Cold chain transport capacity intelligent scheduling method and system based on demand prediction

Abstract

The invention relates to the technical field of intelligent scheduling, and discloses a cold chain transport capacity intelligent scheduling method and system based on demand prediction. The method comprises the following steps: based on historical distribution data of a plurality of distribution points, predicting future demands of each distribution point for a plurality of cold-chain commodities in a future scheduling period; generating a distribution scheduling plan based on the prediction result, including determining a target distribution point visited every day through a value evaluation model, distributing cold chain transport vehicles and planning a distribution route; and generating a loading scheme for the vehicle based on the prediction result and the distribution scheduling plan, and determining a cargo combination, a placement position and a temperature zone attribution. The method can improve the transport capacity utilization rate, and reduces the distribution cost and the risk of goods damage.

Description

Technical Field

[0001] This application relates to the field of intelligent scheduling technology, and in particular to a method and system for intelligent scheduling of cold chain transportation capacity based on demand forecasting. Background Technology

[0002] Cold chain logistics is a crucial link in ensuring the quality and safety of temperature-sensitive goods such as pharmaceuticals and fresh food. With increasing market demand and more complex distribution networks, the traditional cold chain scheduling model based on fixed routes and experience is no longer sufficient to meet the operational needs of high efficiency, low loss, and dynamic response.

[0003] In recent years, some studies have attempted to introduce artificial intelligence and predictive models to optimize the cold chain transportation process. For example, patent document CN120013403A provides a method and system for optimizing cold chain transportation routes based on AI prediction. This method analyzes historical order data, traffic flow, and weather events to predict the probability of route congestion and the freshness requirements of goods, and generates and evaluates multiple candidate routes accordingly to achieve real-time optimization of the transportation process.

[0004] However, this existing technology still has the following limitations: It focuses on dynamic route adjustment but does not systematically address the issues of capacity scheduling and loading optimization. The solution lacks an integrated dynamic temperature zone division and loading optimization mechanism: In cold chain transportation, the spatial layout and placement order of goods in different temperature zones inside the vehicle directly affect the preservation effect and loading and unloading efficiency. This solution lacks intelligent optimization of the internal temperature zone division, goods placement position and unloading order, and cannot achieve refined management of the loading process. The value assessment dimension is relatively singular: In terms of delivery point selection and access scheduling, this method fails to comprehensively consider multi-dimensional value factors such as cumulative order value, cargo damage risk cost, and geographical synergy effect, making it difficult to achieve the optimal balance between cost and service.

[0005] Therefore, the present invention provides a method and system for intelligent scheduling of cold chain transportation capacity based on demand forecasting. Summary of the Invention

[0006] The embodiments in this specification provide the following technical solutions: A method for intelligent scheduling of cold chain transportation capacity based on demand forecasting includes: Step S1, predicting the future demand for various cold chain commodities at each delivery point within a preset scheduling period based on historical delivery data from multiple delivery points; Step S2, generating a delivery scheduling plan based on the forecast results, the delivery scheduling plan including: determining the set of target delivery points to be visited daily within the preset scheduling period based on the forecast results and a pre-created value assessment model, the value assessment model being used to assess the value generated by visiting delivery points on different dates; allocating multiple cold chain transport vehicles to the set of target delivery points to be visited on the corresponding dates based on transportation resource data; creating a delivery route for each cold chain transport vehicle to its assigned delivery point; Step S3, generating a vehicle loading plan for the cold chain transport vehicles based on the forecast results and the delivery scheduling plan, the vehicle loading plan being used to describe the combination, quantity, placement, and temperature zone affiliation of the goods loaded into the vehicle.

[0007] Preferably, the creation process of the value assessment model includes: calculating the comprehensive value for each delivery point and each candidate visit date. The comprehensive value is obtained by weighted summation of the following three value components: the cumulative predicted order value from the start date to the candidate visit date as the first value component; the cumulative estimated loss cost from the start date to the candidate visit date as the second value component, which includes order cancellation losses and cargo damage risk costs; and calculating the third value component based on regional synergy, which is used to quantify the additional value that can be brought by combining visits to the delivery point and its geographically adjacent delivery points in the same delivery route.

[0008] Preferably, the calculation of the third value component based on regional synergy includes: identifying other delivery points to be evaluated located within a preset geographical range of the delivery point on the candidate visit date; predicting the synergy benefits generated when the delivery point is visited in combination with at least one delivery point to be evaluated, compared to visiting them individually; the calculation of synergy benefits takes into account the saved transportation distance costs, the time window matching gain, and the increased costs due to the complexity of route merging.

[0009] Preferably, step S3 includes: a candidate cargo identification sub-step: identifying a set of candidate cargo to be loaded based on the allocation results of the delivery points to be visited by the target cold chain transport vehicle; a dynamic temperature zone division sub-step: dynamically planning the variable temperature zone layout of the vehicle's interior space according to the volume ratio of cargo at different temperature levels in the candidate cargo set and the delivery route environment, wherein the variable temperature zone layout includes at least a non-fixed ratio division of the frozen zone and the refrigerated zone; and a cargo placement optimization sub-step: using an optimization algorithm based on thermodynamic simulation, combined with the variable temperature zone layout, determining the specific placement position of the final cargo combination selected from the candidate cargo set inside the vehicle.

