenarios without creative effort. Unless obvious from the context or otherwise specified, the same reference numerals in the drawings represent the same structures or operations.

[0022] Figure 1 This is a schematic flowchart illustrating a cold chain transportation method based on swarm intelligence, as shown in some embodiments of this specification. Figure 1 As shown, a cold chain transportation method based on swarm intelligence may include the following steps.

[0023] Step 110: Obtain multiple cold chain transportation orders for the target time period and determine the minimum number of orders that can be merged.

[0024] Specifically, multiple cold chain transport orders require delivery times within a target time period. Cold chain transport orders may include required delivery time, cargo weight, volume, refrigeration environment requirements (e.g., humidity requirements, temperature requirements, etc.), and destination.

[0025] The minimum number of orders that can be combined can be determined based on the weight and volume of goods in multiple cold chain transportation orders.

[0026] Specifically, based on the total weight and volume of goods in multiple cold chain transportation orders, combined with vehicle load capacity (e.g., 15 tons) and volume (e.g., 80 m³) limitations, the minimum number of orders that can be combined is calculated. For example, if the total weight of multiple cold chain transportation orders is 12 tons and the total volume is 60 m³, at least one trip is required (without overloading and 75% space utilization). If the total weight of multiple cold chain transportation orders is 18 tons, two trips are required (to avoid the risk of overloading). As an example, the ratio of the total weight of multiple cold chain transportation orders to the vehicle load capacity can be calculated and rounded up to obtain the first quantity. The ratio of the total volume of multiple cold chain transportation orders to the vehicle volume can also be calculated and rounded up to obtain the second quantity. The larger of the first and second quantities is taken as the minimum number of orders that can be combined.

[0027] The minimum number of orders that can be combined refers to the number of combined transport orders necessary to meet vehicle weight and volume restrictions in cold chain transportation. Since only one vehicle is allowed per combined transport order, this minimum number of orders can also be considered the minimum number of vehicles required. This minimum number aims to ensure that the vehicle's load capacity is not exceeded while making reasonable use of its volume space during transportation. For example, assuming a vehicle has a load capacity of 15 tons and a volume of 80 m³, and three cold chain transport orders with cargo weights of 5 tons, 6 tons, and 7 tons, and volumes of 20 m³, 25 m³, and 30 m³ respectively. The total weight of the goods is 5 + 6 + 7 = 18 tons. The ratio of the total weight to the vehicle's load capacity is 18 ÷ 15 = 1.2. Rounding up, the first quantity is 2 truckloads. The total volume of the goods is 20 + 25 + 30 = 75 m³. The ratio of the total volume to the vehicle's volume is 75 ÷ 80 = 0.9375. Rounding up, the second quantity is 1 truckload. Taking the larger of the two values, the minimum number of vehicles required is 2, and therefore the minimum quantity of orders that can be combined is 2.

[0028] Step 120: Determine the range of target transportation order merging quantities based on the number of merged transportation orders in similar historical time periods and the minimum number of merging orders.

[0029] Specifically, it includes: Based on the cargo weight, volume, refrigeration environment requirements and destinations of multiple cold chain transportation orders within the target time period and multiple historical time periods, similar historical time periods are identified for the target time period. Based on the number of merged shipping orders within similar historical time periods, determine the initial range of the target shipping order merging quantity; The range of target transportation order merging quantities is determined based on the initial quantity range for target transportation order merging and the minimum quantity of orders that can be merged.

[0030] Specifically, similar historical time periods can be determined according to the following process: S21. Calculate the total weight and volume of goods for multiple cold chain transportation orders within the target time period; S22. Calculate the mean and variance of the similarity of the refrigeration environment requirements for multiple cold chain transportation orders within the target time period; for example, the mean similarity can be obtained by calculating the cosine similarity of the refrigeration environment requirements of any two cold chain transportation orders, and the variance of the cosine similarity of the refrigeration environment requirements of any two cold chain transportation orders can be obtained as the variance of the similarity. S23. Calculate the mean and variance of the distance between the destinations of any two cold chain transportation orders within the target time period; for example, the length of the shortest transportation path between the destinations of any two cold chain transportation orders can be determined as the distance between the destinations of the two cold chain transportation orders, and the mean and variance of the distance between the destinations of any two cold chain transportation orders can be calculated respectively as the mean and variance of the distance. S24. Construct a feature vector for the target time period based on the total weight, total volume, mean similarity, variance similarity, mean distance, and variance distance of multiple cold chain transportation orders within the target time period. S25. For each historical time period, calculate the cosine similarity between the feature vector of the target time period and the feature vector of the historical time period. The feature vector of the historical time period is constructed in the same way as the feature vector of the target time period, which will not be repeated here. S26. Historical time periods with a cosine similarity greater than the cosine similarity threshold (e.g., 60%) can be used as similar historical time periods to the target time period.

