Intelligent fruit and vegetable eco-friendly baskets and their intelligent scheduling methods

By assessing the health level of crates and real-time inventory, and combining predictive models and optimization algorithms, the task of heavy goods delivery and empty crate recycling is integrated, which solves the problem of uneven crate resources in logistics scheduling and improves operational efficiency and cost-effectiveness.

CN123117956APending Publication Date: 2026-11-17GUANGDONG JINMANKUANG TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611547544.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-08
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing logistics scheduling methods fail to effectively combine vehicle route planning with basket resource allocation, resulting in uneven basket resource flow, waste of high-health baskets and damage to low-health baskets, and an inability to respond to real-time fluctuations in supply and demand, thus affecting operational efficiency.

Method used

By acquiring structural loss monitoring data to assess the health level of the crates, and combining real-time inventory and order demand, predictive models and optimization algorithms are used to integrate heavy goods delivery and empty crate recycling tasks, generating dynamic scheduling instructions to achieve dynamic balance and flow of crate resources.

Benefits of technology

It improves the utilization rate and circulation efficiency of containers, reduces logistics operating costs, and solves the problems of container supply and demand imbalance and low recycling efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123117956A_ABST
    Figure CN123117956A_ABST
Patent Text Reader

Abstract

This application provides an intelligent fruit and vegetable eco-friendly basket and its intelligent scheduling method, including: acquiring the material delivery task demand of each circulation node within a target time period and the structural damage monitoring data of the intelligent fruit and vegetable eco-friendly basket; determining the structural integrity health level of the intelligent fruit and vegetable eco-friendly basket through the structural damage monitoring data; acquiring the real-time inventory quantity of available intelligent fruit and vegetable eco-friendly baskets at each circulation node based on the material delivery task demand; determining the maximum material carrying capacity of the intelligent fruit and vegetable eco-friendly basket based on the structural integrity health level; acquiring the statistical value of the number of unrecovered empty vehicles remaining at each end node within a historical delivery cycle based on the carrying capacity matching relationship; if the statistical value of the number of unrecovered empty vehicles remaining is greater than a preset retention threshold, then determining the corresponding end node as the priority recycling node for empty vehicle recycling; and acquiring the total expected number of vehicles circulating along the optimized equipment travel path.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to intelligent fruit and vegetable eco-friendly baskets and their intelligent scheduling methods. Background Technology

[0002] In modern fruit and vegetable cold chain logistics systems, reusable smart eco-friendly crates serve as a core carrier connecting the production and consumption ends, and their efficient scheduling directly impacts the operating costs and service quality of the entire supply chain. With the rapid development of fresh food e-commerce and increasingly stringent environmental policies, the application scale of smart fruit and vegetable eco-friendly crates continues to expand. How to achieve optimal allocation of crate resources within a complex and ever-changing logistics network has become a key bottleneck restricting cost reduction and efficiency improvement in the industry.

[0003] Current logistics scheduling methods treat vehicle route planning and basket resource allocation as two independent processes—first planning delivery routes based on order demand, then allocating baskets according to the predetermined routes, or first determining the basket allocation plan, then arranging vehicles to perform transportation tasks. This sequential decision-making model ignores the constraints between the two: the delivery routes chosen by vehicles directly affect the turnover speed and backlog of baskets at different locations, while the inventory status of baskets in various locations, in turn, limits the types of orders and loading schemes that vehicles can handle. When a warehouse experiences a shortage of specific-sized baskets, even if vehicles are capable of delivering more orders, their capacity will be idle due to insufficient basket supply; conversely, when customers have a large backlog of empty baskets awaiting recycling, if vehicles only perform one-way delivery tasks, it will cause empty baskets to remain and a global imbalance in basket resources. This disconnect between route planning and basket scheduling leads to two core contradictions in the basket circulation process. On the one hand, the crates themselves exhibit physical wear and tear and varying health conditions. Newly manufactured crates with high health conditions have different load-bearing capacities and applicable scenarios compared to crates with low health conditions after multiple uses. However, existing scheduling methods typically treat all available crates equally, resulting in high-health crates being wasted on short-distance, low-requirement tasks, while low-health crates are sent to long-distance, high-intensity transportation, accelerating their deterioration. On the other hand, the dynamic supply and demand relationship of crates at various nodes exhibits significant spatiotemporal imbalances. Some warehousing centers face crate shortages during the morning peak, while other centers have a large number of empty crates idle. Traditional fixed-cycle allocation mechanisms cannot respond to such real-time fluctuations.

[0004] For example, in a certain operation, Warehouse A ran out of a certain type of crates within two hours due to receiving a large number of orders. Meanwhile, Warehouse B, which was only ten kilometers away, had more than two hundred empty crates of the same type waiting to be cleaned. However, because the dispatching system failed to link this supply and demand gap with the routes of vehicles that were carrying out delivery tasks, Warehouse A had to delay order delivery, while the empty crates in Warehouse B continued to accumulate and occupy storage space.

[0005] Therefore, how to simultaneously consider the health status classification of baskets, dynamic inventory distribution, and recycling needs during vehicle route planning, and achieve deep coupling optimization between route decision-making and basket scheduling, so that vehicles can efficiently recycle empty baskets while completing heavy cargo delivery, and basket resources can be dynamically balanced and circulated between different health levels and spatial locations, has become a key issue that intelligent fruit and vegetable environmental protection basket storage and logistics system urgently needs to overcome. Summary of the Invention

[0006] This invention provides an intelligent scheduling method for intelligent fruit and vegetable eco-friendly baskets, mainly including:

