A method for optimizing the cold chain material distribution path of pre-prepared meals
By constructing a Monte Carlo search tree and optimizing cold chain delivery routes through temperature simulation, the problems of disconnect between production line status and dispatch plans and multi-temperature zone cargo damage in cold chain delivery were solved, achieving low cargo damage and efficient cold chain material delivery.
Patent Information
- Application Number
- CN202611558929.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-09
- Publication Date
- 2026-11-17
AI Technical Summary
The existing cold chain distribution system suffers from problems such as the disconnect between production line status and dispatch plans, crude cost estimation for cargo damage in multiple temperature zones, static rigidity of route plans, and mismatch between loading and unloading sequences. These issues lead to vehicles waiting for cargo or cargo waiting for vehicles, dynamic temperature fluctuations, and increased cargo damage.
By acquiring production and order information, a Monte Carlo search tree is constructed to simulate and generate delivery routes and perform temperature simulations. The routes are dynamically adjusted to optimize vehicle tasks, and real-time data is used for replanning to ensure consistency in loading and unloading sequences and low cargo damage.
It achieves second-level synchronization between production line status and dispatch plan, reduces the accuracy of cargo damage cost estimation, provides dynamic replanning capability, solves the matching problem of loading and unloading sequence, and improves the on-time response rate of cold chain and vehicle full load rate.
Smart Images

Figure CN123117971A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of cold chain logistics, and in particular to a method for optimizing the distribution path of pre-prepared cold chain materials. Background Technology
[0002] In recent years, the prepared food industry has entered a stage of rapid development characterized by large-scale and standardized production. Its core supply chain links—factory packaging and palletizing, cold chain distribution, and final delivery—have placed extremely high demands on timeliness and quality assurance. Prepared food products are diverse, and different materials are highly sensitive to storage and transportation temperature windows, often requiring the mixing of multiple temperatures. At the same time, prepared food factories mostly employ automated production lines operating continuously. Fluctuations in production line pace and equipment alarms causing shutdowns directly affect the completion time of orders, thereby disrupting downstream delivery scheduling.
[0003] In cold chain urban delivery scenarios, vehicles are typically equipped with multi-temperature zone compartments, and three types of coupling challenges exist during the delivery process: Traditional scheduling relies on fixed plans for vehicle dispatch, which does not couple with the actual output rhythm of the packaging / palletizing production line in real time, often resulting in "vehicles waiting for goods" or "goods waiting for vehicles". Opening the door to unload cargo introduces multiple physical field couplings such as heat pulses, heat transfer through the partition, and heat transfer from the outer wall during driving, resulting in dynamic fluctuations in the temperature inside the compartment. Existing route optimization mainly focuses on "time windows and distances," and cargo damage costs are mostly estimated using static coefficients, which cannot reflect the actual overheating damage under specific routes. Frequent disturbances during delivery, such as sudden congestion, production line shutdowns, and temperature exceeding the limit, mean that traditional vehicle routing problem models, which solve the problem once and execute it in a fixed manner, lack the ability to dynamically replan based on real-time conditions. The loading and unloading sequences did not form a closed-loop reverse constraint, and there was a lack of pre-loading verification between the actual volume and the allocated quantity of the multi-temperature zones, resulting in frequent searching during the journey, increased door opening time, and aggravated cargo damage.
[0004] Therefore, we propose an optimization method for the cold chain material distribution path of pre-prepared dishes to solve the above problems. Summary of the Invention
[0005] This invention provides a method for optimizing the cold chain material distribution route of pre-prepared vegetables, which can solve problems such as the disconnect between production line status and dispatch plan, rough estimation of cargo damage costs in multiple temperature zones, static rigidity of route plan and mismatch between loading and unloading sequence in the existing cold chain distribution.
[0006] The first aspect of this invention provides a method for optimizing the cold chain material delivery route for pre-prepared dishes. The method includes: acquiring production information and order information; determining dispatch tasks based on vehicle constraints; constructing a search tree for the dispatch tasks, where each node records vehicle status, customer set, and temperature status of each temperature zone; simulating and generating a complete delivery route during the search process; performing temperature simulation on the complete delivery route to obtain cargo damage costs; selecting nodes based on the total cost including the cargo damage costs to obtain a preliminary delivery route plan; adjusting the preliminary delivery route plan by re-invoking the temperature simulation to update the total cost; and determining the delivery route plan based on the updated total cost; and if an abnormal state is detected during the execution of the delivery route plan, reconstructing the search tree with the current state as the root node to perform a search, obtaining and outputting the remaining delivery route plans.
[0007] Optionally, in a first implementation of the first aspect of the present invention, the method includes: The real-time operating data of the production equipment is obtained as the production information, and the estimated completion time is calculated by combining the order information. The set of orders to be delivered is determined based on the estimated completion time. Based on the set of orders to be delivered, and in conjunction with the vehicle constraints including vehicle load limits and temperature zone volume limits, orders are allocated to obtain a dispatch task that includes the allocated order set and the loading allocation for each temperature zone.
