Operation scheduling optimization method for railway cold chain combined transportation distribution center

Abstract

The invention provides a railway cold chain combined transportation distribution center operation scheduling optimization method, and belongs to the technical field of railway cold chain scheduling. Comprising the following steps: S1, resource pre-allocation and time window constraint embedding; s2, dynamic and static priority collaborative decision making; s3, optimizing a spatial constraint guided path; s4, collaboratively executing the operation process; and S5, performing real-time monitoring and feedback. Through triple technical breakthrough of multi-agent collaborative decision-making, dynamic and static priority coupling and space-time resource pre-allocation, the industrial pain points of unbalanced resource allocation, frequent path conflicts and high goods damage rate of the railway cold chain collecting and distributing center are solved, collaborative optimization of efficiency, quality and cost is realized, and the method is suitable for popularization and application. And a core technical support is provided for intelligent upgrading of cold chain multimodal transport.

Description

Technical Field

[0001] This invention provides an optimization method for the operation scheduling of railway cold chain intermodal transport distribution centers, belonging to the field of railway cold chain scheduling technology. Background Art

[0002] Existing scheduling methods employ a static allocation strategy, applying a uniform processing standard to all goods and failing to identify the differentiated needs of various cold chain products. This results in high-value goods receiving the same treatment as ordinary goods, a lack of targeted allocation of key resources, and limited overall operational efficiency. There are significant deficiencies in equipment collaboration. Gantry cranes, container trucks, forklifts, and other equipment lack effective coordination when operating together, frequently resulting in equipment idleness and mutual waiting, leading to low equipment utilization efficiency. Route planning technology is relatively outdated. Fixed route design ignores the movement restrictions and spatial constraints of track equipment, easily causing channel conflicts during equipment operation and affecting operational continuity. Dynamic adjustment capabilities are weak. Faced with the stringent time requirements and unforeseen circumstances of railway transportation, existing technology lacks a real-time response mechanism, making it difficult to guarantee the continuity of cold chain temperature control. Cargo loss control measures are limited. The lack of a priority ranking mechanism based on spoilage risk leads to perishable products being exposed to unsuitable environments for extended periods, causing economic losses. Multimodal transport connections are not smooth enough. Insufficient coordination when switching between rail, road, and water transport modes makes it difficult to match the operational sequences of each link, affecting service quality. Summary of the Invention

[0003] The goal of this invention is to provide a scheduling optimization method that integrates dynamic and static priority assessment, resource pre-configuration, and shortest path planning, so as to comprehensively improve the operational efficiency of railway cold chain intermodal transport distribution centers, reduce cargo loss, optimize resource utilization, and provide more reliable technical support for cold chain logistics.

[0004] This invention provides a method for optimizing the operation scheduling of railway cold chain intermodal transport distribution centers. This method establishes a multi-level scheduling decision-making system to achieve intelligent optimization management of the cold chain cargo handling process. The specific solution is as follows:

[0005] S1. Resource pre-allocation and time window constraint embedding

[0006] The pre-allocation of loading and unloading resources is based on the train timetable to lock in the loading and unloading lines, rail gantry cranes and cold storage racks in advance. The pre-allocation window is no less than 2 hours before the train arrives at the station. If the train is delayed, priority reassessment is triggered in real time, and resources are dynamically allocated according to the temperature control status of the goods. The pre-matching of platform resources refers to the allocation of target platforms 1 hour before the arrival of refrigerated trucks, and the establishment of forklift-rack-platform resource chain 30 minutes before the goods enter and leave the warehouse.

[0007] S2. Dynamic and Static Priority Collaborative Decision-Making

[0008] First, set the static priority of the goods, and classify them into three levels according to their value, timeliness and perishability: Category A: high value perishable; Category B: conventional cold chain; Category C: temperature-controlled storage. Category A goods are assigned the highest fixed priority. Then, adjust the dynamic priority, that is, monitor the loss rate of goods in real time, and automatically increase the priority when the loss rate exceeds the threshold.

[0009] In the inbound process, cold storage resources are allocated according to static priority, while in the outbound process, scheduling instructions are generated according to dynamic priority.

[0010] S3. Spatial Constraint-Oriented Path Optimization

[0011] Based on the strip-shaped operating range of the rail gantry crane, non-intersecting paths are planned for container trucks to avoid equipment waiting conflicts; a joint path network of forklifts, container trucks, and gantry cranes is constructed, and the shortest movement path between nodes is calculated in real time using the Dijkstra algorithm; dynamic avoidance logic is set up for unloaded equipment to avoid fully loaded equipment and for low-priority task equipment to avoid high-priority task equipment.