[0010] Preferably, the dynamic temperature zone division sub-step includes: based on the candidate cargo set, calculating the estimated total volume of cargo in each temperature zone; based on the environmental prediction data of the delivery route, establishing an optimization model with the goal of minimizing overall temperature control energy consumption and temperature fluctuation risk, and solving for the optimal solution to obtain the optimal spatial division of the frozen zone and refrigerated zone in the vehicle's interior space.

[0011] Preferably, the cargo placement optimization sub-step includes: constructing a simplified thermodynamic model of the internal space of the target cold chain transport vehicle; input parameters include the initial temperature of the cargo, packaging properties, vehicle insulation performance, route ambient temperature, and the expected door opening and unloading sequence; performing transient heat conduction simulation to predict the temperature change curves of different locations inside the vehicle throughout the delivery process, and calculating the temperature stability score of each candidate placement location for various cargoes based on the temperature change curves; according to the degree of temperature sensitivity, and based on the temperature stability score, prioritizing the placement of cargoes with high temperature sensitivity in the area with the highest temperature stability score for the corresponding cargo.

[0012] Preferably, the cargo placement optimization sub-step also includes: after placing the cargo based on the stability score, determining the unloading order according to the delivery route, stacking the pressure-resistant cargo that needs to be unloaded last at the bottom of the front of the carriage, and placing the cargo that needs to be unloaded first on the top layer.

[0013] Preferably, a delivery route is created for each cold chain transport vehicle to its assigned delivery point, including: for each cold chain transport vehicle, the initial location and the location assigned to the delivery point are used as inputs to a path optimization problem; the path optimization problem is solved with the objective of minimizing the total travel cost, which includes the travel distance cost, travel time cost, and temperature control energy consumption cost caused by the vehicle's refrigeration unit; the imposed constraints include: the specified harvest time constraint for each delivery point and the driver's continuous driving time regulation constraint, and a heuristic algorithm that combines real-time traffic information for dynamic evaluation is used to solve the problem.

[0014] A cold chain transportation capacity intelligent scheduling system based on demand forecasting is also provided to implement the aforementioned cold chain transportation capacity intelligent scheduling method based on demand forecasting. The system includes: a demand forecasting module, which predicts the future demand for various cold chain commodities at each delivery point within a preset scheduling period based on historical delivery data from multiple delivery points; a scheduling plan module, which generates a delivery scheduling plan based on the forecast results. The delivery scheduling plan includes: determining the set of target delivery points to be visited each day within the preset scheduling period based on the forecast results and a pre-created value assessment model, whereby the value assessment model is used to evaluate the value generated by visiting delivery points on different dates; allocating multiple cold chain transport vehicles to the set of target delivery points to be visited on the corresponding dates based on transportation capacity resource data; and creating a delivery route for each cold chain transport vehicle to its assigned delivery point; and a loading optimization module, which generates a vehicle loading scheme for cold chain transport vehicles based on the forecast results and the delivery scheduling plan, whereby the vehicle loading scheme describes the combination, quantity, placement, and temperature zone affiliation of the goods loaded into the vehicle.

[0015] Compared with the prior art, the beneficial effects of the present invention are at least as follows: The technical solution provided in this application intelligently selects the best delivery point each day through a precise demand forecasting and value assessment model, optimizes vehicle allocation and route planning, reduces empty mileage and waiting time, and significantly improves vehicle utilization and delivery timeliness. Combined with a loading scheme based on dynamic temperature zone division and thermodynamic simulation, it scientifically plans cargo placement and temperature zone layout, minimizes temperature fluctuations and energy consumption, and effectively reduces cargo damage rate and refrigeration costs. By introducing a regional synergy effect assessment mechanism, it identifies and optimizes the combined delivery of adjacent delivery points, improves the efficiency of route merging, and achieves spatial synergy and time window matching of transportation resources. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of an embodiment of an intelligent cold chain transportation capacity scheduling method based on demand forecasting in this application. Figure 2 This is a schematic diagram of one embodiment of an intelligent cold chain transportation capacity scheduling system based on demand forecasting in this application. Detailed Implementation

[0018] This application provides a method and system for intelligent scheduling of cold chain transportation capacity based on demand forecasting. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data used can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0019] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 An embodiment of a cold chain transportation capacity intelligent scheduling method based on demand forecasting in this application includes: Step S1: Based on historical delivery data from multiple delivery points, predict the future demand for various cold chain products at each delivery point within the preset scheduling period. Specifically, in order to address the problems of high delivery costs, high risk of cargo damage, and low capacity utilization in cold chain logistics caused by information lag and extensive planning, a data-driven approach is first used to mine demand patterns from massive amounts of historical data (such as daily delivery records to each store over the past year). Based on these patterns, combined with relevant information (weather information and calendar) for the future preset scheduling cycle, the future demand for various cold chain products is predicted for each day of the future preset scheduling cycle. The historical delivery data includes at least the daily delivery volume history of each type of cold chain product, and the future demand includes the demand quantity, demand category, and corresponding temperature requirements.

[0020] Step S2: Based on the prediction results, generate a delivery scheduling plan. The delivery scheduling plan includes: determining the set of target delivery points that need to be visited each day within a preset scheduling period based on the prediction results and a pre-created value assessment model. The value assessment model is used to evaluate the value generated by visiting delivery points on different dates. Based on transportation resource data, assign multiple cold chain transport vehicles to the set of target delivery points that need to be visited on the corresponding dates. Create a delivery route for each cold chain transport vehicle to reach its assigned delivery point. Specifically, the forecast results refer to the future demand for various cold chain products at each delivery point within the preset scheduling period. In order to transform the forecasted discrete demand into executable and efficient daily delivery tasks, the questions of when to deliver, which vehicle to use, and which route to take are addressed.