[0008] Optionally, in a second implementation of the first aspect of the present invention, constructing a search tree for the dispatch task includes: The initial root node of the Monte Carlo search tree is established with the distribution center as the initial node, and this root node is used as the search tree. Set expansion rules for the nodes of the search tree: filter candidate customers from the customer set, generate corresponding access child nodes for each candidate customer and establish parent-child node association, and update the vehicle status, the customer set and the temperature status of each temperature zone in the access child nodes.
[0009] Optionally, in a third implementation of the first aspect of the present invention, the method includes: The temperature simulation is specifically a multi-temperature zone heat exchange simulation. The temperature fluctuation in the complete delivery route is calculated through the multi-temperature zone heat exchange simulation to obtain the cumulative amount of over-temperature damage. The cumulative amount of overheating damage is converted into the cargo damage cost, and the cargo damage cost is weighted and summed with the vehicle fixed cost, driving cost and time penalty cost to obtain the total cost; The total cost is backtracked layer by layer along the search tree to update the access records of the corresponding nodes. After iteration, the leaf node with the lowest average cost is selected to restore the initial delivery route plan.
[0010] Optionally, in a fourth implementation of the first aspect of the present invention, the method includes: The preliminary delivery route plan is taken as the current optimal plan; Based on the results of the temperature simulation, customers in the customer set are ranked according to temperature risk, and target customers are selected to apply neighborhood actions to obtain a set of candidate delivery solutions. The neighborhood actions include swapping the access order, reversing the access order, or relocating customers between different vehicles. The temperature simulation is re-invoked for each candidate delivery plan in the candidate delivery plan set to obtain the updated total cost; The solution is iterated by comparing the updated total cost with the total cost of the current optimal solution, and the current optimal solution that meets the termination condition is determined as the delivery route solution.
[0011] Optionally, in a fifth implementation of the first aspect of the present invention, the step of reconstructing the search tree with the current state as the root node and performing a search to obtain and output the remaining delivery route solutions during the execution of the delivery route plan includes: Real-time acquisition of road condition information, actual temperature of each temperature zone, and equipment status signals; When the road network is congested based on the road condition information, the temperature exceeds the limit based on the actual temperature of each temperature zone, or the production line is shut down based on the equipment status signal, the abnormal state is confirmed to be detected. Collect the current status of the vehicle's location, the actual temperature of each temperature zone, and the remaining unvisited customers in the customer set, and reconstruct the Monte Carlo search tree using the current status as the root node; Adjust the search parameters according to the type of abnormal state and perform search iterations to generate remaining delivery route plans.
[0012] Optionally, in a sixth implementation of the first aspect of the present invention, it further includes: Based on the loading allocation of each temperature zone in the vehicle dispatch task, a temperature zone volume adjustment command is generated and sent to the vehicle isolation device to adjust the physical space of each temperature zone; The system receives volume matching status information after physical space adjustment. When loading requirements are met, it generates a loading sequence based on the delivery route plan, performs loading operations according to the loading sequence, and records the loading verification results.
[0013] Optionally, in a seventh implementation of the first aspect of the present invention, the step of filtering candidate customers from the customer set includes: Based on the distance between the vehicle's location in the vehicle status and the location of each unvisited customer in the customer set, combined with the expected driving speed and customer time window, the delivery time margin is calculated. The unvisited customers are sorted according to the remaining delivery time to obtain a priority sequence of candidate customers; Based on the candidate customer priority sequence and the remaining load in the vehicle status, the unvisited customers who meet the conditions are selected as candidate customers.