[0012] S4. Collaborative Execution of Work Processes

[0013] (1) Cold storage operations

[0014] Inbound stage: Receive pre-arrival information → allocate shelves according to static priority → forklifts transport goods to the temporary storage area along the shortest path → put goods on shelves after quality inspection;

[0015] Outbound stage: Dynamic priority triggers orders → Goods with the highest loss rate are removed from shelves first → Forklifts transfer goods to the loading platform via optimized routes.

[0016] (2) Park transfer operations

[0017] Train arrival: Pre-locked loading and unloading line → Gantry crane unloads containers according to static priority sequence → Trucks transfer containers via non-intersecting paths;

[0018] Multimodal transport conversion: Rail to road uses the "platform direct loading" mode, that is, container rail → truck chassis directly; goods that need to be distributed are transferred to cold storage for unpacking and sorting, reducing the exposure time throughout the process.

[0019] S5. Real-time monitoring and feedback

[0020] Deploy a cargo loss rate monitoring system to collect temperature and humidity data in real time, update dynamic priorities, and establish a resource conflict early warning mechanism. When the equipment waiting time exceeds the threshold, trigger path replanning.

[0021] Furthermore, S1 employs multi-agent collaborative rule generation technology;

[0022] Eight types of autonomous intelligent agents (trains, carriages, containers, trucks, rail gantry cranes, forklifts, pallets, and refrigerated trucks) are constructed, each with independent decision-making logic and environmental interaction capabilities. Through real-time state transitions and coordination rules of intelligent agents, autonomous decision-making and conflict resolution in the work process are achieved.

[0023] Based on a multi-agent operational logic architecture, a collaborative simulation system for the operational processes of a railway cold chain intermodal transport distribution center is constructed using the process modeling and track modeling components of a simulation modeling platform. This system covers static infrastructure such as platform facilities, rack structures, yard areas, and track lines, as well as dynamic transportation facilities such as forklifts, cold chain trains, refrigerated vehicles, gantry cranes, and container trucks.

[0024] Multi-agent collaborative rule generation mechanisms include:

[0025] (1) Optimization of intelligent agent collaboration rules for refrigerated trucks: Set the delayed arrival parameter in the refrigerated truck generation module, and realize the advance matching and allocation of refrigerated trucks and platform resources through the queue management module to avoid resource competition conflicts after arrival.

[0026] (2) Platform resource agent collaboration rules: In the platform resource acquisition module, set task priority parameters according to the cargo category attributes, configure the "resource exclusive and uninterrupted" task preemption strategy to ensure that the forklift agent prioritizes the processing of high-priority refrigerated cargo.

[0027] (3) Collaborative rules for loss monitoring agents: Design a cargo loss rate function module, configure loss rate monitoring attributes for each refrigerated cargo agent entering the system, assign initial loss rate values ​​according to probability distribution, and set a custom resource selection algorithm in the cargo removal operation module to realize dynamic priority outbound processing of cargo with the highest loss rate.

[0028] (4) Path planning agent collaboration rules: Set up a transfer and temporary storage area and a dedicated forklift lane between the internal and external areas of the cold storage room, construct a forklift path network topology, and add a resource selection and allocation algorithm to the collaboration logic, which includes the complete process of resource acquisition, movement execution and resource release.

[0029] (5) Train intelligent agent collaboration rules: A delayed arrival mechanism is set in the cold chain train generation module, and the pre-allocation and matching of cold chain trains and loading and unloading line resources is realized through the queue management module.

[0030] (6) Equipment cluster collaboration rules: Set up a refrigerated container temporary storage yard area in the park, and integrate the intelligent agents of equipment such as container front cranes, container trucks, and gantry cranes into the same path network topology. By modifying the train operation logic process, collaborative operation between equipment can be achieved.

[0031] Furthermore, S2 employs dynamic and static priority evaluation techniques;

[0032] In response to the characteristics of cold chain goods, a dual-track priority mechanism of "static grading + dynamic adjustment" will be established:

[0033] Static priority is divided into three levels based on the value and timeliness of goods. Level A: High-value perishable goods; Level B: Conventional cold chain; Level C: High-temperature resistant goods with pre-set differentiated processing timelines. Dynamic priority is based on real-time loss rate monitoring values, which are then dynamically adjusted. (Aiying Rong) [3] The Arrhenius equation, a chemical reaction temperature dependence formula, is used to determine the loss rate of goods to describe the effect of temperature on product quality degradation, as shown in equation (1).