[0031] The maximum and minimum values ​​of the number of transport orders merged within similar historical time periods are used to form the initial range of the target transport order merging quantity.

[0032] The final range of orders that can be merged is dynamically determined based on the initial range of the target shipping orders to be merged (e.g., 5-20 orders) and the minimum number of orders that can be merged (e.g., 8 orders): If the lower limit of the initial range of the target shipping orders to be merged is higher than the minimum number of orders that can be merged (e.g., initial 10-25 orders, minimum 8 orders), then the range of the target shipping orders to be merged is directly taken from the initial range of the target shipping orders to be merged; if the lower limit of the initial range of the target shipping orders to be merged is lower than the minimum number of orders that can be merged (e.g., initial 5-15 orders, minimum 8 orders), then the lower limit of the range of the target shipping orders to be merged is adjusted to the minimum number, for example, the final range is 8-15 orders.

[0033] First, determining the minimum number of orders that can be merged based on cargo weight and volume ensures economies of scale in transportation, preventing resource idleness or excessively high unit costs due to insufficient orders. Second, combining multi-dimensional features and historical data to match similar time periods allows for full utilization of historical experience, making the merging strategy more aligned with actual transportation needs and reducing risks caused by environmental differences. Third, determining the initial range of target transportation order merging quantities based on the merging quantities of transportation orders in similar historical time periods provides a scientifically reasonable boundary for subsequent optimization, avoiding operational complexity caused by blind merging and preventing the optimization space from being too narrow. Finally, combining the initial range of target transportation order merging quantities with the minimum number of mergeable orders determines the final range, achieving a balance between efficiency and feasibility. This ensures that the merged orders meet constraints such as transportation capacity and timeliness, while also improving the overall intelligence level of cold chain transportation management through data-driven decision-making. This process provides a high-quality input range for the wolf pack algorithm, enabling it to efficiently search for the optimal solution within a reasonable space and improving solution efficiency.

[0034] Step 130: Using a wolf pack algorithm based on an improved particle swarm optimization algorithm, multiple target merged transportation orders are generated according to the range of the number of merged transportation orders.

[0035] Specifically, it includes: For any two cold chain transportation orders, calculate the combined index of the two cold chain transportation orders; Initialize the wolf pack based on the merging index of the two cold chain transportation orders, the range of the number of merged transportation orders, and the set of constraints; Construct the first fitness function; Based on the first fitness function, the wolf pack is iteratively optimized multiple times until the first termination condition is met, generating multiple target merged transportation orders. In each iteration, the scout wolves in the wolf pack are identified using the particle swarm optimization algorithm.