[0014] Beneficial effects: By directly using underlying production line micro-data such as real-time packaging speed of packaging machines, number of boxes packaged, palletizing robot cycle time, and alarm shutdown recovery time as upstream inputs for VRP, and dynamically calculating order fulfillment time by taking the later value of remaining packaging / palletizing time and correcting alarms, the static assumption of traditional delivery orders being ready is changed. This allows the dispatch task to be synchronized with the actual output cycle time of the automated production line at the second level, avoiding empty vehicle waiting or order backlog. At the same time, by using the fulfillment time to deduce the latest departure time, the vehicle load rate and cold chain response time rate are improved from the source. The search process naturally avoids high-risk paths involving frequent opening of doors and partitions, large temperature differences, and long-term exposure. The output is not simply the shortest distance path, but the cold chain feasible path with the best overall cost. The accuracy of cargo damage cost assessment is orders of magnitude higher than that of the static model. The customer temperature risk ranking table is obtained by using the simulation results of the preliminary path. Neighborhood actions are applied to high-risk customers in a targeted manner. The incremental simulation of the difference segment is used instead of the full path re-simulation. This overcomes the defects of blind perturbation and computational explosion in the traditional local search in multi-temperature zone VRP. It not only ensures the algorithm's ability to conduct targeted in-depth mining on high-temperature sensitive nodes, but also reduces the computation time of neighborhood evaluation to an engineering-application level. There is no need to re-solve from the distribution center. The remaining path with continuous physical state can be obtained in seconds with the computing power of the vehicle terminal. After the replanning result is smoothly compared with the original solution, the loading order adjustment instruction is issued in linkage, which solves the implementation problem of discontinuous breakpoints and conflicting loading and unloading logic in traditional rescheduling. The route planning results are directly converted into unloading reverse order and loading forward order instructions. Before departure, consistency is verified by adjusting the volume of the adjustable partition, scanning the loading process records and the temperature zone allocation. Only after consistency is verified can the departure release notice be obtained. Attached Figure Description
[0015] Figure 1 This is a schematic diagram of an embodiment of the method for optimizing the cold chain material delivery path of pre-prepared dishes in this invention. Detailed Implementation
[0016] This invention provides a method for optimizing the cold chain material distribution route for pre-prepared dishes, addressing problems in existing cold chain distribution such as the disconnect between production line status and dispatch plans, inaccurate cost estimation for damage in multiple temperature zones, static rigidity of route schemes, and mismatch between loading and unloading sequences. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings 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 can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus 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 apparatus.
[0017] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 One embodiment of the pre-prepared food cold chain material distribution path optimization method in this invention includes: 101. Obtain real-time production information of packaging machines and palletizing robots in the pre-prepared food packaging and palletizing process. Determine the orders to be delivered based on the loading confirmation signal and the order completion time. Obtain the dispatch task according to the vehicle load, volume constraints and loading capacity of each temperature zone. The dispatch task includes the vehicle number, the set of assigned orders, the loading allocation of each temperature zone and the estimated departure time.
[0018] It is understood that the executing entity of this invention can be a pre-prepared food cold chain material distribution path optimization device, or it can be a terminal or a server; the specific implementation is not limited here. This embodiment of the invention will be described using a server as an example.
[0019] Specifically, the process of acquiring real-time production information and obtaining dispatch tasks includes: real-time collection of the current packaging speed, number of boxes already packaged, cumulative production time of the vertical packaging machine, and the number of palletized trays completed and the current palletizing cycle of the palletizing robot; determining the remaining number of boxes to be packaged and the remaining number of boxes to be palletized based on order data; calculating the remaining packaging time based on the current packaging speed and the remaining number of boxes to be packaged; calculating the remaining palletizing time based on the current palletizing cycle and the remaining number of boxes to be palletized; and taking the later of the remaining packaging time and the remaining palletizing time for the same order as the estimated completion time for that order; when a production line alarm signal is collected, the estimated completion time is corrected according to the alarm type and the estimated recovery time to obtain the corrected estimated completion time. Based on the revised estimated packing time and the loading confirmation signal, orders that have completed all packaging and palletizing and have a loading confirmation signal are marked as orders to be delivered, and a set of orders to be delivered is output. Based on the number of boxes, weight per box, box size, and required temperature zone type of each order in the set of orders to be delivered, combined with the vehicle's maximum load capacity and the maximum volume of each temperature zone, multiple orders are combined and assigned to designated vehicles to obtain a dispatch task list. Each dispatch task in the dispatch task list includes the vehicle number, the set of orders assigned, and the loading allocation for each temperature zone. Based on the revised estimated packing time of the orders assigned to each dispatch task, the customer's required delivery time window, and the estimated travel time from the distribution center to the first customer point, the latest departure time of the dispatch task is determined and written as the estimated departure time into the dispatch task. Simultaneously, based on the customer location and delivery time window of the orders assigned to each dispatch task, the initial unloading sequence is determined, and the loading sequence prompt information is obtained in reverse order of the initial unloading sequence. The loading sequence prompt information is output together with the dispatch task. Among them, the corrected estimated assembly time is used to determine the set of orders to be delivered, the set of orders to be delivered is used to obtain the dispatch task list, the dispatch task list is used to construct the Monte Carlo search tree, the estimated departure time is used to calculate the time window penalty cost, and the loading sequence prompt information is used as a reference for the loading sequence when obtaining the delivery route plan.
[0020] 102. For dispatch tasks, construct a Monte Carlo search tree with the distribution center as the root node. Each node of the search tree records the current vehicle location, the set of customers visited, the remaining unvisited customers, the current time, the remaining load, and the estimated temperature status of each temperature zone.