[0034]

[0035] Where: k—the loss rate of the goods;

[0036] k0—a constant initial loss rate parameter;

[0037] E a —Experimental activation energy;

[0038] R—gas constant;

[0039] T—Absolute temperature;

[0040] Cold chain goods arrive at cold chain distribution centers via refrigerated containers or refrigerated transport vehicles, where they undergo storage, transfer, and other operations. To ensure the quality of goods to the greatest extent possible, manual control is typically used to manage the temperature and gas conditions of the environment in which the goods are located, ensuring that the goods are maintained at a suitable low temperature throughout the entire process. Under these conditions, the loss rate remains constant, but the actual amount of loss will be adjusted accordingly based on changes in the existing inventory. Thus, the loss model is given by equation (2), where q′ is the quantity of goods, t is time, and q0 is the initial quantity of products.

[0041] q′=q0-q0·e -kt (2)

[0042] When the monitored value exceeds the threshold, the priority is automatically upgraded and an emergency dispatch plan is triggered. This can effectively reduce the risk of high-value goods being stranded.

[0043] Furthermore, S1 employs a spatiotemporal constrained resource pre-allocation technique;

[0044] To address the time rigidity and spatial constraints of railway cold chain logistics, a resource pre-allocation model is designed:

[0045] (1) Pre-lock loading and unloading lines, gantry cranes and cold storage racks before the train arrives at the station;

[0046] (2) Target platforms are assigned to refrigerated trucks before they arrive at the station;

[0047] (3) Match the forklift-shelf-platform resource chain in advance before the inbound and outbound operations.

[0048] (4) If the train is delayed, priority will be reordered. That is, high-risk cargo will be loaded and unloaded first, and low-priority tasks will be postponed.

[0049] Furthermore, S3 employs conflict avoidance path planning technology;

[0050] Construct a conflict-free mobile path network based on the shortest path algorithm and spatial constraint rules:

[0051] Plan the shortest movement path for forklifts, container trucks, and gantry cranes to reduce unnecessary travel distances and increase forklift movement speed; preset non-intersecting paths for gantry cranes and container trucks; enforce a three-level avoidance mechanism of "empty loads yielding to full loads," "low priority yielding to high priority," and "later-arriving equipment yielding to earlier-arriving equipment"; set up "temporary storage areas" in cold storage to shorten transfer distances.

[0052] This invention solves the industry pain points of unbalanced resource allocation, frequent route conflicts, and high cargo damage rates in railway cold chain distribution centers through a triple technological breakthrough of "multi-agent collaborative decision-making + dynamic and static priority coupling + spatiotemporal resource pre-allocation". It achieves synergistic optimization of efficiency, quality, and cost, and provides core technical support for the intelligent upgrading of cold chain multimodal transport. Attached Figure Description

[0053] Figure 1 This is a flowchart of the present invention;

[0054] Figure 2 This is a logic diagram of the cold chain train operation process under the scheduling algorithm of the present invention;

[0055] Figure 3 This is a simulation diagram for an example embodiment;

[0056] Figure 4 This example compares the container handling capacity of different operating procedures. Detailed Implementation

[0057] This invention provides a method for optimizing the operation scheduling of railway cold chain intermodal transport distribution centers, such as... Figure 1 As shown, the specific solution is as follows:

[0058] S1. Resource pre-allocation and time window constraint embedding

[0059] The pre-allocation of loading and unloading resources is based on the train timetable to lock in the loading and unloading lines, rail gantry cranes and cold storage racks in advance. The pre-allocation window is no less than 2 hours before the train arrives at the station. If the train is delayed, priority reassessment is triggered in real time, and resources are dynamically allocated according to the temperature control status of the goods. The pre-matching of platform resources refers to the allocation of target platforms 1 hour before the arrival of refrigerated trucks, and the establishment of forklift-rack-platform resource chain 30 minutes before the goods enter and leave the warehouse.