[0036] Figure 2 This is a flowchart illustrating the calculation of the combined index of two cold chain transport orders according to some embodiments shown in this specification, such as... Figure 2 As shown, the combined index for two cold chain transportation orders is calculated, including: S11. Based on the weight and volume of the goods in the two cold chain transportation orders, determine whether the two cold chain transportation orders can be merged. Specifically, calculate the sum of the weight and volume of the goods in the two cold chain transportation orders. If the sum of the weight of the goods is greater than the vehicle load or the sum of the volume is greater than the vehicle capacity, then the two orders cannot be merged. If not, the merging index of the two cold chain transportation orders is 0. If yes, then execute S12. S12. Based on the refrigeration environment requirements of the two cold chain transportation orders, calculate the similarity of the refrigeration requirements of the two cold chain transportation orders. Specifically, firstly, extract the refrigeration temperature range of the two cold chain transportation orders and calculate their intersection ratio (e.g., if order A requires 2-6℃ and order B requires 4-8℃, then the temperature similarity is (6-4) / (8-2)≈33%), or use Euclidean distance to measure the numerical difference and then normalize it; in terms of humidity, compare the degree of overlap of the humidity ranges of the two orders (e.g., the overlap rate of 40%-60% and 50%-70% is 10% / 30%≈33%), and sum them according to the preset weights (e.g., temperature accounts for 60%, humidity accounts for 30%, and special requirements account for 10%) to obtain the comprehensive similarity score (0-100%). The higher the score, the stronger the compatibility of the refrigeration environment of the two orders, and the more suitable it is to combine transportation to reduce energy consumption and operating costs. S13. Determine whether two cold chain transportation orders can be merged based on the similarity of their refrigeration requirements. Specifically, if the similarity of the refrigeration requirements of the two cold chain transportation orders is less than the refrigeration requirement similarity threshold (e.g., 40%), then the two orders cannot be merged. If not, the merging index of the two cold chain transportation orders is 0. If yes, then execute S14. S14. Based on the destinations of the two cold chain transport orders, determine the shortest transport routes for both orders. Based on the required delivery times of the two orders, calculate the maximum time difference between them. Based on the shortest transport routes and the maximum time difference, calculate the resource coordination ratio between the two cold chain transport orders. Specifically, calculate the difference between the earliest required delivery time of the order with the earlier delivery time and the deadline for delivery of the order with the later delivery time. Use this difference as the maximum time difference between the two orders. Convert the shortest transport routes of the two orders into the shortest transport time for each order based on the average vehicle speed (e.g., 50 km / h). Calculate the ratio of the maximum time difference to the shortest transport time as the resource coordination ratio between the two cold chain transport orders. S15. Determine whether two cold chain transportation orders can be merged based on their resource coordination ratio. Specifically, if the resource coordination ratio of the two cold chain transportation orders is less than the resource coordination ratio threshold (e.g., 1), then the two orders cannot be merged. If not, the merging index of the two cold chain transportation orders is 0. If yes, then execute S16. S16. Based on the similarity of refrigeration requirements and the resource coordination ratio of the two cold chain transportation orders, calculate the merging index of the two cold chain transportation orders. Specifically, calculate the absolute value of the difference between the resource coordination ratio of the two cold chain transportation orders and a preset resource coordination ratio (e.g., 1.1), using this as the absolute difference in the resource coordination ratio. The merging index of the two cold chain transportation orders can be obtained based on the similarity of their refrigeration requirements and the absolute difference in their resource coordination ratios. The greater the similarity of the refrigeration requirements and the smaller the absolute difference in the resource coordination ratio, the greater the merging index of the two cold chain transportation orders. For example, the merging index of the two cold chain transportation orders can be calculated using the following formula:

[0037] in, This is the combined index of the i-th and j-th cold chain transportation orders. Let represent the similarity of refrigeration requirements between the i-th and j-th cold chain transportation orders. Let be the absolute difference in the resource coordination ratio between the i-th and j-th cold chain transportation orders. and As weight, and Greater than 0,

[0038] The above process comprehensively evaluates multiple dimensions, including cargo weight and volume, refrigeration requirements, destination, and delivery time, ensuring a comprehensive and scientific merging decision. This avoids bias caused by considering only one factor and effectively reduces the risks of merged transportation. By determining whether the weight and volume of the cargo exceed the vehicle's carrying capacity, infeasible merging is eliminated in advance, preventing safety issues such as overloading during transportation and ensuring smooth operation. The similarity of refrigeration requirements is calculated, prioritizing the merging of orders with strong compatibility to reduce additional energy consumption and operating costs caused by differences in refrigeration conditions, thus improving the economics of cold chain transportation. The merging index quantifies the suitability of order merging from multiple dimensions, including cargo, refrigeration, transportation route, and time, providing a scientific basis for wolf pack initialization. This ensures that the initial wolf pack distribution is closer to the feasible solution space, avoiding the low search efficiency caused by blind random initialization.

[0039] In some embodiments, the wolf pack is initialized based on the merging index of the two cold chain transport orders, the range of the target transport order merging quantity, and the set of constraints, including: For each cold chain transportation order, the sampling probability of the cold chain transportation order is determined based on the combined index of any two cold chain transportation orders. When generating each wolf, sample the number of transport orders corresponding to the wolf from the range of target transport order merging quantities. Based on the sampling probability of each cold chain transport order and the number of transport orders corresponding to the wolf, sample cold chain transport orders from multiple centers from multiple cold chain transport orders. Based on the cold chain transport orders from multiple centers, the merging index of any two cold chain transport orders, and the set of constraints, and perform deduplication processing with the generated wolves.

[0040] Specifically, for each cold chain transportation order, the average of its combined index with any other cold chain transportation order can be used as the average combined index of the cold chain transportation order. The ratio of this average combined index to the sum of the average combined indices of all cold chain transportation orders can be used as the sampling probability of the cold chain transportation order. For example, suppose there are three cold chain transportation orders A, B, and C. For cold chain transportation order A, its combined index with cold chain transportation order B is 0.6; the combined index of order A with order C is 0.4. Then the average combined index of cold chain transportation order A with other cold chain transportation orders is (0.6 + 0.4) ÷ 2 = 0.5. The average combined index of cold chain transportation order B with other cold chain transportation orders is assumed to be 0.7, and the average combined index of cold chain transportation order C with other cold chain transportation orders is assumed to be 0.3. The sum of the combined index means of all cold chain transportation orders is the sum of the combined index means of individual cold chain transportation orders A, B, and C, which is 0.5 + 0.7 + 0.3 = 1.5. The sampling probability of cold chain transportation order A is the ratio of its combined index mean to the sum, which is 0.5 ÷ 1.5 ≈ 0.33; the sampling probability of cold chain transportation order B is 0.7 ÷ 1.5 ≈ 0.47; and the sampling probability of cold chain transportation order C is 0.3 ÷ 1.5 = 0.2.