[0007] The process involves: acquiring the material delivery task demand at each circulation node within a target time period and monitoring data on the structural damage of the smart fruit and vegetable eco-friendly baskets; determining the structural integrity health level of the smart fruit and vegetable eco-friendly baskets using the structural damage monitoring data; obtaining the real-time inventory quantity of available smart fruit and vegetable eco-friendly baskets at each circulation node based on the material delivery task demand; determining the maximum material carrying capacity of the smart fruit and vegetable eco-friendly baskets based on the structural integrity health level; comparing the quantity difference between the real-time inventory quantity of available smart fruit and vegetable eco-friendly baskets and the material delivery task demand to obtain the supply-demand gap quantity of smart fruit and vegetable eco-friendly baskets at each circulation node during different delivery time periods; processing the historical vehicle circulation trajectory data corresponding to the supply-demand gap quantity of smart fruit and vegetable eco-friendly baskets using a predictive model based on the supply-demand gap quantity and the maximum material carrying capacity to obtain the average circulation turnover speed of the smart fruit and vegetable eco-friendly baskets in the current delivery network environment; and adjusting the material carrying capacity based on the average circulation turnover speed and the maximum material carrying capacity. The feasibility of heavy-load delivery tasks exceeding a preset quality threshold is screened to determine the load matching relationship between the smart fruit and vegetable eco-friendly baskets and the heavy-load delivery tasks. Based on the load matching relationship, the statistical value of the number of unrecovered empty vehicles remaining at each end node in the historical delivery cycle is obtained. If the statistical value of the number of unrecovered empty vehicles remaining is greater than a preset threshold, the corresponding end node is determined as a priority recovery node for empty vehicle recovery. For the priority recovery nodes and the supply-demand gap of the smart fruit and vegetable eco-friendly baskets, an optimization algorithm is used to perform cross-mutation optimization on the delivery routes corresponding to the load matching relationship to obtain an optimized equipment travel path that simultaneously includes heavy-load delivery tasks and empty vehicle recovery tasks. The total expected number of vehicles circulating along the optimized equipment travel path is obtained. If the total expected number of vehicles circulating is greater than the supply-demand gap of the smart fruit and vegetable eco-friendly baskets, a vehicle scheduling execution instruction including delivery time nodes and recovery time nodes is generated based on the optimized equipment travel path. Furthermore, determining the structural integrity health level of the smart fruit and vegetable eco-friendly basket using the structural loss monitoring data includes: acquiring the initial inventory quantity of each circulation node, as well as the node distance and load distribution data between circulation nodes; obtaining a loss degree parameter by weighting and summing the carrier frame deformation and load-bearing plate wear depth included in the structural loss monitoring data with preset weights; determining the structural integrity health level of the smart fruit and vegetable eco-friendly basket based on the loss degree parameter; if the structural integrity health level is severely damaged, removing the corresponding carrier from the initial inventory quantity to obtain the real-time inventory quantity of available smart fruit and vegetable eco-friendly baskets at each circulation node; and determining the replenishment quantity for each circulation node by the difference between the material delivery task demand and the real-time inventory quantity of available smart fruit and vegetable eco-friendly baskets.A pre-trained support vector machine model is used to classify and map the replenishment quantity and the node distance to obtain the allocation priority between each transfer node; an allocation matrix is ​​constructed based on the allocation priority; corresponding capacity matching values ​​are extracted based on the allocation matrix; and the final delivery route is determined by combining the capacity matching values ​​with the load distribution data. Furthermore, the step of comparing the real-time inventory of available smart fruit and vegetable eco-friendly baskets with the demand for material delivery tasks to obtain the supply-demand gap of smart fruit and vegetable eco-friendly baskets at each distribution node during different delivery periods includes: extracting the corresponding gap severity level based on the supply-demand gap of smart fruit and vegetable eco-friendly baskets; if the gap severity level is higher than a preset level threshold, then using a random forest algorithm to extract features from the supply-demand gap of smart fruit and vegetable eco-friendly baskets and the different delivery periods to obtain the emergency replenishment amount for each distribution node; constructing a spatial distance matrix by combining the emergency replenishment amount with the geographical coordinate information of each distribution node; calculating the transfer cost value between each distribution node based on the spatial distance matrix; obtaining the number of spare vehicles for each distribution node; determining the vehicle scheduling priority for each distribution node by performing a weighted summation based on the transfer cost value and the number of spare vehicles; generating target scheduling instructions for each distribution node based on the vehicle scheduling priority; and controlling the transportation equipment to perform vehicle transfer operations according to the target scheduling instructions. Furthermore, the step of using a predictive model to process historical vehicle circulation trajectory data corresponding to the supply-demand gap quantity of the smart fruit and vegetable eco-friendly baskets, and obtaining the average circulation turnover speed of the smart fruit and vegetable eco-friendly baskets under the current delivery network environment, includes: obtaining the supply-demand gap quantity and the maximum material carrying capacity of the smart fruit and vegetable eco-friendly baskets; filtering out historical vehicle circulation trajectory data that meet the conditions based on the supply-demand gap quantity and the maximum material carrying capacity, wherein the historical vehicle circulation trajectory data includes the round-trip turnover time from the circulation node to the end node and the dwell time en route; and extracting the... The round-trip turnaround time and the stopover time en route are calculated. The average turnaround time for each node is calculated based on the round-trip turnaround time and the stopover time. If the average turnaround time exceeds a preset time threshold, real-time road condition data under the current delivery network environment is obtained. A random forest regression prediction model is used to extract and fuse features from the average turnaround time and the real-time road condition data to obtain the dynamic turnaround characteristics of the smart fruit and vegetable eco-friendly basket. The average turnaround speed of the smart fruit and vegetable eco-friendly basket under the current delivery network environment is determined by weighted calculation based on the dynamic turnaround characteristics and the maximum material carrying capacity. Further, determining the carrying capacity matching relationship between the smart fruit and vegetable eco-friendly basket and the heavy-load delivery task includes: extracting heavy-load delivery tasks exceeding a preset quality threshold from the material delivery task demand; obtaining the material space volume and material stacking restrictions of the heavy-load delivery task.Based on the maximum material carrying capacity, the material space volume, and the material stacking limit, calculate the number of heavy-load orders that a single smart fruit and vegetable eco-friendly basket can carry; obtain the node loading and unloading time of the heavy-load delivery task; calculate the delivery flow efficiency of the heavy-load delivery task based on the average turnover speed and the node loading and unloading time; if the delivery flow efficiency is greater than a preset efficiency threshold, calculate the total number of smart fruit and vegetable eco-friendly baskets required based on the number of heavy-load orders that a single smart fruit and vegetable eco-friendly basket can carry and the total number of orders included in the heavy-load delivery task; determine the carrying capacity matching relationship between the smart fruit and vegetable eco-friendly basket and the heavy-load delivery task based on the number of heavy-load orders that a single smart fruit and vegetable eco-friendly basket can carry and the total number of smart fruit and vegetable eco-friendly baskets required. Furthermore, the step of determining the corresponding end node as the priority recovery node for empty vehicle recovery if the statistical value of the number of unrecovered empty vehicles is greater than a preset threshold value includes: extracting the number of empty vehicles remaining at each end node in the historical delivery cycle based on the load matching relationship; calculating the statistical value of the number of empty vehicles remaining; if the statistical value is greater than the preset threshold value, determining the corresponding end node as the priority recovery node for empty vehicle recovery; obtaining the geographical coordinates of the recovery node; and clustering the geographical coordinates using a hierarchical clustering algorithm. The system is divided into multiple recycling point clusters. For each recycling point cluster, road network data between recycling nodes within the cluster is obtained. The distance and time consumption between adjacent recycling nodes are extracted based on the road network data. Based on the distance and time consumption, the Dijkstra algorithm is used to calculate the shortest traversal path within the recycling point cluster, generating an initial route map for the recycling point cluster. The maximum loading capacity of the transportation equipment performing the recycling task is obtained. If the total number of empty vehicles remaining in the recycling point cluster is less than the maximum loading capacity, a corresponding recycling dispatch order is generated based on the initial route map. Furthermore, the step of using an optimization algorithm to perform crossover and mutation optimization on the delivery routes corresponding to the load matching relationship to obtain an optimized equipment driving path that simultaneously includes heavy-load delivery tasks and empty vehicle recovery tasks includes: obtaining heavy-load delivery tasks and recovery tasks as gene nodes; arranging the gene nodes using the chromosome encoding mechanism of a genetic algorithm to generate an initial gene coding sequence; selecting two parent gene coding sequences according to a preset crossover probability for the initial gene coding sequence; exchanging fragments of some gene nodes in the parent gene coding sequences to obtain a crossover recombination sequence; randomly selecting mutation nodes according to a preset mutation probability for the crossover recombination sequence; adjusting the insertion position of the recovery node corresponding to the mutation node to obtain a mutation-optimized sequence; and calculating the cumulative load capacity per vehicle corresponding to the mutation-optimized sequence based on the heavy-load delivery tasks in the mutation-optimized sequence and the supply-demand gap of the smart fruit and vegetable environmental protection baskets.If the cumulative load of a single vehicle is less than the preset maximum load threshold of the equipment, the modified optimization sequence is decoded to obtain an optimized equipment driving path that includes both heavy-load delivery tasks and empty vehicle recovery tasks. Furthermore, the step of generating vehicle scheduling execution instructions containing delivery time nodes and collection time nodes based on the optimized equipment travel path includes: extracting the spatial coordinate sequence of each station based on the optimized equipment travel path; calculating the physical distance set between adjacent stations using the spatial coordinate sequence; calculating the basic travel time sequence between adjacent stations based on the physical distance set and a preset equipment travel speed threshold; accumulating the delivery time node and collection time node corresponding to the optimized equipment travel path based on the basic travel time sequence and a preset station loading and unloading operation time; constructing an initial time axis sequence based on the delivery time node and collection time node; obtaining the number of newly delivered smart fruit and vegetable environmental protection baskets and the number of empty collection vehicles; summing the number of newly delivered smart fruit and vegetable environmental protection baskets and the number of empty collection vehicles to obtain the expected total number of vehicle circulations; if the expected total number of vehicle circulations is greater than the supply-demand gap of smart fruit and vegetable environmental protection baskets, then using a decision tree algorithm to classify the initial time axis sequence to obtain a target time axis sequence with priority labels; and generating vehicle scheduling execution instructions containing delivery time nodes and collection time nodes based on the target time axis sequence with priority labels. ;