[0060] S2. Dynamic and Static Priority Collaborative Decision-Making

[0061] First, set the static priority of the goods, and classify them into three levels according to their value, timeliness and perishability: Category A: high value perishable; Category B: conventional cold chain; Category C: temperature-controlled storage. Category A goods are assigned the highest fixed priority. Then, adjust the dynamic priority, that is, monitor the loss rate of goods in real time, and automatically increase the priority when the loss rate exceeds the threshold.

[0062] In the inbound process, cold storage resources are allocated according to static priority, while in the outbound process, scheduling instructions are generated according to dynamic priority.

[0063] S3. Spatial Constraint-Oriented Path Optimization

[0064] Based on the strip-shaped operating range of the rail gantry crane, non-intersecting paths are planned for container trucks to avoid equipment waiting conflicts; a joint path network of forklifts, container trucks, and gantry cranes is constructed, and the shortest movement path between nodes is calculated in real time using the Dijkstra algorithm; dynamic avoidance logic is set up for unloaded equipment to avoid fully loaded equipment and for low-priority task equipment to avoid high-priority task equipment.

[0065] S4. Collaborative Execution of Work Processes

[0066] (1) Cold storage operations

[0067] Inbound stage: Receive pre-arrival information → allocate shelves according to static priority → forklifts transport goods to the temporary storage area along the shortest path → put goods on shelves after quality inspection;

[0068] Outbound stage: Dynamic priority triggers orders → Goods with the highest loss rate are removed from shelves first → Forklifts transfer goods to the loading platform via optimized routes.

[0069] (2) Park transfer operations

[0070] Train arrival: Pre-locked loading and unloading line → Gantry crane unloads containers according to static priority sequence → Trucks transfer containers via non-intersecting paths;

[0071] Multimodal transport conversion: Rail to road uses the "platform direct loading" mode, that is, container rail → truck chassis directly; goods that need to be distributed are transferred to cold storage for unpacking and sorting, reducing the exposure time throughout the process.

[0072] S5. Real-time monitoring and feedback

[0073] Deploy a cargo loss rate monitoring system to collect temperature and humidity data in real time, update dynamic priorities, and establish a resource conflict early warning mechanism. When the equipment waiting time exceeds the threshold, trigger path replanning.

[0074] This invention optimizes the operational processes of railway cold chain distribution centers by driving dynamic scheduling through a multi-agent collaborative decision-making framework, combining dynamic and static priority evaluation, spatiotemporal resource allocation with conflict avoidance path planning. Specific technologies are as follows:

[0075] 1. S1 employs multi-agent collaborative rule generation technology;

[0076] As shown in Table 1, eight types of autonomous intelligent agents are constructed (train, carriage, container, truck, rail gantry crane, forklift, pallet, and refrigerated truck), each with independent decision-making logic and environmental interaction capabilities. Through real-time state transitions and coordination rules of intelligent agents, autonomous decision-making and conflict resolution in the work process are achieved.

[0077] Table 1 Categories of Autonomous Agents

[0078]

[0079]

[0080] Based on a multi-agent operational logic architecture, a collaborative simulation system for the operational processes of a railway cold chain intermodal transport distribution center is constructed using the process modeling and track modeling components of a simulation modeling platform. This system covers static infrastructure such as platform facilities, rack structures, yard areas, and track lines, as well as dynamic transportation facilities such as forklifts, cold chain trains, refrigerated vehicles, gantry cranes, and container trucks.

[0081] Under the traditional scheduling algorithm framework, the coordination rules among various intelligent agents have the following technical defects: the cold chain train intelligent agent executes the operation instructions according to a fixed time sequence and lacks a dynamic interaction mechanism with other intelligent agents; the forklift intelligent agent adopts a simple "first-come, first-served" scheduling rule and cannot identify differences in cargo priority; the platform resource allocation adopts a static preset mode and cannot be dynamically adjusted according to the real-time operation load.

[0082] To address the technical shortcomings of traditional collaborative rules, this invention designs the following novel multi-agent collaborative rule generation mechanism, such as... Figure 2 :

[0083] (1) Optimization of intelligent agent collaboration rules for refrigerated trucks: Set the delayed arrival parameter in the refrigerated truck generation module, and realize the advance matching and allocation of refrigerated trucks and platform resources through the queue management module to avoid resource competition conflicts after arrival.

[0084] (2) Platform resource agent collaboration rules: In the platform resource acquisition module, set task priority parameters according to the cargo category attributes, configure the "resource exclusive and uninterrupted" task preemption strategy to ensure that the forklift agent prioritizes the processing of high-priority refrigerated cargo.