[0021] Specifically, the process of constructing the Monte Carlo search tree includes: for each dispatch task, establishing a root node with the distribution center as the initial node, the root node recording the initial location of the vehicle, the set of all unvisited customers, the estimated departure time, the vehicle's full load capacity, and the initial set temperature of each temperature zone, and setting the number of visits and the initial value of the cumulative cost of the root node. Set child node expansion rules for the root node and each subsequent node: From the set of unvisited customers, filter out a candidate customer set based on the urgency of the customer's delivery time window and the distance between the customer's location and the current vehicle location. For each candidate customer in the candidate customer set, obtain a corresponding access child node. Each access child node records the current vehicle location, the set of visited customers, the set of remaining unvisited customers, the estimated arrival time, the remaining load capacity, and the estimated temperature status of each temperature zone after visiting that customer. If the remaining load capacity corresponding to the current node is insufficient to serve any customer in the candidate customer set, or if the candidate customer set is empty, obtain a child node to return to the distribution center. The child node to return to the distribution center records the status of the vehicle after returning to the distribution center, including the vehicle location, the set of visited customers, the set of remaining unvisited customers, the estimated return time, and the estimated temperature status of each temperature zone. Each child node is associated with its parent node, and an initial value for the number of visits and cumulative cost is set for each child node, resulting in a search tree structure for Monte Carlo search. The node state data and inter-node relationships in the search tree structure are used as the basis for state updates when selecting child nodes layer by layer from the root node, expanding new child nodes, and randomly completing the complete delivery path. Further, the process of filtering the candidate customer set includes: obtaining the material temperature zone type, required delivery time window, and customer location for each customer from the remaining unvisited customer set; determining the estimated arrival time of the vehicle for each customer based on the distance between the current vehicle location recorded by the current node and each customer's location, combined with the current time and the estimated driving speed; and obtaining the delivery time margin for each customer based on the time difference between the estimated arrival time and the boundary of the corresponding customer's required delivery time window. The remaining unvisited customers are sorted in ascending order of their remaining delivery time to obtain a priority sequence of candidate customers. In the priority sequence of candidate customers, customers with more urgent delivery times are ranked higher. The remaining loading capacity of the current vehicle is then checked in turn according to the priority sequence of candidate customers to determine whether it meets the needs of the corresponding customers. Customers whose loading needs are met are selected and added to the candidate customer set. The process stops when the number of customers in the candidate customer set reaches a set number. For each candidate customer in the candidate customer set, a corresponding access sub-node is obtained, and the sorting position of the candidate customer in the candidate customer priority sequence is used as the priority identifier of the access sub-node. The current vehicle location, the set of visited customers, the set of remaining unvisited customers, the estimated arrival time, the remaining load capacity, and the estimated temperature status of each temperature zone recorded in the access sub-node are all updated according to the status after visiting the candidate customer. The priority identifier is stored in the corresponding access sub-node. The priority identifier is used to obtain new sub-nodes in the order of priority identifiers when expanding sub-nodes, and the sub-node with the earlier priority identifier is visited first during the selection phase.
[0022] 103. During the Monte Carlo search tree process, candidate customers are gradually expanded from the current node to obtain branches. A complete delivery path is randomly completed during the simulation phase. Multi-temperature zone heat exchange simulation is performed on the complete delivery path to calculate the temperature fluctuations caused by heat transfer through the vehicle walls, heat pulses from door opening, and infiltration from adjacent temperature zones. The cumulative amount of overheating damage is obtained and converted into cargo damage cost. The cargo damage cost is weighted and summed with vehicle fixed cost, driving cost, refrigeration energy consumption cost, and time window penalty cost to obtain the total cost of the complete delivery path. The total cost is then propagated back along the search path layer by layer to update the number of visits and cumulative cost of each parent node. After repeated iterations, the leaf node with the minimum average cost is selected from the root node to reconstruct the initial delivery path scheme.