[0008] The intelligent fruit and vegetable eco-friendly basket is characterized by comprising:

[0009] The basket includes a bottom and sidewalls surrounding the bottom, the sidewalls and the bottom enclosing a space for accommodating fruits and vegetables; a reinforcing structure is provided on the bottom and / or the sidewalls to improve the load-bearing capacity of the basket; a structural loss monitoring tag is provided on the basket to provide structural loss monitoring data of the smart fruit and vegetable eco-friendly basket to determine the structural integrity health level and maximum material load capacity of the smart fruit and vegetable eco-friendly basket; the smart fruit and vegetable eco-friendly basket is scheduled using an intelligent scheduling method.

[0010] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0011] This invention addresses the problems of container supply-demand imbalance and low recycling efficiency in logistics and distribution networks. It assesses container health levels and determines maximum carrying capacity through structural loss monitoring data, analyzes gaps by combining real-time inventory and order demand, and uses a random forest regression model to predict container turnover speed, thereby accurately selecting the carrying capacity matching relationship for heavy goods delivery tasks. After identifying terminal nodes with excessive numbers of unrecycled empty containers, a genetic algorithm is used to optimize delivery routes through cross-mutation, integrating empty container recycling tasks with heavy goods delivery tasks into the same travel path. A scheduling instruction is generated when the expected total container turnover meets the supply-demand gap. This method achieves integrated dynamic scheduling and recycling under container health status awareness, improving container utilization and turnover efficiency, reducing logistics operating costs, and solving the problems of empty container return and supply-demand mismatch in traditional scheduling models. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of the framework of step S101 in the intelligent scheduling method of the intelligent fruit and vegetable environmental protection basket of the present invention.

[0013] Figure 2 This is a schematic diagram of the framework of step S103 in the intelligent scheduling method of the intelligent fruit and vegetable environmental protection basket of the present invention.

[0014] Figure 3 This is a schematic diagram of the framework of step S104 in the intelligent scheduling method of the intelligent fruit and vegetable environmental protection basket of the present invention.

[0015] Figure 4 This is a three-dimensional structural diagram of the intelligent fruit and vegetable eco-friendly basket of the present invention. In the diagram, 1-basket body, 2-structural loss monitoring tag, 3-side wall reinforcing rib, 4-ventilation hole. Detailed Implementation

[0016] To further understand the content of this invention, a detailed description of the invention is provided in conjunction with the accompanying drawings and embodiments. The specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0017] like Figure 4 As shown, this embodiment provides an intelligent fruit and vegetable environmental protection basket, including a basket body 1, a reinforced structure, and a structural damage monitoring tag 2.

[0018] The basket body 1 includes a bottom and side walls surrounding the bottom. The bottom is a rectangular flat plate structure, and the side walls include a front side wall, a rear side wall, a left side wall, and a right side wall. The four side walls extend vertically upwards from the four edges of the bottom, and the corners of adjacent side walls are connected by rounded transitions. The side walls and the bottom enclose a top-open storage space for holding fruits and vegetables. The basket body is integrally injection molded from environmentally friendly plastic material, preferably made of recyclable and environmentally friendly materials such as polypropylene (PP) or polyethylene (PE).

[0019] In this embodiment, multiple ventilation holes 4 are provided on the side wall. The ventilation holes are circular and elongated, and arranged in a matrix on the side wall. The ventilation holes are used to maintain air circulation inside the basket during the storage and transportation of fruits and vegetables, preventing the fruits and vegetables from rotting due to heat and stuffiness.

[0020] Reinforcing structures are installed on the bottom and side walls to improve the load-bearing capacity of the basket. Specifically, the reinforcing structures include bottom reinforcing ribs and side wall reinforcing ribs 3. The bottom reinforcing ribs are located on the outer bottom surface and are arranged in a cross pattern, including longitudinal reinforcing ribs extending along the length of the basket and transverse reinforcing ribs extending along the width of the basket. The bottom reinforcing ribs are integrally injection molded with the bottom and protrude slightly from the outer bottom surface. The side wall reinforcing ribs 3 are located on the outer surface of the side walls and extend along the height of the side walls, with multiple vertical reinforcing ribs spaced apart on each side wall. The side wall reinforcing ribs are also integrally injection molded with the side walls.