[0041] When generating each wolf, multiple cold chain transportation orders are sorted from highest to lowest sampling probability. Based on the sorting result, the highest-ranking cold chain transportation order is selected, and its current probability is randomly generated. If the current probability of this cold chain transportation order is less than its sampling probability, it becomes the central cold chain transportation order. Otherwise, the same operation is performed on the next cold chain transportation order according to the sorting result, until the number of central cold chain transportation orders reaches the number of transportation orders to be merged for the corresponding wolf. Specifically, the process of generating each wolf mainly revolves around the selection and matching of cold chain transportation orders. The core objective is to select the central order from multiple cold chain transportation orders according to specific rules until the number meets the number of transportation orders to be merged for the corresponding wolf. First, the sampling probability of each cold chain transportation order is sorted from highest to lowest. This step determines the priority order of each cold chain transportation order in the sampling selection; cold chain transportation orders with higher sampling probabilities are ranked higher and are more likely to be selected first. Next, the process begins by drawing from the highest-ranked cold chain transport orders, and a current probability is randomly generated for each order. Then comes the crucial judgment phase: the current probability of this order is compared to the previously determined sampling probability. If the current probability is less than the sampling probability, this cold chain transport order is selected as the central cold chain transport order. Conversely, if the current probability is not less than the sampling probability, it means that this order did not meet the criteria to become the central order in this sampling, and it is disregarded. Instead, the process of drawing, generating current probabilities, and comparing and judging is repeated for the next cold chain transport order in the previously ranked order. The entire process is a cyclical and orderly screening process. Each cycle is based on the results of the previous cycle, either successfully determining a central cold chain transport order or eliminating an order that does not meet the criteria and moving on to the next candidate order. In this process, the ranking result plays a key role in guiding the process, determining the order in which orders are examined, ensuring that orders with higher sampling probabilities have more opportunities to participate in the judgment. Randomly generating current probabilities increases the uncertainty of the selection, making each generation of orders random and diverse, avoiding the selection of orders entirely according to a fixed pattern. As the cycle continues, eligible cold chain transportation orders are continuously identified as the central cold chain transportation orders. When the number of central cold chain transportation orders finally reaches the number of transportation orders merged for that alpha wolf, the selection process for the central cold chain transportation orders for that alpha wolf ends.

[0042] For example, suppose there are three cold chain transportation orders, A, B, and C, with sampling probabilities of 0.6, 0.3, and 0.1 respectively. The target number of transportation orders to be merged is between 1 and 2. To generate a "wolf," we first sample from this range, assuming the number of transportation orders to be merged corresponding to this "wolf" is 1. Then, we sort them in descending order of sampling probability: A, B, C. Starting with order A, we randomly generate a current probability for A, say 0.5. Since 0.5 < 0.6, order A is selected as the central cold chain transportation order. At this point, the target number of central orders for this "wolf" has been reached, and the process ends. If the current probability generated for A is 0.65, 0.65 > 0.6, A does not meet the condition, so we move on to order B. We randomly generate a current probability for B; if it is 0.2, 0.2 < 0.3, B is selected as the central cold chain transportation order. Suppose the initial number of transportation orders to be merged corresponding to this "wolf" is 2. After A is excluded because its current probability is not less than its sampling probability, we then examine B and C in turn. Assuming B is selected, another cold chain transportation order from a center needs to be selected. The remaining orders are then examined sequentially, and the current probability for the next order is generated and compared with the sampling probability until a second cold chain transportation order from a center that meets the conditions is selected. Furthermore, the order combinations and other aspects of the already generated wolves are deduplicated to ensure the uniqueness of each wolf.

[0043] The set of constraints may include vehicle load and vehicle volume.