[0021] First, the delivery point selection sub-step is executed. Since not all delivery points require immediate delivery, a value assessment model is used to intelligently determine the optimal daily delivery combination, achieving an optimal balance between meeting customer service levels and controlling delivery frequency. For example, delivering to a store whose inventory is about to run out tomorrow has a high value (avoiding lost sales and maintaining customer relationships), while replenishing a store with sufficient inventory in advance may have a lower value than the additional transportation costs. Based on the forecast results and the pre-created value assessment model, the set of target delivery points that need to be visited daily within a preset scheduling cycle is determined. The specific method for determining the target delivery point set based on the value assessment model will be explained in detail later. The delivery point selection sub-step realizes a proactive and economically driven delivery departure mechanism, avoiding resource waste caused by fixed-frequency delivery and reducing total logistics costs.

[0022] To match tasks with the most suitable transportation resources and ensure task feasibility and high resource utilization, a capacity allocation sub-step is executed. The principle is based on matching according to multi-dimensional constraints, including vehicle capacity (tonnage, volume), temperature zone capability (whether it has multiple temperature zones), available time (driver schedule, vehicle maintenance plan), and business planning (some vehicles are fixed to serve specific areas). For example, if a target set for a certain day includes 15 delivery points and a total of 5 tons of goods need to be delivered, including 3 tons of refrigerated and 2 tons of frozen goods, a usable van with a load capacity of 6 tons and independent refrigerated and frozen dual temperature zones is selected from the fleet and assigned to this group of tasks. The capacity allocation sub-step ensures that the transportation capacity and tasks are accurately matched in terms of physical and temporal attributes, improves vehicle utilization, and ensures that the cold chain remains uninterrupted.

[0023] To plan the shortest, lowest-cost, or most reliable travel sequence given a list of delivery points and vehicle constraints, a delivery route creation sub-step is executed. This involves establishing a vehicle routing problem model, adding cold chain-specific time windows (such as delivery point receiving times) and temperature control energy consumption constraints, and using optimization algorithms (such as genetic algorithms) to find the optimal or near-optimal solution. For example, to plan routes for the 15 delivery points of the aforementioned vehicles, the algorithm considers not only the shortest distance but also congested roads during morning rush hour, peak harvest periods, and minimizing the number of times vehicle doors are opened to maintain a stable temperature inside the vehicle. The final result is a detailed route starting from the warehouse, passing through each delivery point in sequence, and finally returning to the warehouse. This delivery route creation sub-step can shorten vehicle travel time, reduce fuel consumption and refrigeration energy consumption, improve delivery timeliness, and ensure product quality.

[0024] Step S3: Based on the prediction results and delivery scheduling plan, generate a vehicle loading plan for cold chain transport vehicles. The vehicle loading plan is used to describe the combination, quantity, placement and temperature zone of the goods loaded into the vehicle.

[0025] Specifically, to address the issue of loading goods onto trucks safely, efficiently, and with good quality within limited space, thereby reducing transportation costs and damage rates, the loading problem is transformed into a three-dimensional spatial layout optimization problem. This is compounded by the unique constraints of the cold chain: physical isolation of goods in different temperature zones, avoiding crushing lighter goods with heavier ones, and ensuring stacking stability. For example, when loading 200 boxes of goods destined for 15 distribution points, the unloading sequence is first determined based on the delivery route. The pressure-resistant goods to be unloaded last are stacked at the bottom front of the truck, while the fragile fresh produce to be unloaded first is placed on top. Separating the frozen and refrigerated sections ensures no temperature interference, ultimately resulting in a visual loading guide. This process maximizes the load capacity per vehicle, reduces the number of vehicles required, standardizes loading and unloading operations, minimizes cargo damage, and reduces rummaging time during unloading by optimizing the layout, thus improving operational efficiency.

[0026] By closing the gaps between the above steps, and through a closed-loop approach of prediction, planning, and loading, the cold chain can be upgraded from experience-driven to data-intelligent driven, achieving the comprehensive effects of reducing total logistics costs, improving the quality of goods delivery, and enhancing the efficiency of transportation capacity resource allocation.

[0027] In one specific embodiment, the process of creating a value assessment model includes the following steps: For each delivery point and each candidate visit date, the overall value is calculated. The overall value is obtained by weighted summation of the following three value components: The cumulative predicted order value from the start date to the candidate access date will be used as the first value component; The cumulative estimated loss cost from the start date to the candidate access date is used as the second value component. The cumulative estimated loss cost includes order cancellation losses and cargo damage risk costs. The third value component is calculated based on regional synergy. This third value component is used to quantify the additional value that can be obtained by combining visits to the delivery point with its geographically adjacent delivery points on the same delivery route.

[0028] Specifically, in order to establish a mathematical model that can quantify the comprehensive value of visiting a specific delivery point on a specific date and provide a scientific numerical basis for subsequent decision-making, the decision of whether a delivery point should be delivered on a certain day depends not only on its own urgency of demand (first and second value components) but also on its ability to share logistics resources with other nearby delivery points (third value component). In order to measure how much sales revenue might be lost in the future if replenishment is not carried out today, i.e., the sales potential brought by replenishment, high-demand, high-value delivery points are identified and their supply is prioritized. From the start date (current inventory count date) to the candidate visit date (each day within the future preset scheduling cycle), the predicted daily sales revenue is accumulated as the first value component.