[0021] Structural damage monitoring tag 2 is affixed to the outer surface of one side wall of the basket, located near the top edge of the side wall. The tag is a rectangular sheet structure, with its back adhered to the outer surface of the side wall using adhesive. The tag stores the identification information of the smart fruit and vegetable eco-friendly basket and structural damage monitoring data. This data characterizes the degree of structural damage to the basket caused by factors such as collisions, wear, and aging during use. This data can be read by external devices via RFID or near-field communication, and is used to determine the structural integrity health level and maximum load-bearing capacity of the smart fruit and vegetable eco-friendly basket.

[0022] like Figures 1-3 The intelligent scheduling method for the intelligent fruit and vegetable eco-friendly basket in this embodiment may specifically include:

[0023] Step S101: Obtain the demand for goods delivery orders at each logistics node within the target time period and the structural damage monitoring data of the target logistics container. The structural damage monitoring data includes the deformation of the container frame and the wear depth of the load-bearing plate. Determine the structural integrity health level of the target logistics container through the damage degree parameter in the structural damage monitoring data. The health level is divided into intact level, slightly damaged level and severely damaged level. Obtain the real-time inventory quantity of available target logistics containers at each logistics node based on the demand for goods delivery orders.

[0024] The system acquires the demand for goods delivery orders at each logistics node within a target time period, as well as structural damage monitoring data for the target logistics containers. This data includes container frame deformation and load-bearing plate wear depth. It also acquires the initial inventory quantity at each logistics node, as well as node distances and load distribution data between nodes. A damage severity parameter is obtained by weighting and summing the container frame deformation and load-bearing plate wear depth using preset weights. The specific calculation formula is D = w1 × F + w2 × W, where D is the damage severity parameter, F is the container frame deformation (in millimeters), W is the load-bearing plate wear depth (in millimeters), and w1 and w2 are the corresponding weighting coefficients. These weighting coefficients are determined by analyzing historical failure data. The percentage of containers scrapped due to different damage types is statistically analyzed; frame deformation accounts for 0.35% of the scrapped containers, and load-bearing plate wear accounts for 0.65%. Based on this, w1 is set to 0.35, and w2 to 0.65. For example, if a container frame has a deformation of 8.0 mm and a wear depth of 12.0 mm, then the damage level parameter D = 0.35 × 8.0 + 0.65 × 12.0 = 10.6. The health level of the target logistics container is determined based on this damage level parameter. When D is less than 5.0, it is considered slightly damaged; when D is between 5.0 and 12.0, it is considered moderately damaged; and when D is greater than 12.0, it is considered severely damaged. If the health level is severely damaged, the corresponding container is removed from the initial inventory quantity to obtain the real-time inventory quantity of available target logistics containers at each logistics node. The replenishment quantity for each logistics node is determined by the difference between the demand for goods delivery orders and the real-time inventory quantity of available target logistics containers. A pre-trained support vector machine model is used to classify and map the replenishment quantity and the node distance to obtain the allocation priority between logistics nodes. An allocation matrix is ​​constructed based on the allocation priority. The corresponding capacity matching value is extracted from the allocation matrix, and the final delivery route is determined by combining the capacity matching value with the load distribution data.

[0025] In one implementation, the system first acquires the demand for goods delivery orders at each logistics node within the target time period, as well as the structural damage monitoring data of the target logistics container.

[0026] Specifically, each logistics node typically includes warehousing centers or distribution centers located in different geographical locations. The target time period can be set daily or weekly based on the actual business rhythm. The demand for goods delivery orders reflects the total amount of goods to be shipped out within that period and the corresponding container demand. Simultaneously, the system receives real-time structural damage monitoring data of the target logistics containers through a data acquisition network deployed at the logistics nodes. This data includes the deformation of the container frame and the wear depth of the load-bearing plates; these data serve as the basic input for assessing the physical condition of the containers.

[0027] For example, the target logistics container is typically a metal pallet or composite material crate used to carry heavy goods. The container frame deformation is primarily acquired using strain gauges or laser rangefinders installed at key stress points to characterize the degree of structural torsion or bending that occurs during long-term load-bearing and handling. The wear depth of the load-bearing plates is collected using ultrasonic thickness gauges or surface profile scanning equipment, reflecting material loss from the container's bottom bearing surface caused by cargo friction or forklift operations. This underlying physical monitoring data objectively reflects the container's true physical wear condition.

[0028] It should be noted that determining the structural integrity health level of the target logistics container through the loss degree parameter in the structural loss monitoring data is a process of multi-dimensional data fusion analysis.

[0029] In one possible implementation, the system constructs a structural loss assessment model to calculate the loss degree parameter. This model first normalizes the collected container frame deformation and load-bearing plate wear depth to eliminate differences between different physical dimensions. Subsequently, the system assigns different weighting coefficients to these two types of monitoring data. The weighting coefficients are determined as follows: the system extracts 500 failure cases from a historical failure database. Each case includes the frame deformation before failure, the load-bearing plate wear depth, and the corresponding actual failure time. The correlation between each monitoring parameter and the failure event is calculated using correlation analysis, yielding a correlation of 0.65 for the frame deformation and 0.35 for the load-bearing plate wear depth. After normalizing the correlation, the normalized values ​​are used as weighting coefficients, i.e., w1=0.65, w2=0.35. The system uses this combination of weighting coefficients to calculate the comprehensive loss degree parameter.

[0030] For example, if historical data shows that the risk of overall container collapse due to frame deformation is much higher than that of surface wear on the load-bearing plate, then the container frame deformation is given a higher weight. The system multiplies the normalized container frame deformation by its corresponding weight, and adds the normalized load-bearing plate wear depth by its corresponding weight to obtain a comprehensive value, which is the wear level parameter. To improve the accuracy of the assessment, a nonlinear compensation factor K is introduced for correction, specifically calculated as K = 1 + α × e^(β × D), where D is the ratio of the current deformation or wear depth to the material's physical limit, α is the compensation strength coefficient (valued at 0.3), and β is the acceleration coefficient (valued at 2.5). The material physical limit is determined when the frame deformation reaches 15 mm or the load-bearing plate wear depth exceeds 40% of the original thickness. At this point, D approaches 1.0, and the compensation factor is significantly increased to reflect the accelerated wear effect. The corrected wear level parameter is calculated by multiplying the original linear weighted value by the compensation factor K, thus ensuring that the calculation results accurately map the current structural fatigue state of the target logistics container. Furthermore, the system determines the structural integrity health level of the target logistics container based on the calculated damage level parameters. These health levels are categorized into three levels: intact, slightly damaged, and severely damaged. Specifically, the system presets a first damage threshold of 0.35 and a second damage threshold of 0.70. These thresholds are determined based on statistical analysis of numerous logistics container failure cases. 0.35 corresponds to the critical point where the container begins to show visible structural changes but its load-bearing capacity decreases by no more than 20%, while 0.70 corresponds to the critical point where the container's load-bearing capacity decreases by more than 50% and there are significant safety hazards. When the damage level parameter of the target logistics container is less than 0.35, the system determines its structural integrity health level to be intact, indicating that the container structure is stable, can operate at full load, and has no safety hazards. When the damage level parameter is greater than or equal to 0.35 and less than 0.70, it is classified as slightly damaged. In this case, the system calculates usage restrictions based on the specific damage level parameter value. The load-bearing capacity is adjusted to 100 × (1-D) percent of the original rated load, and the stacking layer limit is the original allowable layer number × (1.2-D) rounded down, where D is the damage level parameter value. For example, when the damage level parameter is 0.50, the load-bearing capacity should be adjusted to 50% of the original rated load, and the stacking layer limit is the original allowable layer number × 0.7 rounded down. When the damage level parameter is greater than or equal to 0.70, it is classified as severely damaged, meaning the container faces a significant risk of breakage or collapse and must be immediately removed from circulation and scrapped or overhauled. This tiered strategy effectively avoids a one-size-fits-all scrapping mechanism and maximizes the container's service life.