[0023] Specifically, the process of performing Monte Carlo search and heat exchange simulation includes: starting from the root node, calculating selection criteria based on the number of visits and cumulative costs recorded for each child node, prioritizing child nodes with lower average costs and fewer visits, and repeating the selection operation until the selected node meets the expansion conditions or becomes a leaf node, thus obtaining the currently selected node; expanding the currently selected node: when the currently selected node records a set of remaining unvisited customers and has not completed all customer visits, filtering the remaining unvisited customers based on the urgency of their time windows, their distance from the current vehicle location, and the remaining load capacity of the current vehicle to obtain a set of candidate customers, obtaining a child node for each candidate customer in the candidate customer set, and associating it with the currently selected node to obtain the expanded search tree; Select the latest obtained child node from the expanded search tree as the simulation starting point. Starting from the state recorded at the simulation starting point, randomly select customers from the set of remaining unvisited customers and add them to the current vehicle route. Update the remaining load and estimated arrival time after each customer is added. When the remaining load is insufficient to serve the next customer or the time window constraint does not allow continued service, arrange for the vehicle to return to the distribution center and activate the next vehicle until all customers have been visited, and obtain the complete delivery route. The complete delivery route includes the customer visit sequence of each vehicle, the estimated arrival time of each customer, the time when the vehicle returns to the distribution center, and the operating status of each temperature zone at each stage of the journey. Multi-temperature zone heat exchange simulation is performed on the complete delivery route: The driving and unloading processes of each vehicle in the complete delivery route are divided into multiple discrete time intervals according to a set time step. The heat transferred to each temperature zone through the vehicle wall, the heat transferred to the corresponding temperature zone by external hot air each time the door is opened for unloading, and the heat conducted between adjacent temperature zones through the partition are accumulated in each time interval. The current temperature of each temperature zone is updated according to the accumulated heat in each time interval, and the temperature change sequence of the temperature zone changes with the execution of the delivery route is obtained. The Arrhenius equation is introduced, and the actual temperature of the section in the temperature change sequence that exceeds the allowable temperature range of the pre-prepared food materials is substituted into the deterioration rate function based on the exponential law. The function is integrated or discretely accumulated over the corresponding duration to obtain the cumulative amount of overheat damage reflecting the exponential deterioration law. The cumulative amount of overheat damage is multiplied by the set unit cargo loss cost coefficient to obtain the cargo loss cost of the complete delivery route. The total cost of the complete delivery route is obtained by adding the cost of cargo damage, the fixed usage cost of the vehicle, the cost of driving distance, the cost of refrigeration energy consumption, and the time window penalty cost calculated based on the deviation between the expected arrival time of each customer and the customer's expected time window in the complete delivery route. The total cost is used as the simulation result and backtracked layer by layer along the parent-child relationship of the search tree, starting from the child node corresponding to the simulation starting point, updating the number of visits and the cumulative cost of each parent node. The updated number of visits and the cumulative cost are used as the basis for selecting child nodes in subsequent iterations. The above selection, expansion, simulation and backtracking process is repeated until the number of iterations reaches a preset threshold. The leaf node with the smallest ratio of cumulative cost to number of visits is selected from the child nodes of the root node. Based on the set of visited customers and vehicle status recorded in the leaf node, a preliminary delivery route plan is reconstructed. The preliminary delivery route plan includes the customer visit order, unloading order and expected arrival time of each vehicle. The preliminary delivery route plan is used as the input for subsequent neighborhood adjustment.
[0024] Furthermore, the process of performing multi-temperature zone heat exchange simulation includes: obtaining the customer access sequence of each vehicle and the estimated arrival time of each customer in the complete delivery route; dividing the delivery journey of each vehicle into multiple continuous task segments based on the unloading time of each customer and the number of times the truck door is opened during the unloading process; obtaining the task segment sequence of the vehicle; and updating the temperature state of each temperature zone in sequence according to the task segment sequence: for the driving segment, updating the temperature state of each temperature zone at the end of the driving segment based on the estimated driving time and the heat transfer parameters of the truck wall corresponding to the segment in the complete delivery route. For each unloading segment, the temperature state of the corresponding temperature zone at the end of the segment is updated based on the number of door openings, the duration of door openings, and the heat exchange parameters. Simultaneously, the heat infiltration between adjacent temperature zones is updated based on the temperature difference at the beginning of the segment, resulting in the temperature state of each temperature zone at the end of each task segment. The temperature states of each temperature zone at the end of each task segment for each vehicle are arranged chronologically to obtain a temperature change sequence that varies with the delivery route. Based on the temperature change sequence, time segments exceeding the allowable temperature range of the corresponding pre-prepared food materials in each temperature zone are extracted. The temperature variables and corresponding time variables within these segments are substituted into the degradation model according to the Arrhenius equation and accumulated to obtain the cumulative over-temperature damage for that temperature zone. The cumulative over-temperature damage for each temperature zone is multiplied by the unit loss cost coefficient corresponding to that temperature zone, and the weighted results of each temperature zone are summed to obtain the loss cost of the entire delivery route. This loss cost is then used to calculate the total cost of the entire delivery route.
[0025] 104. Perform neighborhood adjustments on the initial delivery route plan based on the customer access order, and re-invoke the multi-temperature zone heat exchange simulation for each adjusted candidate plan to update the cargo damage cost. Accept or reject the adjustment based on the criterion of minimizing the total cost, and determine the delivery route plan that includes the customer access order, unloading order, and loading order obtained by unloading in reverse order.