[0029] To quantify the risk costs of delayed replenishment, including explicit order cancellation losses and implicit damage to reputation and spoilage losses, and to balance delivery costs with delay risks, thus avoiding customer loss or goods spoilage due to excessive pursuit of delivery efficiency, a model is established based on historical data to predict the relationship between out-of-stock duration and order cancellation rate to estimate the first loss cost. In addition, for different types of goods, an inventory time-quality decay-value depreciation model is established to predict the second loss cost. For example, the spoilage rate of leafy green vegetables increases exponentially over time. The two loss costs are added together as the second value component.

[0030] To capture the optimization potential in geospatial areas, we assess the additional value created by arranging multiple nearby delivery points on the same day and along the same route. This breaks the limitations of single-decision approaches and generates efficient delivery clusters. We calculate a third value component based on regional synergy effects. The specific calculation method will be explained in detail later. The third value component is used to quantify the additional value brought by combining visits to the delivery point and its geographically adjacent delivery points on the same delivery route.

[0031] Business managers can assign weights to the three components based on their priorities (e.g., if revenue is prioritized, increase the weight of the first value component; if quality is prioritized, increase the weight of the second value component). By calculating the comprehensive value of each delivery point on different visit dates daily, the day with the highest comprehensive value is selected as the recommended visit day, and finally, the daily set of delivery points is generated.

[0032] In one specific embodiment, calculating the third value component based on regional synergy includes the following steps: Identify other delivery points to be evaluated that are located within a preset geographic area of ​​the delivery point on the candidate visit date; Predict the synergistic benefits of combining visits to a delivery point with at least one delivery point to be evaluated, compared to visiting them individually; The calculation of synergistic benefits takes into account the cost savings in transportation distance, the time window matching gain, and the increased cost due to the complexity of route merging.

[0033] Specifically, in order to perform refined and quantifiable calculations of the value of regional synergy effects, potential synergy partners are screened and the specific benefits of combined visits are predicted by pre-setting geographical scope.

[0034] First, by identifying other delivery points within a preset geographical range of the current delivery point on the candidate visit date, a reasonable potential collaborative delivery circle is defined for the currently evaluated delivery point. Typically, the synergistic effect diminishes with distance; beyond a certain range, the cost of detours may outweigh the savings from collaboration. Therefore, based on the principle of geographical proximity, delivery points to be evaluated are identified within a preset radius (e.g., within 10 kilometers) or based on travel time (e.g., within a 20-minute drive). Next, the synergistic benefits of combining the delivery point with at least one of the evaluated delivery points are predicted compared to individual visits. For example, the estimated cost of delivering to three delivery points A, B, and C individually is 180, 170, and 190, respectively, totaling 540. By planning the optimal route through combined delivery, the total cost of the optimal route is calculated to be 380 (because the shared distance and vehicle fixed costs are less than separate deliveries). Therefore, the synergistic benefit is 540 - 380 = 160.

[0035] The calculation of synergistic benefits needs to consider the saved transportation distance costs, the time window matching degree gain, and the increased costs due to route merging complexity. The calculation formula is: Synergistic Benefits = α × Saved Transportation Distance Costs + β × Time Window Matching Degree Gain - γ × Increased Costs Due to Route Complexity. Here, α is the weight of saved transportation distance costs, reflecting the actual savings in fuel, tolls, vehicle depreciation, etc. β is the weight of time window matching degree gain, used to measure the contribution of combined delivery to meeting the delivery time requirements of each delivery point. γ is the weight of increased costs due to route complexity. Combined delivery may increase the number of stops and loading / unloading times, thus introducing hidden costs. γ is used to offset these complexities into costs. The values ​​of α, β, and γ can be adjusted by managers according to business strategies. For example, in a customer experience-first model, β can be increased, and the system will be more inclined to generate combinations with high time window matching degree. In a model of extreme cost reduction, α can be increased to make the system focus more on saving direct transportation costs. This method transforms the spatial relationships in the delivery grid into precise optimization drivers by establishing a quantifiable and adjustable method for calculating collaborative benefits. It can intelligently identify and prioritize delivery combinations that can reduce costs and improve service quality, thereby improving the efficiency and service level of the entire cold chain delivery.

[0036] In one specific embodiment, step S3 specifically includes the following steps: Candidate cargo identification sub-step: Based on the allocation results of the delivery points to be visited by the target cold chain transport vehicle, identify the set of candidate cargo to be loaded; Dynamic temperature zone division sub-step: Based on the volume ratio of goods at different temperature levels in the candidate goods set and the delivery route environment, dynamically plan the variable temperature zone layout of the vehicle's interior space. The variable temperature zone layout includes at least a non-fixed ratio division of the frozen zone and the refrigerated zone. Cargo placement optimization sub-step: Using an optimization algorithm based on thermodynamic simulation, combined with a variable temperature zone layout, the specific placement position of the final cargo combination selected from the candidate cargo set inside the vehicle is determined.

[0037] Specifically, in order to solve the problem of balancing space utilization and temperature control in the cold chain loading process, the temperature zone layout is dynamically planned and the cargo placement is optimized based on thermodynamic simulation. This maximizes the loading capacity within the cargo compartment while ensuring that all types of goods are in the most suitable temperature environment.