[0031] Understandably, through the aforementioned rigorous monitoring and grading mechanism, the logistics management system can clearly grasp the actual availability status of each target logistics container. This provides a reliable basis for subsequent inventory scheduling and order matching, effectively preventing cargo damage or safety accidents caused by container structural failure.

[0032] In one embodiment, the real-time inventory of available target logistics containers at each logistics node is obtained based on the demand for goods delivery orders. When performing inventory statistics, the system does not simply accumulate all containers within a node, but rather dynamically filters them based on the aforementioned structural integrity and health levels. First, the system removes target logistics containers with a severely damaged health level from the total inventory of the logistics node, as these containers are no longer operational. Next, the system extracts containers with intact and slightly damaged health levels as the basic available inventory. For slightly damaged containers, the system sets their load-bearing capacity limit to 70% of the container's standard rated load-bearing capacity; for example, a container with a standard rated load-bearing capacity of 1000 kg will have its load-bearing capacity limit reduced to 700 kg after slight damage. Subsequently, the system analyzes the demand for goods delivery orders and extracts the weight and volume attributes of the goods in the orders. For heavy cargo orders where the weight of a single item exceeds the carrying capacity of a slightly damaged container, the system only includes intact containers in the available real-time inventory for that order; for orders where the weight of a single item does not exceed the carrying capacity of a slightly damaged container, both intact and slightly damaged containers are included in the available real-time inventory.

[0033] Preferably, the system also deducts the allocated target logistics containers in real time based on the consumption progress of goods delivery order demand, thereby obtaining the real-time inventory quantity of available target logistics containers at each logistics node. As orders are continuously placed and containers continue to circulate, the system periodically re-collects structural loss monitoring data and updates the health level to ensure that the real-time inventory quantity always reflects the actual safe carrying capacity of the logistics node, achieving precise matching of logistics resource supply and demand.

[0034] Step S102: Determine the maximum cargo carrying capacity of the target logistics container based on the structural integrity health level, and compare the quantity difference between the real-time inventory quantity of the available target logistics containers and the demand quantity of the cargo delivery order to obtain the supply and demand gap of target logistics containers at each logistics node in different delivery time periods.

[0035] The system obtains the demand for goods delivery orders at each logistics node during different delivery periods. It then calculates the difference between the real-time inventory of available target logistics containers at each node and the demand for goods delivery orders to determine the container supply-demand gap at each node during different delivery periods. A gap less than 15% of demand is classified as Level 1 (mild gap), 15% to 40% as Level 2 (moderate gap), and more than 40% as Level 3 (severe gap). If the gap severity reaches Level 2 or higher, a random forest algorithm is used to extract features from the container supply-demand gap, different delivery periods, historical replenishment records, and the node's average daily throughput. The random forest model consists of 200 decision trees, with 70% of the features randomly selected from each tree for training. The input is the aforementioned four types of feature vectors, and the output is the predicted emergency replenishment amount for each logistics node. The distance matrix D is constructed by calculating the Euclidean distance between logistics nodes using their geographical coordinates. The allocation cost C = D × unit distance transportation rate + fixed operating fee, where the unit distance transportation rate is 3.5 yuan per kilometer, and the fixed operating fee is 150 yuan per allocation. The number of spare containers at each logistics node is obtained, defined as the total number of currently idle, unallocated, and intact containers at that node. The allocation cost is normalized to a range of 0 to 1 to obtain the cost index K, and the number of spare containers is normalized to a range of 0 to 1 to obtain the inventory index S. The container scheduling priority P = 0.6 × S - 0.4 × K, with higher P values ​​indicating higher priority. Target scheduling instructions are generated for each logistics node based on the descending order of container scheduling priority. These instructions include the sending node number, receiving node number, number of containers to be allocated, and required delivery time. These target scheduling instructions are used to control transport vehicles to execute container allocation operations.

[0036] The system reads the structural integrity health level of the target logistics container and calls the built-in load-bearing capacity reduction algorithm. The container's factory standard maximum load capacity is set to 1000 kg. When the health level is identified as intact, the system assigns a reduction coefficient of 1, determining its maximum cargo load capacity to be 1000 kg. When the health level is identified as slightly damaged, the system extracts its specific damage level parameter. Let the damage parameter be d, then take a value of 0.45. Substitute this value into the linear attenuation function C = C0 × (1 - 0.5d) to calculate the current load capacity, where C0 is the standard load capacity and the coefficient 0.5 is the safety redundancy coefficient. The calculated maximum cargo load capacity of the container is 775 kg. The system then divides the 24 hours into six 4-hour delivery periods. Taking the first period as an example, the system analyzes the demand for delivery orders during that period, identifying 150 heavy-load orders with a single item weight exceeding 775 kg and 300 light-load orders with a single item weight below 775 kg. The system synchronously retrieves the real-time inventory of available target logistics containers for that time period. Classified by carrying capacity, it calculates that there are 120 containers with a maximum carrying capacity greater than 775 kg and 250 containers with a maximum carrying capacity less than 775 kg. The system compares the demand for heavy-load orders (150) with the inventory of heavy-load containers (120), and the demand for light-load orders (300) with the inventory of light-load containers (250), to obtain the supply-demand gap for target logistics containers at that logistics node during that delivery period: a shortage of 30 heavy-load containers and a shortage of 50 light-load containers. This gap data, along with its timestamp, is recorded in the logistics scheduling database for subsequent inter-node allocation.