[0026] Specifically, the process of performing neighborhood adjustment and determining delivery route plan includes: obtaining a preliminary delivery route plan as the current optimal plan, setting the maximum number of iterations, initial acceptance probability and cooling-off coefficient for neighborhood search, and defining a neighborhood action set, which includes three actions: swapping the access order of two customers within the same vehicle route, reversing the access order of a segment of customers within the same vehicle route, and relocating customers between different vehicles. A multi-temperature zone heat exchange simulation is performed on the current optimal solution to obtain the temperature deviation record of each customer at the expected arrival time. The customers are sorted from high to low according to the degree of temperature deviation, resulting in a customer temperature risk ranking table. The customer temperature risk ranking table is used to prioritize customers with larger temperature deviations as neighborhood adjustment targets. Based on the customer temperature risk ranking table, a set number of customers with the largest temperature deviations are selected from the current optimal solution. Actions in the neighborhood action set are applied to each selected customer to obtain a candidate delivery solution set. Each candidate delivery solution in the candidate delivery solution set contains the adjusted customer access order and the corresponding unloading order. For each candidate delivery plan in the candidate delivery plan set, the multi-temperature zone heat exchange simulation is called again to obtain the cargo loss cost of each candidate delivery plan. Combined with the travel distance cost and time window penalty cost, the total cost of each candidate delivery plan is obtained, forming the candidate plan total cost evaluation result. The candidate delivery plan corresponding to the minimum value in the candidate plan total cost evaluation result is compared with the current optimal plan. If the total cost of the candidate delivery plan is lower than the total cost of the current optimal plan, the candidate delivery plan is updated to the current optimal plan. If the total cost of the candidate delivery plan is not lower than the total cost of the current optimal plan, the candidate delivery plan is accepted within the allowable range of the initial acceptance probability and the current optimal plan is updated. Otherwise, the current optimal plan is kept unchanged, and the acceptance probability is reduced according to the cooling coefficient to obtain the updated current optimal plan. Repeat the above comparison and update process until the maximum number of iterations is reached or the total cost of the current optimal solution no longer decreases after multiple consecutive rounds. The final current optimal solution is taken as the delivery route solution. Based on the unloading order in the delivery route solution, the loading order is obtained in reverse order of unloading. The delivery route solution and the loading order are output together. The delivery route solution is used for delivery execution and dynamic replanning, and the loading order is used to place the goods in reverse order of unloading when loading.
[0027] Furthermore, the process of re-invoking the multi-temperature zone heat exchange simulation for the candidate delivery scheme includes: obtaining the difference information between the candidate delivery scheme and the current optimal scheme, including the changed customer access order, the changed path segments, and the changed unloading order; based on the difference information, extracting the delivery process segments corresponding to the changed customer access order and path segments from the candidate delivery scheme to obtain the set of path segments to be simulated; and simultaneously extracting the temperature change sequence and overheat damage accumulation corresponding to the unchanged path segments from the simulation results corresponding to the current optimal scheme as retained simulation data. For each path segment to be simulated in the set of path segments to be simulated, multi-temperature zone heat exchange simulation is performed sequentially to obtain the updated temperature change sequence and the updated cumulative over-temperature damage for each path segment to be simulated. The updated temperature change sequence and the updated cumulative over-temperature damage are then merged with the retained simulation data in chronological order of delivery execution time to obtain the complete temperature change sequence and the complete cumulative over-temperature damage for the candidate delivery scheme. The cargo damage cost of the candidate delivery scheme is calculated based on this. The cargo damage cost is used as the basis for calculating the total cost of the candidate delivery scheme by combining the travel distance cost and the time window penalty cost.
[0028] 105. During the delivery process according to the delivery route plan, when road network congestion, temperature exceeding the allowable range, or shutdown of the front-end packaging and palletizing production line is detected, the Monte Carlo search tree is reconstructed with the vehicle's current location, the current actual temperature of each temperature zone, and the set of remaining unvisited customers as the new root node, and a search is performed to update the remaining delivery route plan and send the remaining delivery route plan to the vehicle terminal.