[0038] By executing the candidate cargo identification sub-step, an accurate list of cargo to be loaded is determined for subsequent loading optimization. Based on the generated delivery scheduling plan, all delivery points allocated to the target cold chain transport vehicle and their corresponding predicted demand cargo lists are extracted. Then, by executing the dynamic temperature zone division sub-step, the space ratio of the frozen and refrigerated zones in the compartment is intelligently adjusted according to the current temperature composition of the cargo. That is, based on the estimated total volume of frozen and refrigerated goods in the candidate cargo set, the optimal space allocation ratio is calculated. For example, if the total volume of frozen goods is 12 cubic meters, the total volume of refrigerated goods is 6 cubic meters, and the total volume of the compartment is 20 cubic meters, through dynamic planning, the partition is adjusted so that the frozen zone occupies 13 cubic meters and the refrigerated zone occupies 7 cubic meters. The specific dynamic temperature zone division sub-step will be explained in detail later.

[0039] To address the optimal placement of goods within a given temperature zone layout, a cargo placement optimization sub-step is implemented. This ensures that goods fit comfortably while minimizing temperature fluctuations in the environment surrounding temperature-sensitive goods during delivery. A thermodynamic simulation-based optimization algorithm simulates temperature field changes during delivery, prioritizing the placement of temperature-sensitive goods in simulated temperature-stable zones. For example, simulations show that during summer deliveries, areas near the doors and side walls experience significant temporary temperature increases after repeated door openings. By dividing the cargo compartment into virtual grids and calculating temperature stability scores, calculations reveal that the center area furthest from the doors in the compartment has the highest stability score for goods like tuna. Therefore, tuna is prioritized for placement in the center of this highest-stable area. For temperature-tolerant goods such as bok choy, placement closer to the doors facilitates earlier unloading. This method, through dynamic temperature zone division and precise placement based on thermodynamic simulation, achieves a triple improvement in refrigerated cargo compartment loading capacity, energy efficiency, and cargo quality preservation, ensuring the safe delivery of core perishable goods.

[0040] In one specific embodiment, the dynamic temperature zone division sub-step specifically includes the following steps: Based on the candidate cargo set, the estimated total volume of cargo in each temperature zone is calculated. Based on the environmental prediction data of the delivery route, an optimization model is established with the goal of minimizing overall temperature control energy consumption and temperature fluctuation risk. The optimal solution is obtained to obtain the optimal spatial division of the frozen and refrigerated areas in the vehicle interior space.

[0041] Specifically, in order to establish optimized physical boundary conditions and ensure that the generated partitioning scheme is feasible within the vehicle's hardware capabilities, the feasible area for temperature zone partitioning is first constructed based on the vehicle's internal dimensions and the range of partition movement. For example, in a multi-temperature transport vehicle, the interior of the compartment is a regular rectangular space with dimensions of 5 meters long, 2.2 meters wide, and 2.1 meters high. The movable heat insulation panels inside can move on guide rails between 1 meter and 4 meters away from the front door of the compartment, thus determining the vehicle's spatial constraint parameters.

[0042] To ensure that the cargo can fit, the loading requirements are converted into space requirements as a constraint that the partitioning scheme must meet. Based on the cargo's specifications (box and basket dimensions) and quantity, the theoretical volume of its stacking thickness is calculated. The necessary loading and unloading gaps and cold air circulation channels are considered. The theoretical volume is then converted into an empirical coefficient (such as 1.2) to obtain the actual required volume as the estimated total volume of cargo in each temperature zone. For example, after quantification, the estimated total volume of the frozen zone and the refrigerated zone are 11 cubic meters and 8 cubic meters, respectively.

[0043] Since the external ambient temperature directly affects the heat load of the carriage, stopping and opening the doors will cause temperature fluctuations inside the carriage. To provide input data for evaluating the temperature control performance and energy consumption under different partitioning schemes, and to ensure the optimization results align with actual operating conditions, based on environmental prediction data for the delivery route, the estimated ambient temperature curve from 9:00 AM to 5:00 PM is expected to gradually increase from 25°C to 35°C and then decrease to 30°C. The route includes four delivery points, each with an estimated 15-minute stop for unloading, during which the carriage doors are fully open. This data is used for heat load and temperature fluctuation simulation, establishing an optimization model aimed at minimizing overall temperature control energy consumption and temperature fluctuation risk. The length of the frozen section is used as a decision variable; for example, if the length of the frozen section is x, then the length of the refrigerated section is (5-x) meters. Constraints are set, including volume constraints, which include: the total volume of the frozen section: Total volume of the cold storage area: Where η is the effective loading coefficient (e.g., 0.85), x ranges from [1,4], and the optimization objective is to minimize the total risk cost, where total cost = Temperature fluctuation risk can be estimated using a simplified model. For example, temperature fluctuation risk is positively correlated with the surface area of ​​the area; the larger the surface area, the greater the impact of opening doors. It is also related to the sensitivity of goods to temperature. Therefore, it can be constructed as: Temperature fluctuation risk = Energy consumption risk can be estimated based on the volume of each temperature zone and the temperature difference between the set temperature and the ambient temperature. The theoretical power consumption of the chiller required to maintain the temperature can be estimated as follows: Energy consumption risk = Energy consumption risk is a simplified indicator used to compare the relative energy consumption of different partitioning schemes; it does not represent the actual energy consumption value. w1 and w2 are weights that can be adjusted according to business priorities (e.g., increase w1 when cargo damage costs are high; increase w2 when electricity costs are sensitive). Then, an optimization algorithm, such as enumeration, is used to solve the above optimization model. Assuming the optimal solution is x=2.8, the partitioning scheme is: a frozen zone with a length of 2.8 meters and a theoretical volume of approximately 12.9 cubic meters; and a refrigerated zone with a length of 2.2 meters and a theoretical volume of 10.1 cubic meters. This scheme satisfies both the volume requirement and minimizes the total risk cost. This method, through mathematical optimization, eliminates the uncertainty and suboptimal nature of relying on manual experience in partitioning temperature zones, providing a quantitative basis for the optimal space allocation scheme for each task.