[0044] For cold chain transport orders that are not selected (i.e., not selected as the center), they are merged sequentially according to the sorting results. During merging, the target merged transport order containing the cold chain transport order with the highest merging index that satisfies the constraint set is selected. Only when the constraint set is satisfied is the target merged transport order practically feasible and will not cause problems such as overloading or exceeding vehicle capacity. For example, suppose there are 10 cold chain transport orders. After initializing the wolfpack operation, orders A, B, and C are selected as the center cold chain transport orders, corresponding to target merged transport orders 1, 2, and 3 respectively. The remaining order DJ is not selected. These unselected orders are sorted by sampling probability as D, E, F, G, H, I, and J. Order D is processed first, and it is checked whether it meets the vehicle load and volume constraints with the target merged transport orders containing orders A, B, and C. If the merging index with target merged transport order 1 is 80, with target merged transport order 2 is 60, with target merged transport order 3 is 70, and with target merged transport order 1, the constraint is satisfied, then order D is merged into target merged transport order 1. Next, process order E. If its merging index with target merged transport order 1 is 75, and its merging index with target merged transport order 2 is 85, and both satisfy the constraints, then merge it into target merged transport order 2. Continue in this manner, processing the remaining unselected orders one by one, until all orders are merged, resulting in a merging scheme that satisfies the constraints while optimizing transportation efficiency as much as possible.

[0045] If the overlap of cold chain transportation orders from multiple centers of the two wolves exceeds the overlap threshold (e.g., 90%), then the two wolves will be deduplicated, and only one of them will be retained.

[0046] First, by determining the sampling probability of each cold chain transportation order based on the merging index of any two cold chain transportation orders, the applicability between these orders is fully considered. This ensures that the sampling probability reasonably reflects the potential value of merging cold chain transportation orders, giving orders with a high probability of merging and high applicability a greater chance of becoming central orders, laying a solid foundation for the construction of subsequent merged transportation orders. Second, by sampling from the range of target transportation order merging quantities to determine the number of "wolves" corresponding to each order, and combining this with sampling the central cold chain transportation orders based on the sampling probability, the diversity and randomness of the initial solution are increased, avoiding getting trapped in local optima. This helps the wolf pack algorithm search in a wider solution space, increasing the likelihood of finding the global optimum. Third, by using constraints such as vehicle load capacity and volume as criteria, merging operations are performed on unselected cold chain transportation orders, ensuring the feasibility of the target merged transportation orders in actual transportation. This effectively avoids problems such as overloading or exceeding vehicle capacity, reducing transportation risks. Meanwhile, merging unselected cold chain transportation orders according to their merging index and constraints prioritizes grouping orders with high suitability, minimizing transportation costs, maximizing resource utilization, and optimizing transportation efficiency. Finally, deduplication is performed on wolves with an overlap greater than a threshold, reducing redundant solutions, avoiding repeated calculations on similar solutions, improving the algorithm's efficiency, and enabling the wolf pack algorithm to iteratively optimize towards the optimal solution more efficiently.

[0047] The first fitness function is correlated with the mean of the merging index of each target consolidated transport order. The merging index of a target consolidated transport order can be the mean of any two cold chain transport orders included in the target consolidated transport order. The higher the mean of the merging index of each target consolidated transport order, the higher the first fitness value. The first fitness function can also be correlated with other indicators, such as the cost and load standard deviation of each target consolidated transport order. The lower the cost and the higher the average load of each target consolidated transport order, the higher the first fitness value.

[0048] For example, the mean of the consolidation index for all target consolidated transportation orders can be calculated as the global consolidation index mean. The sum of the costs for all target consolidated transportation orders can be used as the total cost. The standard deviation of the load for all target consolidated transportation orders can be calculated as the load standard deviation. The global consolidation index mean, total cost, and load standard deviation can be normalized, with the normalized values ​​ranging from [0,1]. Based on the normalized global consolidation index mean, total cost, and load standard deviation, a first fitness function can be constructed. For example, the first fitness function can be:

[0049] in, The first fitness function is... This is the normalized global combined index mean. The normalized total cost, The normalized standard deviation of load capacity. , and As weight, , and Greater than 0, .

[0050] During the iteration process, the wolf pack updates its algorithm based on the evaluation results of the first fitness function. Wolves with higher fitness represent better multi-target combined transportation orders, attracting other wolves to approach and imitate their superior characteristics. Simultaneously, scout wolves identified through particle swarm optimization jump out of the current search range, exploring new possible solutions in a broader solution space, introducing new information to the wolf pack and preventing them from getting trapped in local optima. As the number of iterations increases, the wolf pack continuously adjusts its position, i.e., continuously optimizes the target combined transportation order solution, gradually improving its overall fitness. The first termination condition can be set as reaching a preset maximum number of iterations, at which point the algorithm is considered to have fully searched the solution space; or it can be set as follows: in several consecutive iterations, the improvement in the overall fitness of the wolf pack is less than a certain minimum threshold, meaning that further optimization space is limited and continued iteration yields little benefit. When the first termination condition is met, the algorithm stops iterating. At this point, the solution represented by the wolf with higher fitness in the wolf pack represents the generated multi-target combined transportation order, which can effectively improve the efficiency and feasibility of cold chain transportation.