[0029] Specifically, the process of performing dynamic replanning includes: during the delivery process of vehicles according to the delivery route plan, real-time acquisition of road network congestion information, the current actual temperature of each temperature zone of the vehicle, and the equipment status signals of the front-end packaging and palletizing production line. When the road network congestion information shows that the estimated travel time of the current travel segment or the subsequent travel segment exceeds the set threshold, or when the current actual temperature of any temperature zone exceeds the allowable temperature range, or when the equipment status signal shows that the front-end packaging and palletizing production line has stopped and the stoppage will cause the orders that have not yet been dispatched to be delayed, the dynamic replanning conditions are determined to be triggered, and replanning trigger information including the triggering cause type, triggering time, and the set of affected customers is obtained. Based on the replanning trigger information, the system collects the vehicle's current location, current time, current actual temperature of each temperature zone, set of remaining unvisited customers, remaining load capacity, and the required delivery time window for each remaining customer. This data is then processed to obtain the initial replanning state data. The initial replanning state data also records the original access order of the remaining unexecuted parts of the delivery route plan, which is used as a reference benchmark for replanning. A Monte Carlo search tree is constructed using the initial replanning state data as the new root node. The trigger cause type in the replanning trigger information is used as the basis for adjusting the search parameters. When the trigger cause type is road network congestion, the estimated subsequent travel time is adjusted according to the additional travel time caused by congestion. When the trigger cause type is temperature exceeding the limit in a temperature zone, the weighting coefficient of cargo damage cost in the total cost calculation is increased. When the trigger cause is a front-end production line shutdown, the affected orders are temporarily removed from the remaining customer set and marked as delayed delivery. A Monte Carlo tree search is then performed, with the number of iterations set to a predetermined proportion of the Monte Carlo tree search iteration count. After the search, an updated remaining delivery route plan is output, including the remaining customer access order, estimated arrival time, and corresponding temperature change predictions for each temperature zone. The updated remaining delivery route plan is compared with the original access order of the remaining unexecuted parts of the delivery route plan to determine if the remaining customer access order has changed. If it has changed, the departure status of the corresponding vehicles is further determined: for vehicles that have not yet departed, an updated loading order adjustment instruction is obtained based on the unloading order in the updated remaining delivery route plan, and this instruction is merged with the updated remaining delivery route plan; for vehicles already en route, only the remaining delivery route and unloading prompts are updated, without generating a loading order adjustment instruction, thus merging to obtain a replanning execution plan to be issued. If the order of remaining customer visits remains unchanged, the updated remaining delivery route plan will be used as the replanning execution plan to be issued. The replanning execution plan to be issued will be sent to the vehicle terminal, and the estimated arrival time of each customer in the replanning execution plan to be issued will be sent back to the distribution center dispatch terminal for the distribution center to confirm the receiving arrangements and complete the dynamic update of the delivery route.
[0030] 106. After determining the delivery route plan and before executing the delivery, the following steps are also included: Based on the loading allocation of each temperature zone corresponding to each vehicle in the delivery route plan, a temperature zone volume adjustment instruction is obtained. The temperature zone volume adjustment instruction includes the target volume that each temperature zone needs to be adjusted to. The temperature zone volume adjustment instruction is sent to the vehicle adjustable partition control device to adjust the physical space of each temperature zone of the vehicle. The actual adjusted volume of each temperature zone is received from the adjustable partition control device. The actual adjusted volume of each temperature zone is compared with the corresponding temperature zone loading allocation recorded in the vehicle dispatch task to obtain volume matching status information. The volume matching status information includes the determination result of whether each temperature zone meets the loading requirements. When all temperature zones in the volume matching status information meet the loading requirements, a loading condition ready signal is obtained based on the volume matching status information. Based on the loading condition ready signal, the loading operation is carried out according to the loading sequence in the delivery route plan. During the loading process, the actual loading temperature zone, loading position, and loading time of each box are collected simultaneously to obtain the loading process record. Based on the loading process record, the actual number of boxes loaded in each temperature zone is accumulated. The actual number of boxes loaded in each temperature zone is checked against the loading allocation quantity of that temperature zone in the dispatch task to obtain the loading verification result. The loading verification result includes the verification conclusion of whether the loading quantity in each temperature zone is consistent. After all temperature zones are checked to be consistent, the vehicle departure release notice is obtained. The vehicle departure release notice is used to start the delivery task.
[0031] Furthermore, the process of obtaining the loading verification results includes: grouping the loading process records according to the temperature zone type to obtain the packing record set corresponding to each temperature zone; for each temperature zone's packing record set, removing records that were abnormally removed from the carriage during the loading process and not reloaded, to obtain the valid loading record set for that temperature zone; and obtaining the actual number of boxes loaded in that temperature zone based on the number of record entries in the valid loading record set. The actual number of boxes loaded in each temperature zone is compared with the loading allocation for that temperature zone in the dispatch task to obtain a loading verification result that includes the actual number of boxes loaded in each temperature zone, the loading allocation, and the difference between the two. When the difference between the actual number of boxes loaded and the loading allocation for all temperature zones in the loading verification result is zero, a vehicle dispatch release notice is obtained, and the loading verification result is sent to the delivery dispatch terminal for the delivery dispatch terminal to confirm that the vehicle has completed loading and then execute the delivery task.
[0032] The present invention also provides a pre-prepared food cold chain material distribution path optimization device, the pre-prepared food cold chain material distribution path optimization device includes a memory and a processor, the memory stores computer-readable instructions, when the computer-readable instructions are executed by the processor, the processor performs the steps of the pre-prepared food cold chain material distribution path optimization method in the above embodiments.
[0033] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the steps of the pre-prepared food cold chain material distribution path optimization method.
[0034] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.
[0035] 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 the present invention, 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 the present invention. 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.
[0036] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention 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 the present invention.