[0085] (3) Collaborative rules for loss monitoring agents: Design a cargo loss rate function module, configure loss rate monitoring attributes for each refrigerated cargo agent entering the system, assign initial loss rate values ​​according to probability distribution, and set a custom resource selection algorithm in the cargo removal operation module to realize dynamic priority outbound processing of cargo with the highest loss rate.

[0086] (4) Path planning agent collaboration rules: Set up a transfer and temporary storage area and a dedicated forklift lane between the internal and external areas of the cold storage room, construct a forklift path network topology, and add a resource selection and allocation algorithm to the collaboration logic, which includes the complete process of resource acquisition, movement execution and resource release.

[0087] (5) Train intelligent agent collaboration rules: A delayed arrival mechanism is set in the cold chain train generation module, and the pre-allocation and matching of cold chain trains and loading and unloading line resources is realized through the queue management module.

[0088] (6) Equipment cluster collaboration rules: Set up a refrigerated container temporary storage yard area in the park, and integrate the intelligent agents of equipment such as container front cranes, container trucks, and gantry cranes into the same path network topology. By modifying the train operation logic process, collaborative operation between equipment can be achieved.

[0089] 2. S2 employs dynamic and static priority evaluation techniques;

[0090] In response to the characteristics of cold chain goods, a dual-track priority mechanism of "static grading + adjustment" will be established:

[0091] Static priority is divided into three levels based on the value and timeliness of goods. Level A: High-value perishable goods; Level B: Conventional cold chain; Level C: High-temperature resistant goods with pre-set differentiated processing timelines. Dynamic priority is based on real-time loss rate monitoring values, which are then dynamically adjusted. (Aiying Rong) [3] The Arrhenius equation, a chemical reaction temperature dependence formula, is used to determine the loss rate of goods to describe the effect of temperature on product quality degradation, as shown in equation (1).

[0092]

[0093] Where: k—the loss rate of the goods;

[0094] k0—a constant initial loss rate parameter;

[0095] E a —Experimental activation energy;

[0096] R—gas constant;

[0097] T—Absolute temperature;

[0098] Cold chain goods arrive at cold chain distribution centers via refrigerated containers or refrigerated transport vehicles, where they undergo storage, transfer, and other operations. To ensure the quality of goods to the greatest extent possible, manual control is typically used to manage the temperature and gas conditions of the environment in which the goods are located, ensuring that the goods are maintained at a suitable low temperature throughout the entire process. Under these conditions, the loss rate remains constant, but the actual amount of loss will be adjusted accordingly based on changes in the existing inventory. Thus, the loss model is given by equation (2), where q′ is the quantity of goods, t is time, and q0 is the initial quantity of products.

[0099] q′=q0-q0·e -kt (2)

[0100] When the monitored value exceeds the threshold, the priority is automatically upgraded and an emergency dispatch plan is triggered. This can effectively reduce the risk of high-value goods being stranded.

[0101] 3. S1 employs spatiotemporal constrained resource pre-allocation technology;

[0102] To address the time rigidity and spatial constraints of railway cold chain logistics, a resource pre-allocation model is designed:

[0103] (1) Pre-lock loading and unloading lines, gantry cranes and cold storage racks before the train arrives at the station;

[0104] (2) Target platforms are assigned to refrigerated trucks before they arrive at the station;

[0105] (3) Match the forklift-shelf-platform resource chain in advance before the inbound and outbound operations.

[0106] (4) If the train is delayed, priority will be reordered. That is, high-risk cargo will be loaded and unloaded first, and low-priority tasks will be postponed.

[0107] 4. S3 employs conflict avoidance path planning technology.

[0108] Construct a conflict-free mobile path network based on the shortest path algorithm and spatial constraint rules:

[0109] Plan the shortest movement path for forklifts, container trucks, and gantry cranes to reduce unnecessary travel distances and increase forklift movement speed; preset non-intersecting paths for gantry cranes and container trucks; enforce a three-level avoidance mechanism of "empty loads yielding to full loads," "low priority yielding to high priority," and "later-arriving equipment yielding to earlier-arriving equipment"; set up "temporary storage areas" in cold storage to shorten transfer distances.

[0110] To better illustrate the technical solution of the present invention, the present invention will be described in detail below with reference to specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.