[0051] In some embodiments, the particle swarm optimization algorithm is used to identify scout wolves within a wolf pack, including: Construct a second fitness function; Based on the first fitness value of each wolf in the wolf pack, multiple candidate wolves are determined from the wolf pack. Specifically, the wolves in the wolf pack are sorted from largest to smallest according to the first fitness value of each wolf in the pack. Based on the sorting results, a certain proportion of candidate wolves are selected. For example, the bottom 10% of the wolves in the sorting results are selected as multiple candidate wolves. Initialize the population based on multiple candidate wolves, where each individual in the population represents at least a subset of the multiple candidate wolves; The population is iteratively optimized according to the second fitness function until the second termination condition is met, and the scout wolves in the wolf pack are determined. The second fitness function is related to the merged similarity of any two scout wolves and the global merged similarity of each scout wolf.

[0052] Specifically, the merged similarity between any two scout wolves can be determined based on the overlap of cold chain transportation orders across multiple centers for both scout wolves; the higher the overlap, the higher the merged similarity. The global merged similarity of a scout wolf can be determined based on the average overlap of cold chain transportation orders between that scout wolf and other wolves in the pack across multiple centers; the higher the average overlap, the higher the global merged similarity. The higher the average merged similarity between any two scout wolves, and the higher the average global merged similarity for each scout wolf, the higher the second fitness value.

[0053] The second termination condition can be a maximum iteration limit, convergence of the second fitness value, etc.

[0054] Based on the first fitness value, a certain proportion of wolves are selected as candidates from the pack. This focuses on relatively less capable individuals, whose areas often have more room for exploration, providing the possibility of discovering new solutions and preventing the algorithm from prematurely converging to local optima. A second fitness function related to the merged similarity is constructed to measure the quality of scout wolves at both local and global levels. The merged similarity of any two scout wolves is determined based on the overlap of central cold chain transportation orders, while the global merged similarity of each scout wolf is determined based on the average overlap with other wolves in the pack. This encourages scout wolves to maintain a certain degree of difference within a local range, avoiding excessive similarity that could limit the search direction.

[0055] Step 140: For each target combined transport order, determine the optimal transport route corresponding to the target combined transport order based on the cold chain transport orders included in the target combined transport order.

[0056] Specifically, it includes: For each target combined transportation order, a global reachable path map is constructed based on the electronic map, where the global reachable path map consists of multiple reachable road segments; The optimal transportation path for the target merged transportation order is determined by using the particle swarm optimization algorithm based on the global reachability path graph corresponding to the target merged transportation order.

[0057] Specifically, the process begins by accurately locating the origin and destination of the order using an electronic map, which serves as the starting anchor point for route planning. The electronic map possesses a vast and detailed road database. Based on the origin and destination information, it employs built-in route search algorithms, such as Dijkstra's algorithm or A* algorithm, to filter out all possible road segments connecting the two points from the massive amount of road data. This filtering process is not a simple listing but rather a comprehensive consideration of multiple factors. On one hand, it considers the special attributes of cold chain transport vehicles, such as their height, weight, and length, excluding road segments with height, weight, or length restrictions that prevent vehicles from passing. On the other hand, it combines real-time traffic information to eliminate road segments currently experiencing severe congestion or road construction that could lead to long delays. Only those road segments that meet the criteria become reachable road segments. These reachable road segments are not isolated; the electronic map uses road intersections, key landmarks, and other nodes as nodes to connect these reachable road segments, forming a global reachable route map.

[0058] In some embodiments, the optimal transportation path for the target merged transportation order is determined using a particle swarm optimization algorithm based on the globally reachable path graph corresponding to the target merged transportation order, including: For each reachable route, the weight of the reachable route is calculated based on historical transportation records; Construct a third fitness function, where the third fitness function is related to the weights of the reachable segments included in the transport path corresponding to the particle; Initialize the population based on the global reachability path graph corresponding to the target merged transportation orders; Based on the third fitness function, the population is iteratively optimized until the third termination condition is met, and the optimal transportation path corresponding to the target merged transportation order is determined.

[0059] Specifically, historical transportation records document the actual performance of each route segment during past transportation processes, such as traffic speed, congestion frequency, and accident rate. Based on these historical transportation records, the weight of reachable routes is calculated. For example, the higher the average traffic speed, the lower the congestion frequency, and the lower the accident rate, the greater the weight of the reachable route.