[0037] Step S103: For the quantity of supply and demand gap of the target logistics containers and the maximum cargo carrying capacity, a random forest regression prediction model is used to process the historical container flow trajectory data corresponding to the quantity of supply and demand gap of the target logistics containers. The historical container flow trajectory data includes the round-trip turnaround time and the transit time from the logistics node to the terminal node. The quantity of supply and demand gap is used as the first input feature, the maximum cargo carrying capacity is used as the second input feature, the round-trip turnaround time is used as the third input feature, and the transit time is used as the fourth input feature to construct a feature vector, which is then input into the random forest regression prediction model to obtain the average flow and turnover speed of the target logistics containers in the current delivery network environment as the output result.

[0038] The supply-demand gap and maximum cargo carrying capacity of the target logistics containers are obtained. Based on these gaps, historical container transit trajectory data that meets the criteria are selected. This data includes the round-trip turnaround time and transit dwell time from logistics nodes to terminal nodes. The round-trip turnaround time and transit dwell time are extracted from the historical trajectory data, and the average transit time for each logistics node is calculated. Simultaneously, real-time traffic data under the current delivery network environment is acquired. The historical average transit time, transit dwell time, and real-time traffic congestion index are normalized. The normalized feature vector is input into a random forest regression prediction model to predict the expected transit time T under the current delivery network environment, where T is in hours. The baseline transit speed V = D / T is calculated based on the standard delivery distance D between logistics nodes and the expected transit time T. The benchmark turnover speed V is weighted and calculated with the maximum cargo carrying capacity M. The weighting coefficient is determined according to the container load rate. When the load rate exceeds 0.8, the weighting coefficient is 0.7, otherwise it is 0.9, to obtain the average turnover speed of the target logistics container in the current distribution network environment.

[0039] In one implementation, based on the quantity of supply and demand gaps of target logistics containers at each logistics node during different delivery periods and the maximum cargo carrying capacity of the target logistics containers obtained in the aforementioned steps, the system further triggers a flow rate prediction mechanism.

[0040] Specifically, the system aims to accurately predict the average turnover rate of target logistics containers in the current delivery network environment by analyzing historical circulation patterns and combining them with the current supply and demand status and container physical properties. This provides a time-based decision-making basis for subsequent cross-node scheduling. Furthermore, the system needs to extract historical container circulation trajectory data corresponding to the number of supply-demand gaps in the target logistics containers.

[0041] Understandably, historical container flow trajectory data records the spatial displacement and time consumption of containers in past delivery tasks, while also linking it to the container configuration information for the corresponding tasks. Specifically, this trajectory data mainly includes four types of characteristics: the quantity of supply-demand gap for target logistics containers, the maximum cargo carrying capacity, the round-trip turnaround time from logistics nodes to terminal nodes, and the transit dwell time. Among them, the quantity of supply-demand gap reflects the tightness of container resources under the current delivery task, and the maximum cargo carrying capacity characterizes the upper limit of container load. The round-trip turnaround time refers to the total time it takes for a container to depart from the originating logistics node loaded with goods, arrive at the terminal node to complete unloading, and return to the originating logistics node or designated recycling node empty or loaded with recycled goods. The transit dwell time refers to the cumulative time that a container spends in a stationary state in a non-destination area during transportation due to traffic congestion, transshipment, driver rest, or waiting for loading and unloading operations. The system retrieves historical waybill records from the distributed database and trajectory coordinates reported by IoT positioning terminals. It cleans and aggregates historical container flow trajectory data for the same or similar delivery routes over the past 6 months, extracts the above four types of time features and container configuration features, and then structures and splices them together to construct a multi-dimensional feature input vector.

[0042] In one possible implementation, the system uses a random forest regression prediction model to process the multi-dimensional feature input vector constructed above.

[0043] It's important to note that the Random Forest regression prediction model is a machine learning algorithm based on ensemble learning. Its core mechanism involves constructing a large number of decision trees and combining the predictions from all trees to output the final regression value. Compared to a single decision tree, Random Forest effectively reduces model variance and improves its ability to fit complex nonlinear relationships and its generalization performance by introducing sample and feature perturbations. In this business scenario, the supply-demand gap reflects local scheduling pressure, maximum carrying capacity affects the cargo density and loading / unloading efficiency of a single transport, while round-trip turnaround time and transit dwell time directly characterize the smoothness of the road network and the efficiency of operational nodes. These factors interact in complex ways, and the Random Forest model can adaptively capture the potential mapping relationship between these multidimensional features and the final turnover speed.

[0044] Specifically, during the training phase of the random forest regression prediction model, the system first uses a bootstrap sampling method to randomly sample with replacement from the original training set containing a large number of historical feature vectors, generating multiple different training subsets. For each training subset, the system independently constructs a regression decision tree. During the node splitting process of the decision tree, the system does not search for the optimal splitting attribute among all available features, but rather randomly extracts a feature subset and selects the best splitting feature and splitting threshold within that subset based on the principle of minimizing mean squared error.

[0045] For example, at a certain internal node, the system might randomly select two features: the number of supply-demand gaps and the duration of transit stops. Calculations show that using a two-hour transit stop as a threshold for splitting minimizes the sum of the variances of the target values ​​of the left and right child nodes after the split, i.e., maximizes the reduction in mean squared error. Therefore, this node is split according to this rule. In this way, the system recursively partitions the sample space downwards until a preset stopping condition is met, such as the tree reaching its maximum depth or the number of samples in a leaf node falling below a set threshold. Ultimately, the system trains hundreds of regression decision trees with diverse structures in parallel, collectively forming a random forest regression prediction model.

[0046] For example, after completing model training, the system enters the real-time prediction phase. The system inputs the target logistics container supply-demand gap quantity, maximum cargo carrying capacity, and historical container flow trajectory data extracted in real time under the current delivery network environment into the trained random forest regression prediction model. This multi-dimensional feature input vector simultaneously enters each regression decision tree within the model. Each decision tree guides the input vector to traverse downwards from the root node according to the splitting rules learned during the training phase.

[0047] For example, the first decision tree first determines whether the number of supply-demand gaps is greater than 50. If so, it proceeds to the left subtree to further determine the maximum carrying capacity. The second decision tree might prioritize determining whether the round-trip turnaround time exceeds 8 hours. After a series of logical judgments, the input vector falls into a leaf node of each tree. The predicted value stored in the leaf node comes from the average of the target values ​​of historical samples that fell into that node during training. For example, if a node has accumulated 120 historical records with turnaround times of 2.3 hours, 2.5 hours, 2.1 hours, etc., the average is taken as 2.4 hours as the predicted value. The random forest collects the predicted values ​​from all decision trees and takes the arithmetic mean. Assuming that 100 trees output values ​​of 2.4, 2.6, 2.3, etc., the final average is 2.45 hours, which is the predicted turnaround time of logistics containers under the current delivery network.