Claims
1. A method for optimizing the cold chain material distribution path of pre-prepared meals, characterized in that, Includes the following steps: Obtain production and order information, and determine vehicle dispatch tasks based on vehicle constraints; A search tree is constructed for the vehicle dispatch task, and the nodes of the search tree record the vehicle status, customer set, and temperature status of each temperature zone; During the search process of the search tree, a complete delivery route is simulated and generated. Temperature simulation is performed on the complete delivery route to obtain the cost of cargo damage. The node is selected based on the total cost including the cost of cargo damage to obtain a preliminary delivery route plan. The initial delivery route plan is adjusted, the total cost is updated by re-invoking the temperature simulation, and the delivery route plan is determined based on the updated total cost. If an abnormal state is detected during the execution of the delivery route plan, the search tree is reconstructed with the current state as the root node to perform a search, obtain the remaining delivery route plan, and output it.
2. The pre-prepared meal cold chain material distribution path optimization method according to claim 1, characterized in that, include: The real-time operating data of the production equipment is obtained as the production information, and the estimated completion time is calculated by combining the order information. The set of orders to be delivered is determined based on the estimated completion time. Based on the set of orders to be delivered, and in conjunction with the vehicle constraints including vehicle load limits and temperature zone volume limits, orders are allocated to obtain a dispatch task that includes the allocated order set and the loading allocation for each temperature zone.
3. The pre-prepared meal cold chain material delivery route optimization method of claim 1, wherein, The construction of a search tree for the dispatch task includes: The initial root node of the Monte Carlo search tree is established with the distribution center as the initial node, and this root node is used as the search tree. Set expansion rules for the nodes of the search tree: filter candidate customers from the customer set, generate corresponding access child nodes for each candidate customer and establish parent-child node association, and update the vehicle status, the customer set and the temperature status of each temperature zone in the access child nodes.
4. The pre-prepared meal cold chain material delivery route optimization method of claim 1, wherein, include: The temperature simulation is specifically a multi-temperature zone heat exchange simulation. The temperature fluctuation in the complete delivery route is calculated through the multi-temperature zone heat exchange simulation to obtain the cumulative amount of over-temperature damage. The cumulative amount of overheating damage is converted into the cargo damage cost, and the cargo damage cost is weighted and summed with the vehicle fixed cost, driving cost and time penalty cost to obtain the total cost; The total cost is backtracked layer by layer along the search tree to update the access records of the corresponding nodes. After iteration, the leaf node with the lowest average cost is selected to restore the initial delivery route plan.
5. The pre-prepared meal cold chain material delivery route optimization method of claim 1, wherein, include: The preliminary delivery route plan is taken as the current optimal plan; Based on the results of the temperature simulation, customers in the customer set are ranked according to temperature risk, and target customers are selected to apply neighborhood actions to obtain a set of candidate delivery solutions. The neighborhood actions include swapping the access order, reversing the access order, or relocating customers between different vehicles. The temperature simulation is re-invoked for each candidate delivery plan in the candidate delivery plan set to obtain the updated total cost; The solution is iterated by comparing the updated total cost with the total cost of the current optimal solution, and the current optimal solution that meets the termination condition is determined as the delivery route solution.
6. The pre-prepared meal cold chain material delivery route optimization method of claim 1, wherein, If an abnormal state is detected during the execution of the delivery route plan, the search tree is reconstructed with the current state as the root node to perform a search, obtain the remaining delivery route plans, and output them, including: Real-time acquisition of road condition information, actual temperature of each temperature zone, and equipment status signals; When the road network is congested based on the road condition information, the temperature exceeds the limit based on the actual temperature of each temperature zone, or the production line is shut down based on the equipment status signal, the abnormal state is confirmed to be detected. Collect the current status of the vehicle's location, the actual temperature of each temperature zone, and the remaining unvisited customers in the customer set, and reconstruct the Monte Carlo search tree using the current status as the root node; Adjust the search parameters according to the type of abnormal state and perform search iterations to generate remaining delivery route plans.
7. The pre-prepared meal cold chain material delivery route optimization method of claim 2, wherein, Also includes: Based on the loading allocation of each temperature zone in the vehicle dispatch task, a temperature zone volume adjustment command is generated and sent to the vehicle isolation device to adjust the physical space of each temperature zone; The system receives volume matching status information after physical space adjustment. When loading requirements are met, it generates a loading sequence based on the delivery route plan, performs loading operations according to the loading sequence, and records the loading verification results.
8. The pre-prepared meal cold chain material delivery route optimization method of claim 3, wherein, The step of filtering candidate customers from the customer set includes: Based on the distance between the vehicle's location in the vehicle status and the location of each unvisited customer in the customer set, combined with the expected driving speed and customer time window, the delivery time margin is calculated. The unvisited customers are sorted according to the remaining delivery time to obtain a priority sequence of candidate customers; Based on the candidate customer priority sequence and the remaining load in the vehicle status, the unvisited customers who meet the conditions are selected as candidate customers.