[0060] Each particle in the population represents a possible transportation route, and the third fitness function is used to quantitatively evaluate these routes. The third fitness function is related to the weights of reachable segments, and can also be related to other factors, such as the total path length and the delivery timeout of each cold chain transportation order in the target merged transportation order. The larger the sum of the weights of the reachable segments in the particle's corresponding transportation route, the smaller the total path length, and the smaller the sum of the delivery timeouts of each cold chain transportation order in the target merged transportation order, the higher the third fitness value.

[0061] In each iteration, each particle adjusts its direction and speed according to the particle swarm optimization (PSO) algorithm's update rules, based on its current position (i.e., the proposed transportation path), its individual optimal position (i.e., the optimal path found in its own history), and the swarm's optimal position (i.e., the optimal path found among all particles), thereby generating a new transportation path. Then, the fitness value of the new path is calculated using the third fitness function and compared and updated with the previous optimal value. This process is repeated until the third termination condition is met, such as reaching the preset maximum number of iterations or the fitness value converging to a certain range. When the algorithm ends, the transportation path corresponding to the optimal particle in the swarm is the optimal transportation path for the target merged transportation order. It maximizes transportation efficiency, reduces costs, and minimizes risks while ensuring transportation feasibility.

[0062] Figure 3 This is a schematic diagram of a cold chain transportation system based on swarm intelligence, as shown in some embodiments of this specification. Figure 3 As shown, a cold chain transportation system based on swarm intelligence may include an order acquisition module, a quantity constraint module, an order merging module, and a route generation module.

[0063] The order acquisition module is used to acquire multiple cold chain transportation orders within a target time period and determine the minimum number of orders that can be merged. The quantity constraint module is used to determine the range of target transportation order merging quantities based on the number of transportation orders merged in similar historical time periods and the minimum number of orders that can be merged. The order merging module is used to generate multiple target merged transportation orders based on the target transportation order merging range using a wolf pack algorithm improved from the particle swarm algorithm. The route generation module is used to determine the optimal transportation route for each target merged transportation order, based on the cold chain transportation orders included in the target merged transportation order.

[0064] A swarm intelligence-based cold chain transportation system can be used to implement a swarm intelligence-based cold chain transportation method, which will not be elaborated here.

[0065] Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments described herein. Other variations may also fall within the scope of this specification. Therefore, alternative configurations of the embodiments described herein are intended to be illustrative rather than limiting, and should be considered consistent with the teachings of this specification. Accordingly, the embodiments described herein are not limited to those explicitly introduced and described herein.

Claims

1. A cold chain transportation method based on swarm intelligence, characterized in that, include: Obtain multiple cold chain transportation orders for the target time period and determine the minimum number of orders that can be merged; Based on the number of shipping orders merged within similar historical time periods and the minimum number of orders that can be merged, determine the range of target shipping order merging quantities; By using a wolf pack algorithm based on an improved particle swarm optimization algorithm, multiple target merged transportation orders are generated according to the range of the number of target transportation orders to be merged. For each target combined transport order, the optimal transport route corresponding to the target combined transport order is determined based on the cold chain transport orders included in the target combined transport order.

2. The cold chain transportation method based on swarm intelligence according to claim 1, characterized in that, Determine the minimum quantity of orders that can be combined, including: Based on the weight and volume of goods in multiple cold chain transportation orders, determine the minimum quantity of orders that can be combined. Based on the number of shipping orders consolidated over similar historical time periods and the minimum number of orders that can be consolidated, the target range for consolidating shipping orders is determined, including: Based on the weight, volume, refrigeration environment requirements, and destination of multiple cold chain transportation orders and multiple historical time periods, similar historical time periods are identified. Based on the number of merged transportation orders in similar historical time periods, determine the initial range of the target transportation order merging quantity; The range of target transportation order merging quantities is determined based on the initial quantity range for target transportation order merging and the minimum quantity of orders that can be merged.

3. The cold chain transportation method based on swarm intelligence according to claim 2, characterized in that, Using a wolf pack algorithm improved from particle swarm optimization, multiple target merged transportation orders are generated based on the range of target transportation order merging quantities, including: For any two cold chain transportation orders, calculate the combined index of the two cold chain transportation orders; Initialize the wolf pack based on the merging index of the two cold chain transportation orders, the range of the target transportation order merging quantity, and the set of constraints; Construct the first fitness function; Based on the first fitness function, the wolf pack is iteratively optimized multiple times until the first termination condition is met, generating multiple target merged transportation orders. In each iteration, the scout wolves in the wolf pack are identified using the particle swarm optimization algorithm.