[0111] This invention will describe in detail the technical implementation of the intelligent dispatching system for railway cold chain intermodal transport distribution centers through the following comprehensive application examples. Taking an international cold chain distribution center as an example, it will demonstrate the synergistic application effect and implementation details of the aforementioned four key technologies.

[0112] The international cold chain company operates a cold chain intermodal distribution center covering an area of ​​305,000 square meters with a storage capacity of up to 227,000 tons. It has completed the construction of three cold storage facilities, a refrigerated container yard, and a supporting exhibition building.

[0113] The distribution center adopts a three-dimensional design concept, with the cold storage building having an eight-story structure above ground and one floor below ground. It features an ultra-low temperature storage area of ​​-23℃ and a conventional cold storage area of ​​0℃, and optimizes the use of storage space through an "I"-shaped through-type design. The platform operation area on the first floor of the cold storage has 24 loading and unloading platforms on the north and south sides respectively, while the storage functional area includes four large-scale cold storage warehouses, each with an area of ​​1,300 square meters.

[0114] In terms of the overall layout of the park, the northern area is planned to have a 54,600-square-meter port empty container storage area, while the southern area has a large-scale railway freight station container yard. The yard is equipped with rail-mounted gantry cranes to handle the loading and unloading of refrigerated containers on four dedicated loading and unloading lines, and achieves seamless connection between various modes of transportation such as highways, railways, and waterways through a comprehensive road traffic network system.

[0115] This embodiment presents a comparative verification experiment between traditional scheduling algorithms and the novel scheduling algorithm of this invention in terms of workflow efficiency and cargo loss rate. To ensure a high degree of consistency between the simulation results and actual operation, the experiment uses seconds as the unit of time measurement. Based on the actual eight-hour daily working schedule of the distribution center, a simulation model was built using Anylogic software, such as... Figure 3 The model ran continuously for three days, totaling 86,400 seconds of simulation, and the results are as follows:

[0116] (1) Comparison of inventory turnover time:

[0117] Table 2 Comparison of Inventory Turnover Time

[0118]

[0119] As shown in Table 2, the traditional scheduling method resulted in an inbound time of 1695 seconds. This method reduces the inbound time to 1060 seconds, a reduction of 37.4%, and the outbound time from 2960 seconds to 2192 seconds, a reduction of 25.9%. This two-way efficiency improvement reduces the overall inventory turnover time from 46445 seconds to 45768 seconds, a reduction of 1.5%. This method deeply optimizes the pre-inbound preparation process for cold chain palletized goods, advancing the average shelving time by 10 minutes. While this timely shelving mechanism increases the storage cycle by 736 seconds, it effectively avoids the quality risks caused by goods remaining in the temporary storage area under the traditional operating model, and simultaneously creates an operational window for dynamic storage location optimization.

[0120] (2) Comparison of working indicators of fixed equipment in the park

[0121] Table 3 Comparison of Operating Indicators of Fixed Equipment

[0122]

[0123] As shown in Table 3, after adopting this method, the average storage time of containers at the port was reduced from 12 hours to 9 hours, a decrease of 36.9%; the dwell time of refrigerated containers in empty container yards was reduced from 11 hours to 9 hours, and the turnover efficiency was improved by 23.7%; and the storage time of refrigerated containers in railway yards benefited from the static priority algorithm and the optimization of loading and unloading processes, with the container storage time being reduced sharply from 10.1 minutes to 7.8 minutes, a decrease of approximately 22.7%.

[0124] (3) Comparison of performance indicators of mobile devices in the park

[0125] Table 4 Comparison of Mobile Device Performance Indicators

[0126]

[0127] As shown in Table 4, compared with the traditional method, the average dwell time of unloading trains is reduced by 131 seconds, a decrease of 4.9%, and the average dwell time of loading is reduced by 1974.8 seconds, a decrease of 24%. In terms of refrigerated truck operations, the unloading dwell time is reduced by 913.4 seconds, a decrease of 52.9%, and the loading dwell time is reduced by 238.9 seconds, a decrease of 9.48%. In addition, the optimized operation time of internal container trucks is 3733.9 seconds per vehicle.

[0128] (4) Cargo handling capacity analysis

[0129] The simulation model yielded the following curves showing the processing capacity of refrigerated pallets and refrigerated containers at the distribution center under two different operational processes: Operation Process 1 represents the traditional scheduling method, and Operation Process 2 represents the proposed method.