[0044] In one specific embodiment, the goods placement optimization sub-step specifically includes the following steps: Construct a simplified thermodynamic model of the interior space of the target cold chain transport vehicle; Input parameters include the initial temperature of the goods, packaging properties, vehicle insulation performance, route ambient temperature, and the expected door opening and unloading sequence. Perform transient heat conduction simulation to predict the temperature change curves of different locations inside the vehicle throughout the delivery process, and calculate the temperature stability score of each candidate placement location for various goods based on the temperature change curves. Based on the degree of temperature sensitivity and the temperature stability score, goods with high temperature sensitivity are prioritized for placement in the area with the highest temperature stability score for the corresponding goods.

[0045] Specifically, to create a model that reflects the temperature distribution and variation patterns inside the refrigerated truck compartment and lay the foundation for simulation calculations, a simplified three-dimensional thermodynamic model of the interior space of the target refrigerated transport vehicle is constructed. The complex real-world compartment is abstracted into a simplified model containing key thermal elements. The finite difference method is typically used to divide the interior space into several small units, each with specific heat capacity and thermal resistance properties, to simulate the heat transfer process between cargo, air, and the compartment walls. For example, in a refrigerated truck component model, the compartment is abstracted as a cuboid, with walls composed of multiple layers of material, each layer assigned a corresponding thermal conductivity and thickness. The interior space is divided into multiple 3D meshes, with each mesh unit representing a small piece of cargo. The region provides initial states and dynamic boundary conditions for the thermodynamic model by inputting multi-dimensional parameters, ensuring that each simulation closely matches the specific circumstances of the corresponding delivery task. The input parameters include two types: physical property parameters (used to describe inherent thermal characteristics) and operating condition parameters (used to describe the dynamic changes in the task execution process). Physical properties include the specific heat capacity of the goods, the rate of heat generation from respiration, the thermal conductivity of the goods packaging materials, and the thermal resistance of the vehicle insulation layer, etc. Operating condition parameters include the initial temperature of the goods, the route ambient temperature (external ambient temperature), and the time for unloading by opening the door, etc. Through the refined input of multi-dimensional parameters, the simulation can reflect the combined effects of the goods' characteristics and the external environment, making the simulation results closer to reality.

[0046] To dynamically predict the temperature change history of each grid cell inside the vehicle over time from loading to unloading at the final station, and to assess temperature stability, a set of differential equations describing energy conservation is solved based on initial and boundary conditions. The entire delivery time is divided into small time steps using the finite element method. Within each step, the inflow and outflow of heat into each grid cell are calculated, and its temperature value is updated. This yields the temperature change curves of various points inside the vehicle over time throughout the entire delivery cycle. Based on these temperature change curves, the temperature fluctuation index for each candidate placement location is calculated as a temperature stability score. For example, if a certain type of cargo is placed in a grid cell inside the vehicle, its temperature curve during the entire simulation is analyzed. Two indicators are calculated: the maximum positive deviation (the difference between the highest temperature and the required temperature) and the cumulative high-temperature time (the total duration of time the temperature is higher than the required temperature). Combining these two indicators, a temperature stability score is generated.

[0047] Based on the degree of temperature sensitivity and temperature stability scores, goods with high temperature sensitivity are prioritized for placement in the area with the highest temperature stability score for their respective goods. For example, if three types of goods are to be loaded: A (core product, high value, extremely sensitive to temperature fluctuations), B (regular product, relatively sensitive), and C (gift, not sensitive), given the stability scores of each location, and assuming all physical constraints are met, goods A will be prioritized for placement in the area with the highest stability score for goods A. From the remaining locations, goods B will then be placed in the area with the highest stability score for goods B, and finally, goods C will be assigned a location.

[0048] Furthermore, the temperature fluctuation index of each candidate placement location for various goods is calculated as a temperature stability score. This includes: placing a certain type of goods in a grid within the carriage and analyzing its temperature curve throughout the simulation process, calculating two indicators, namely a first indicator and a second indicator. The first indicator is the maximum positive deviation, which refers to the difference between the highest temperature and the required temperature. The second indicator is the cumulative high temperature time, which refers to the total duration during which the temperature is higher than the required temperature. The two indicators are combined to generate a temperature stability score.

[0049] Furthermore, by combining the two indicators, a temperature stability score is generated, including: The maximum and minimum values ​​of the first indicator are statistically calculated in history. The first interval range of the first indicator is calculated based on the maximum and minimum values. The first interval range is divided into multiple different first sub-intervals, and a corresponding first score is set for each different first sub-interval. The maximum and minimum values ​​of the second indicator are statistically calculated in history. The range of the second indicator is calculated based on the maximum and minimum values. The range of the second indicator is divided into multiple different second sub-ranges, and a corresponding second score is set for each different second sub-range. The average of the first and second scores is used as the final temperature stability score.