[0048] In the actual operation of a large regional logistics distribution center, the system identified a supply-demand gap of 30 target logistics containers for the current period, and determined the maximum cargo carrying capacity of this batch of containers to be 775 kg. Simultaneously, the system extracted historical container flow trajectory data from the distribution center to multiple surrounding terminal outlets, calculating the recent average round-trip turnaround time to be 6.5 hours and the average transit time to be 45 minutes. The system converted these four dimensions into feature vectors and input them into a random forest regression prediction model. The model's 500 decision trees independently evaluated each vector. Some decision trees, based on the feature of a large gap leading to scheduling tension, predicted a decrease in flow speed, outputting a predicted value of 15 km / h; others focused on the feature of short transit times, outputting a predicted value of 18 km / h. The system then averaged these 500 predicted values, ultimately accurately calculating the average flow turnaround speed of this batch of target logistics containers in the current distribution network environment to be 16.5 km / h.

[0049] Understandably, through the data processing and feature mapping process based on the random forest regression prediction model described above, the system can deeply integrate static container physical attributes, dynamic supply and demand gaps, and historical spatiotemporal flow patterns. The final output average turnover rate not only reflects the physical movement speed of a single container but also comprehensively embodies the overall container turnover efficiency of the current delivery network under specific supply and demand pressures and road network conditions.

[0050] Step S104: Based on the average turnover speed and the maximum cargo carrying capacity, perform feasibility screening on heavy cargo delivery tasks that exceed a preset weight threshold included in the cargo delivery order demand, and determine the carrying capacity matching relationship between the target logistics container and the heavy cargo delivery task. The carrying capacity matching relationship includes the number of heavy cargo orders that a single target logistics container can carry and the total number of target logistics containers required.

[0051] Obtain the cargo space volume and cargo stacking limits for the filtered heavy cargo delivery tasks. Based on the maximum cargo carrying capacity of the target logistics container, the cargo space volume, and the cargo stacking limits, calculate the number of heavy cargo orders that a single target logistics container can carry. Specifically: based on the mass constraint, calculate the number of orders that can be loaded under the maximum carrying capacity of the container, N1 = M container / M single cargo; based on the volume constraint, calculate the number of orders that can be loaded under the available space of the container, N2 = V container / V single cargo; based on the stacking limit, calculate the number of orders that can be loaded under the allowed stacking layers, N3 = H limit, where M container is the maximum carrying capacity of the container, M single cargo is the cargo weight of a single order, V container is the available volume of the container, V single cargo is the cargo volume of a single order, and H limit is the maximum allowed stacking layers. The final number of orders that can be carried is N = min(N1, N2, N3). The loading and unloading time Tloading of the heavy goods delivery task is obtained. Based on the average turnover speed Vturnover of the target logistics container, the delivery efficiency E = Vturnover / (1+Tloading) of the heavy goods delivery task is calculated. If E is greater than 0.8, the total number of target logistics containers required is calculated as Q / N, rounded up, based on the number of heavy goods orders N that a single target logistics container can carry and the total number of orders Q included in the heavy goods delivery task. The carrying capacity matching relationship between the target logistics containers and the heavy goods delivery task is determined based on the number of heavy goods orders that a single target logistics container can carry and the total number of target logistics containers required.

[0052] In one implementation, after obtaining the average turnover speed of the target logistics container in the current delivery network environment, the system further performs feasibility screening and matching operations on specific tasks in the demand for goods delivery orders, based on the container's maximum cargo carrying capacity.

[0053] It should be noted that the demand for goods delivery orders includes a massive number of terminal delivery instructions, and the system first needs to identify heavy goods delivery tasks from them.

[0054] Specifically, the system pre-sets a quality threshold, which is typically defined based on the historical quality distribution boundaries of regular express parcels and bulk logistics goods. When the quality of goods in a single delivery order exceeds this pre-set threshold, the system marks it as a heavy goods delivery task. These tasks place more stringent business requirements on the physical load-bearing capacity of containers and the timeliness of delivery. For the identified heavy goods delivery tasks, the system employs a two-dimensional feasibility screening mechanism.

[0055] Specifically, this screening mechanism considers both capacity and timeliness dimensions. In terms of capacity, the system not only extracts the weight of individual shipments in heavy cargo delivery tasks and compares it to the maximum cargo carrying capacity of the target logistics container, eliminating extreme abnormal orders where the weight of a single shipment directly exceeds the physical carrying capacity limit of the container, but also further considers the packaging size characteristics of the heavy cargo to assess the utilization rate of the container's internal three-dimensional space, preventing falsely feasible situations where the weight is within acceptable limits but the volume cannot be accommodated. In terms of timeliness, the system calculates the estimated basic one-way transportation time for the heavy cargo delivery task based on the spatial road network distance between the originating logistics node and the terminal node, combined with the previously calculated average turnover speed. Subsequently, the system adds the estimated dwell time for loading and unloading heavy cargo to this estimated basic one-way transportation time to obtain the complete estimated fulfillment period. The system then rigorously compares this estimated fulfillment period with the promised fulfillment time window issued by the order system. If the expected fulfillment period exceeds the fulfillment window, the heavy goods delivery task is determined to be unable to reach the terminal node on time under the current flow rate, and is thus removed from the candidate task pool of the current batch. After this series of logical checks, the system retains heavy goods delivery tasks that meet both container carrying capacity and delivery time requirements as the candidate task set for subsequent scheduling. Furthermore, for heavy goods delivery tasks that pass the feasibility screening, the system needs to determine the carrying capacity matching relationship between the target logistics container and these tasks.

[0056] Specifically, the system first calculates the number of heavy goods orders that a single target logistics container can carry. The system extracts the average cargo weight or the weight of a specific single shipment from the filtered heavy goods delivery tasks, and calculates based on weight, volume, and stacking constraints. In terms of weight, the maximum cargo carrying capacity of the target logistics container is divided by the weight of a single shipment and rounded down to obtain the number of orders N_weight that can be carried under weight constraints. In terms of volume, the container's cargo space volume is divided by the volume of a single shipment and rounded down to obtain the number of orders N_volume that can be carried under volume constraints. In terms of stacking, the number of orders N_stack that can be carried is calculated based on the cargo stacking height limit. The system takes the minimum of these three values ​​as the maximum number of heavy goods orders that a single container can hold in a single full load, i.e., N_final = min(N_weight, N_volume, N_stack). This calculation process comprehensively considers the non-disassembly attribute of heavy goods and multi-dimensional physical constraints to ensure that loading operations neither exceed load limits nor violate stacking requirements.

[0057] Understandably, after determining the carrying capacity of a single container, the system further calculates the total number of target logistics containers required.