[0130] Data shows that the standard refrigerated pallet processing efficiency jumped from 244 pallets per hour in the original scheme to 342 pallets per hour, an absolute increase of 98 pallets per hour, representing a 40.2% increase in processing capacity. Dynamic simulation curves further reveal that, throughout the continuous simulation period, the processing volume curve of this method consistently remained higher than the baseline of the traditional method, with the peak processing efficiency gap widening to a maximum of 128 pallets per hour.

[0131] Refrigerated container handling capacity curve as shown in the figure Figure 4 .

[0132] Data shows that during continuous operation simulation, the cumulative processing volume curve of this method consistently outperforms the traditional method by a stable advantage.

[0133] (5) Comparison of cargo loss rates

[0134] Table 5 Comparison of Cargo Loss Rates under Different Scheduling Methods

[0135]

[0136]

[0137] As shown in Table 5, the cargo loss rate decreased significantly from 6.916% using the traditional method to 4.258%, a reduction of 2.658 percentage points, equivalent to a relative reduction of 38.4% in the loss rate.

Claims

1. A method for optimizing the operation scheduling of a railway cold chain intermodal transport distribution center, characterized in that, The following steps are involved: S1. Resource pre-allocation and time window constraint embedding The pre-allocation of loading and unloading resources is based on the train timetable to lock in the loading and unloading lines, rail gantry cranes and cold storage racks in advance. The pre-allocation window is no less than 2 hours before the train arrives at the station. If the train is delayed, priority reassessment is triggered in real time, and resources are dynamically allocated according to the temperature control status of the goods. The pre-matching of platform resources refers to the allocation of target platforms 1 hour before the arrival of refrigerated trucks, and the establishment of forklift-rack-platform resource chain 30 minutes before the goods enter and leave the warehouse. S2. Dynamic and Static Priority Collaborative Decision-Making First, set the static priority of the goods, and classify them into three levels according to their value, timeliness and perishability: Category A: high value perishable; Category B: conventional cold chain; Category C: temperature-controlled storage. Category A goods are assigned the highest fixed priority. Then, adjust the dynamic priority, that is, monitor the loss rate of goods in real time, and automatically increase the priority when the loss rate exceeds the threshold. In the inbound process, cold storage resources are allocated according to static priority, while in the outbound process, scheduling instructions are generated according to dynamic priority. S3. Spatial Constraint-Oriented Path Optimization Based on the strip-shaped operating range of the rail gantry crane, non-intersecting paths are planned for container trucks to avoid equipment waiting conflicts; a joint path network of forklifts, container trucks, and gantry cranes is constructed, and the shortest movement path between nodes is calculated in real time using the Dijkstra algorithm; dynamic avoidance logic is set up for unloaded equipment to avoid fully loaded equipment and for low-priority task equipment to avoid high-priority task equipment. S4. Collaborative Execution of Work Processes (1) Cold storage operations Inbound stage: Receive pre-arrival information → allocate shelves according to static priority → forklifts transport goods to the temporary storage area along the shortest path → put goods on shelves after quality inspection; Outbound phase: Dynamic priority triggers orders → Goods with the highest loss rate are removed from shelves first → Forklifts transfer goods to the loading platform via optimized routes; (2) Park transfer operations Train arrival: Pre-locked loading and unloading line → Gantry crane unloads containers according to static priority sequence → Trucks transfer containers via non-intersecting paths; Multimodal transport conversion: Rail to road uses the "platform direct loading" mode, that is, container rail → truck chassis directly; goods that need to be distributed are transferred to cold storage for unpacking and sorting, reducing the exposure time throughout the process; S5. Real-time monitoring and feedback Deploy a cargo loss rate monitoring system to collect temperature and humidity data in real time, update dynamic priorities, and establish a resource conflict early warning mechanism. When the equipment waiting time exceeds the threshold, trigger path replanning.