[0050] Specifically, for example, if the maximum value of the first indicator is 2 and the minimum value is 0, then the first interval range of the first indicator is [0,2]. Suppose this first interval range is divided into 10 different first sub-intervals, for example, each sub-interval is divided into intervals of 0.2. Then, from the first to the tenth first sub-interval, the corresponding first score decreases by 10 from 100, becoming 100, 90, ..., 10. The calculation method for the second score is the same as that for the first score.

[0051] In one specific embodiment, the goods placement optimization sub-step further includes: After placing the goods based on the stability rating, the unloading order is determined according to the delivery route. The pressure-resistant goods that need to be unloaded last are piled at the bottom of the front of the truck, and the goods that are unloaded first are placed on the top layer.

[0052] Specifically, in order to standardize loading and unloading operations and reduce the risk of cargo damage caused by multiple moves, this step optimizes the layout by determining the unloading sequence according to the delivery route. The pressure-resistant goods that need to be unloaded last are stacked at the bottom front of the truck bed, while the goods that need to be unloaded first are placed on the top layer. This reduces the time spent searching during unloading and improves operational efficiency.

[0053] In one specific embodiment, creating a delivery route for each cold chain transport vehicle to its assigned delivery point includes the following steps: For each cold chain transport vehicle, the initial location and the location assigned to the delivery point are used as inputs to the route optimization problem; The goal is to minimize the total driving cost, which includes the cost of driving distance, the cost of driving time, and the energy consumption cost of temperature control due to the vehicle's cold engine operation. The imposed constraints include: the specified pickup time for each delivery point and the regulations on the continuous driving time of drivers. A heuristic algorithm that combines real-time traffic information for dynamic evaluation is used to solve the problem.

[0054] Specifically, in order to abstract the delivery task into a computable mathematical optimization problem, the initial position of the cold chain transport vehicle and its assigned delivery points are modeled as nodes on a complete graph. The edge weights between nodes are composite costs (total travel costs) determined by multiple factors. The total travel cost includes travel distance cost, travel time cost, and temperature control energy consumption cost. The travel distance cost is calculated based on mileage and vehicle fuel per unit mileage, the travel time cost is calculated based on travel time and driver labor cost per unit time, and the temperature control energy consumption cost is obtained by multiplying the electricity consumption during the delivery period by the corresponding electricity price.

[0055] To ensure that the generated routes are not only cost-effective but also legal, compliant, and feasible, harvest time constraints are set. Since each delivery point has a corresponding harvest time period, it is necessary to ensure that the vehicle arrives within the corresponding harvest time period. In addition, legal constraints on the driver's continuous driving time are added. For example, the regulations require that the driver must rest for at least 20 minutes after driving continuously for 4 hours. Therefore, when the cumulative driving time is close to 4 hours, rest time is added to the subsequent nodes and included in the total time to ensure the practical feasibility of the plan.

[0056] Subsequently, to efficiently solve combinatorial optimization problems and provide optimal feasible solutions, a heuristic algorithm combining real-time traffic information for dynamic evaluation is adopted. Specifically, a heuristic algorithm (such as the legacy algorithm) is used to solve vehicle routing problems with time windows. The algorithm's evaluation function guides the search direction by minimizing the total travel cost. During the algorithm's iteration, real-time traffic information is also used to obtain the latest traffic speeds of each road segment, dynamically updating travel time, travel cost, and temperature control energy consumption cost. For example, the algorithm initially generates a route with the initial point, delivery point A, delivery point B, and delivery point C. However, during evaluation, it is found that the route from A to B is currently congested. At this time, an alternative route is immediately queried, such as the initial point, delivery point A, delivery point C, and delivery point B. Although the distance is slightly longer, the total travel time and temperature control energy consumption are reduced because the route is unobstructed. After calculating the total travel cost, the algorithm uses the alternative route as the final route. This method can plan reasonable routes for cold chain transport vehicles and enable the delivery plan to have anti-interference and real-time optimization capabilities to cope with traffic uncertainties.

[0057] The above describes a method for intelligent scheduling of cold chain transportation capacity based on demand forecasting in an embodiment of this application. The following describes a system for intelligent scheduling of cold chain transportation capacity based on demand forecasting in an embodiment of this application. Please refer to [link / reference]. Figure 2 One embodiment of the intelligent cold chain transportation capacity scheduling system based on demand forecasting in this application includes: The demand forecasting module, based on historical delivery data from multiple delivery points, predicts the future demand for various cold chain products at each delivery point within a preset scheduling period. The scheduling module generates a delivery scheduling plan based on the forecast results. The delivery scheduling plan includes: determining the set of target delivery points that need to be visited each day within a preset scheduling period based on the forecast results and a pre-created value assessment model. The value assessment model is used to evaluate the value generated by visiting delivery points on different dates; assigning multiple cold chain transport vehicles to the set of target delivery points that need to be visited on the corresponding dates based on transportation resource data; and creating a delivery route for each cold chain transport vehicle to reach its assigned delivery point. The loading optimization module generates vehicle loading plans for cold chain transport vehicles based on prediction results and delivery scheduling plans. The vehicle loading plans describe the combination, quantity, placement, and temperature zone affiliation of the goods loaded into the vehicles.