4. The cold chain transportation method based on swarm intelligence according to claim 3, characterized in that, Calculate the combined index of two cold chain transportation orders, including: S11. Based on the weight and volume of the goods in the two cold chain transportation orders, determine whether the two cold chain transportation orders can be merged. If not, the merging index of the two cold chain transportation orders is 0. If yes, then execute S12. S12. Based on the refrigeration environment requirements of the two cold chain transportation orders, calculate the similarity of the refrigeration requirements of the two cold chain transportation orders. S13. Determine whether two cold chain transportation orders can be merged based on the similarity of their refrigeration requirements. If not, the merging index of the two cold chain transportation orders is 0. If yes, proceed to S14. S14. Based on the destinations of the two cold chain transportation orders, determine the shortest transportation route for the two cold chain transportation orders. Based on the required delivery time of the two cold chain transportation orders, calculate the maximum time difference between the two cold chain transportation orders. Based on the shortest transportation route and the maximum time difference between the two cold chain transportation orders, calculate the resource coordination ratio between the two cold chain transportation orders. S15. Determine whether the two cold chain transportation orders can be merged based on the resource coordination ratio of the two cold chain transportation orders. If not, the merging index of the two cold chain transportation orders is 0. If yes, then execute S16. S16. Calculate the merging index of the two cold chain transportation orders based on the similarity of their refrigeration requirements and the resource coordination ratio.

5. A cold chain transportation method based on swarm intelligence according to claim 3, characterized in that, Based on the merging index of the two cold chain transportation orders, the range of the target transportation order merging quantity, and the set of constraints, initialize the wolf pack, including: For each cold chain transportation order, the sampling probability of the cold chain transportation order is determined based on the combined index of any two cold chain transportation orders. When generating each wolf, sample the number of transport orders corresponding to the wolf from the range of target transport order merging quantities. Based on the sampling probability of each cold chain transport order and the number of transport orders corresponding to the wolf, sample cold chain transport orders from multiple centers from multiple cold chain transport orders. Based on the cold chain transport orders from multiple centers, the merging index of any two cold chain transport orders, and the set of constraints, and perform deduplication processing with the generated wolves.

6. A cold chain transportation method based on swarm intelligence according to claim 3, characterized in that, The particle swarm optimization algorithm is used to identify scout wolves within the wolf pack, including: Construct a second fitness function; Based on the first fitness value of each wolf in the wolf pack, identify multiple candidate wolves from the wolf pack; The population is initialized based on multiple wolf candidates; The population is iteratively optimized according to the second fitness function until the second termination condition is met, and the scout wolves in the wolf pack are determined. The second fitness function is related to the merged similarity of any two scout wolves and the global merged similarity of each scout wolf.

7. A cold chain transportation method based on swarm intelligence according to claim 3, characterized in that, The first fitness function is related to the mean of the merging index of each target merged transportation order.

8. A cold chain transportation method based on swarm intelligence according to any one of claims 1-7, characterized in that, Based on the cold chain transportation orders included in the target consolidated transportation orders, determine the optimal transportation route corresponding to the target consolidated transportation orders, including: For each target combined transportation order, a global reachable path map is constructed based on the electronic map, where the global reachable path map consists of multiple reachable road segments; The optimal transportation path for the target merged transportation order is determined by using the particle swarm optimization algorithm based on the global reachability path graph corresponding to the target merged transportation order.

9. A cold chain transportation method based on swarm intelligence according to claim 8, characterized in that, Using the particle swarm optimization algorithm, the optimal transportation path for the target merged transportation order is determined based on the globally reachable path graph, including: For each reachable route, the weight of the reachable route is calculated based on historical transportation records; Construct a third fitness function, where the third fitness function is related to the weights of the reachable segments included in the transport path corresponding to the particle; Initialize the population based on the global reachability path graph corresponding to the target merged transportation orders; Based on the third fitness function, the population is iteratively optimized until the third termination condition is met, and the optimal transportation path corresponding to the target merged transportation order is determined.

10. A cold chain transportation system based on swarm intelligence, characterized in that, A cold chain transportation method based on swarm intelligence as described in any one of claims 1-9, comprising: The order acquisition module is used to acquire multiple cold chain transportation orders within a target time period and determine the minimum number of orders that can be merged. The quantity constraint module is used to determine the range of target transportation order merging quantities based on the number of transportation orders merged in similar historical time periods and the minimum number of orders that can be merged. The order merging module is used to generate multiple target merged transportation orders based on the target transportation order merging range using a wolf pack algorithm improved from the particle swarm algorithm. The route generation module is used to determine the optimal transportation route for each target merged transportation order, based on the cold chain transportation orders included in the target merged transportation order.

Images