2. The method for optimizing the operation scheduling of a railway cold chain intermodal transport distribution center according to claim 1, characterized in that, S1 employs a multi-agent cooperative rule generation method, the specific process of which is as follows: Eight types of autonomous intelligent agents are constructed: trains, carriages, containers, trucks, rail gantry cranes, forklifts, pallets, and refrigerated trucks. Each type of intelligent agent has independent decision-making logic and environmental interaction capabilities. Through the real-time state transitions and coordination rules of intelligent agents, autonomous decision-making and conflict resolution of work processes are realized. Based on a multi-agent operation logic architecture, a collaborative simulation system for the operation process of a railway cold chain intermodal transport distribution center is constructed using the process modeling and track modeling components of a simulation modeling platform. The system covers static infrastructure such as platform facilities, racking structures, yard areas, and rail lines, as well as dynamic transportation facilities such as forklift equipment, cold chain trains, refrigerated vehicles, gantry cranes, and container trucks. Multi-agent collaborative rule generation mechanism: (1) Optimization of intelligent agent collaboration rules for refrigerated trucks: Set the delayed arrival parameter in the refrigerated truck generation module, and realize the advance matching and allocation of refrigerated trucks and platform resources through the queue management module to avoid resource competition conflicts after arrival; (2) Platform resource agent collaboration rules: In the platform resource acquisition module, set task priority parameters according to the cargo category attributes, configure the resource exclusive and uninterrupted task preemption strategy, and ensure that the forklift agent prioritizes the processing of high-priority refrigerated cargo. (3) Collaborative rules for loss monitoring agents: Design a cargo loss rate function module, configure loss rate monitoring attributes for each refrigerated cargo agent entering the system, assign initial loss rate values ​​according to probability distribution, and set a custom resource selection algorithm in the cargo removal operation module to realize dynamic priority outbound processing of cargo with the highest loss rate. (4) Path planning agent collaboration rules: Set up a transfer and temporary storage area and a dedicated forklift lane between the internal and external areas of the cold storage room, construct a forklift path network topology, and add a resource selection and allocation algorithm to the collaboration logic, which includes the complete process of resource acquisition - movement execution - resource release. (5) Train intelligent agent collaboration rules: A delayed arrival mechanism is set in the cold chain train generation module, and the pre-allocation and matching of cold chain trains and loading and unloading line resources is realized through the queue management module; (6) Equipment cluster collaboration rules: Set up a refrigerated container temporary storage yard area in the park, and integrate the container front crane, container truck and gantry crane equipment intelligent agents into the same path network topology. By modifying the train operation logic process, collaborative operation between equipment can be realized.

3. The method for optimizing the operation scheduling of a railway cold chain intermodal transport distribution center according to claim 1, characterized in that, S2 employs a dynamic and static priority evaluation method, including: In response to the characteristics of cold chain goods, a dual-track priority mechanism of "static grading + dynamic adjustment" will be established: Dynamic priority is based on real-time loss rate monitoring value to dynamically adjust priority. The loss rate of goods is determined by the Arrhenius equation, which is the chemical reaction temperature dependence formula, to describe the effect of temperature on product quality degradation, as shown in equation (1). Where: k—the loss rate of the goods; k0—a constant initial loss rate parameter; E a —Experimental activation energy; R—gas constant; T—Absolute temperature; The loss model is given by equation (2), where q′ is the quantity of goods, t is time, and q0 is the initial quantity of products; q′=q0-q0·e -kt (2) When the monitored value exceeds the threshold, the priority is automatically upgraded and the emergency dispatch plan is triggered.

4. The method for optimizing the operation scheduling of a railway cold chain intermodal transport distribution center according to claim 1, characterized in that, S1 adopts a time- and space-constrained resource pre-allocation method. To address the time rigidity and space constraints of the railway cold chain, a resource pre-allocation model is designed: (1) Pre-lock loading and unloading lines, gantry cranes and cold storage racks before the train arrives at the station; (2) Target platforms are assigned to refrigerated trucks before they arrive at the station; (3) Match the forklift-rack-platform resource chain in advance before inbound and outbound operations; (4) If the train is delayed, priority will be reordered; that is, high-risk goods will be loaded and unloaded first, and low-priority tasks will be postponed.

5. The method for optimizing the operation scheduling of a railway cold chain intermodal transport distribution center according to claim 1, characterized in that, S3 employs a conflict-avoidance path planning method, specifically: Construct a conflict-free mobile path network based on the shortest path algorithm and spatial constraint rules: Plan the shortest travel path for forklifts, container trucks, and gantry cranes to reduce unnecessary travel distances and increase forklift travel speed; preset non-intersecting paths for gantry cranes and container trucks; enforce a three-level avoidance mechanism of "empty loads yielding to fully loaded loads", "low priority yielding to high priority", and "later-arriving equipment yielding to earlier-arriving equipment"; set up "temporary storage areas" in cold storage to shorten transfer distances.

Images