[0058] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0059] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0060] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for intelligent scheduling of cold chain transportation capacity based on demand forecasting, characterized in that, The method includes: Step S1: Based on historical delivery data from multiple delivery points, predict the future demand for various cold chain products at each delivery point within the preset scheduling period. Step S2: Based on the prediction results, generate a delivery scheduling plan. The delivery scheduling plan includes: determining the set of target delivery points that need to be visited each day within a preset scheduling period based on the prediction results and a pre-created value assessment model. The value assessment model is used to evaluate the value generated by visiting delivery points on different dates. Based on transportation resource data, assign multiple cold chain transport vehicles to the set of target delivery points that need to be visited on the corresponding dates. Create a delivery route for each cold chain transport vehicle to reach its assigned delivery point. Step S3: Based on the prediction results and delivery scheduling plan, generate a vehicle loading plan for cold chain transport vehicles. The vehicle loading plan is used to describe the combination, quantity, placement and temperature zone of the goods loaded into the vehicle.

2. The method according to claim 1, characterized in that, The process of creating a valuation model includes: For each delivery point and each candidate visit date, the overall value is calculated. The overall value is obtained by weighted summation of the following three value components: The cumulative predicted order value from the start date to the candidate access date will be used as the first value component; The cumulative estimated loss cost from the start date to the candidate access date is used as the second value component. The cumulative estimated loss cost includes order cancellation losses and cargo damage risk costs. The third value component is calculated based on regional synergy. This third value component is used to quantify the additional value that can be obtained by combining visits to the delivery point with its geographically adjacent delivery points on the same delivery route.

3. The method according to claim 1, characterized in that, The third value component is calculated based on regional synergy effects, including: Identify other delivery points to be evaluated that are located within a preset geographic area of ​​the delivery point on the candidate visit date; Predict the synergistic benefits of combining visits to a delivery point with at least one delivery point to be evaluated, compared to visiting them individually; The calculation of synergistic benefits takes into account the cost savings in transportation distance, the time window matching gain, and the increased cost due to the complexity of route merging.

4. The method according to claim 1, characterized in that, Step S3 includes: Candidate cargo identification sub-step: Based on the allocation results of the delivery points to be visited by the target cold chain transport vehicle, identify the set of candidate cargo to be loaded; Dynamic temperature zone division sub-step: Based on the volume ratio of goods at different temperature levels in the candidate goods set and the delivery route environment, dynamically plan the variable temperature zone layout of the vehicle's interior space. The variable temperature zone layout includes at least a non-fixed ratio division of the frozen zone and the refrigerated zone. Cargo placement optimization sub-step: Using an optimization algorithm based on thermodynamic simulation, combined with a variable temperature zone layout, the specific placement position of the final cargo combination selected from the candidate cargo set inside the vehicle is determined.

5. The method according to claim 4, characterized in that, The dynamic temperature zone division sub-steps include: Based on the candidate cargo set, the estimated total volume of cargo in each temperature zone is calculated. Based on the environmental prediction data of the delivery route, an optimization model is established with the goal of minimizing overall temperature control energy consumption and temperature fluctuation risk. The optimal solution is obtained to obtain the optimal spatial division of the frozen and refrigerated areas in the vehicle interior space.

6. The method according to claim 4, characterized in that, The sub-steps for optimizing goods placement include: Construct a simplified thermodynamic model of the interior space of the target cold chain transport vehicle; Input parameters include the initial temperature of the goods, packaging properties, vehicle insulation performance, route ambient temperature, and the expected door opening and unloading sequence. Perform transient heat conduction simulation to predict the temperature change curves of different locations inside the vehicle throughout the delivery process, and calculate the temperature stability score of each candidate placement location for various goods based on the temperature change curves. Based on the degree of temperature sensitivity and the temperature stability score, goods with high temperature sensitivity are prioritized for placement in the area with the highest temperature stability score for the corresponding goods.

7. The method according to claim 6, characterized in that, The goods placement optimization sub-step also includes: After placing the goods based on the stability rating, the unloading order is determined according to the delivery route. The pressure-resistant goods that need to be unloaded last are piled at the bottom of the front of the truck, and the goods that are unloaded first are placed on the top layer.

8. The method according to claim 1, characterized in that, Create delivery routes for each cold chain transport vehicle to its assigned delivery point, including: For each cold chain transport vehicle, the initial location and the location assigned to the delivery point are used as inputs to the route optimization problem; The goal is to minimize the total driving cost, which includes the cost of driving distance, the cost of driving time, and the energy consumption cost of temperature control due to the vehicle's cold engine operation. The imposed constraints include: the specified harvest time for each delivery point and the legal constraints on the driver's continuous driving time. The solutions are obtained using a heuristic algorithm that combines real-time traffic information for dynamic evaluation.

9. A cold chain transportation capacity intelligent scheduling system based on demand forecasting, used to implement the cold chain transportation capacity intelligent scheduling method based on demand forecasting as described in any one of claims 1-8, characterized in that, The system includes: The demand forecasting module, based on historical delivery data from multiple delivery points, predicts the future demand for various cold chain products at each delivery point within a preset scheduling period. The scheduling module generates a delivery scheduling plan based on the forecast results. The delivery scheduling plan includes: determining the set of target delivery points that need to be visited each day within a preset scheduling period based on the forecast results and a pre-created value assessment model. The value assessment model is used to evaluate the value generated by visiting delivery points on different dates; assigning multiple cold chain transport vehicles to the set of target delivery points that need to be visited on the corresponding dates based on transportation resource data; and creating a delivery route for each cold chain transport vehicle to reach its assigned delivery point. The loading optimization module generates vehicle loading plans for cold chain transport vehicles based on prediction results and delivery scheduling plans. The vehicle loading plans describe the combination, quantity, placement, and temperature zone affiliation of the goods loaded into the vehicles.

Images