[0058] In one embodiment, the system does not simply divide the total demand for heavy goods orders by the capacity of a single container. Instead, it deeply integrates the dynamic variable of average turnover speed to reflect the reuse characteristics of containers over time. The system first obtains the total available time for the current delivery period from the scheduling management module. This time is determined by the start and end times of delivery shifts. Simultaneously, the system extracts the average delivery distance from the originating node to each terminal node from the historical delivery database. This distance is calculated based on actual path data from the past 30 days. Based on the average turnover speed and the average delivery distance, the system calculates the standard time span required for a single round trip. The system divides the total available time by this standard time span to obtain the theoretical number of turnovers N. If N is not an integer, the system uses the integer part as the complete number of turnovers and calculates the time percentage corresponding to the decimal part. This percentage is multiplied by the capacity of a single container to obtain the capacity contribution of the last incomplete turnover, thus deriving the comprehensive capacity contribution value of a single container throughout the entire delivery period. This contribution value characterizes the maximum task handling capacity of a single container considering turnover speed limitations. Finally, the system aggregates the total number of all heavy cargo delivery tasks that passed the feasibility screening, divides it by the comprehensive transportation capacity contribution value, and rounds up to calculate the total number of target logistics containers actually required to complete the batch of heavy cargo delivery tasks. This calculation logic combines static carrying capacity attributes with dynamic flow rate.

[0059] For example, in the home appliance delivery scenario of a regional distribution center, the system sets a preset weight threshold of 30 kg and selects 120 large home appliance heavy goods delivery tasks. All of these tasks have passed the timeliness feasibility verification based on the average turnover speed. The maximum cargo carrying capacity of the target logistics container is known to be 500 kg, the average weight of a single home appliance is 80 kg, the average volume of a single home appliance is 0.8 cubic meters, and the effective loading space of the container is 5 cubic meters. The system first calculates the theoretical carrying capacity as 6.25 orders based on the weight constraint, and then calculates the theoretical carrying capacity as 6.25 orders based on the volume constraint. Considering the stability requirements of cargo stacking and the space utilization loss during actual loading operations, the system rounds down both calculation results, ultimately determining that a single container can carry a maximum of 6 heavy goods orders per trip. Simultaneously, based on the current average turnover speed of 20 km / h, the system calculates that within an 8-hour delivery period, a single container can complete two complete round trips. Therefore, the overall transport capacity contribution of a single container is 12 orders, where N = min(⌊M / m⌋, ⌊V / v⌋) × T, where N is the overall transport capacity contribution, M is the maximum load capacity of the container, m is the mass of a single item, V is the effective space of the container, v is the volume of a single item, and T is the number of turnovers. The system ultimately divides the total of 120 tasks by 12, determining that the total number of target logistics containers required to complete this batch of appliance deliveries is 10.

[0060] Step S105: Obtain the statistical value of the number of unrecycled empty containers remaining in each terminal node during the historical delivery cycle according to the bearing matching relationship. If the statistical value of the number of unrecycled empty containers remaining is greater than the preset threshold of the number of remaining containers, then determine the corresponding terminal node as the recycling node that will prioritize the recycling of empty containers.

[0061] Based on the carrying capacity matching relationship, extract the statistical value of the number of empty containers remaining in each terminal node during the historical period. If the statistical value is greater than a preset threshold for the number of remaining containers, such as 100 empty containers, then the corresponding terminal node is determined to be a recycling node that prioritizes empty container recycling. Obtain the geographical coordinates of the recycling nodes, and use a hierarchical clustering algorithm to cluster the geographical coordinates to obtain multiple recycling node clusters. For each recycling node cluster, obtain the road network data between each recycling node within the cluster, and extract the distance and time values ​​between adjacent recycling nodes based on the road network data. Use the nearest neighbor algorithm or genetic algorithm to solve the traveling salesman problem, calculate the shortest traversal path within the recycling node cluster, and generate an initial route map for the recycling node cluster. Obtain the schedulable recycling vehicles and their maximum loading capacity from the vehicle management module. If the total number of empty containers remaining in the recycling node cluster is less than or equal to the maximum loading capacity, and the number of remaining containers in any single recycling node does not exceed the maximum loading capacity, then generate a corresponding recycling dispatch order based on the initial route map. If the total amount of waste within a cluster exceeds the load capacity of a single vehicle or there is a single point of overload, the recycling node cluster will be split into multiple sub-clusters, and multiple recycling dispatch orders will be generated for each sub-cluster.

[0062] After parsing the load matching data determined in the pre-processing stage, the system connects to the historical delivery record database to extract the total number of fully loaded logistics containers received by each terminal node within the past 7-day historical delivery cycle. Subsequently, the system retrieves the number of actually scanned and returned empty containers from the reverse logistics database for the same period. The system uses the formula V_retention = V_issued - V_recovered - V_abnormal to calculate the statistical value of the number of unrecovered empty containers remaining at each terminal node, where V_retention is the number of containers remaining, V_issued is the total number of issues, V_recovered is the number of containers recovered, and V_abnormal is the number of abnormal containers marked by the system as damaged, lost, or misappropriated. To ensure the accuracy of recovery scheduling, the system generates a preset retention threshold V_threshold based on a 1.5 elastic fault tolerance coefficient multiplied by the daily average order throughput Q_daily for each terminal node. The calculation formula is V_threshold = Q_daily × 1.5. This coefficient is set based on industry experience to cover container turnover delays caused by order fluctuations on weekends and holidays. The system compares the statistical value of the retention quantity for each node with the corresponding preset threshold. Taking a terminal node at a municipal central hospital as an example, the system reads from the historical delivery record database that the node received 500 target logistics containers in the past 7 days, while the reverse logistics database records 320 recovered empty containers and 5 abnormal containers marked by the system. The system calculates that V_retention = 500 - 320 - 5 = 175 containers. Simultaneously, the system extracts the node's daily average Q_retention rate as 100 containers, calculating the V_threshold = 100 × 1.5 = 150 containers. The system executes comparison logic, determines that the retention statistical value of 175 containers is greater than the preset threshold of 150, and then marks the municipal central hospital terminal node in the scheduling system as a priority node for empty container recovery.

[0063] Step S106: Based on the vehicle's load capacity and volume constraints, and combined with the demand for heavy cargo delivery and empty container recycling, a load-bearing matching relationship is established between vehicles and delivery / recycling tasks. This matching relationship records the specific delivery and recycling node combinations that each vehicle can carry. For the recycling nodes that prioritize empty container recycling and the target logistics container supply-demand gap, a genetic algorithm's chromosome encoding mechanism is used to perform crossover and mutation optimization operations on the delivery routes corresponding to the load-bearing matching relationship. This crossover and mutation optimization operation includes exchanging the order of path nodes and adjusting the insertion position of recycling nodes, resulting in an optimized vehicle travel path that simultaneously includes both heavy cargo delivery and empty container recycling tasks.

[0064] The algorithm obtains the priority recycling nodes for empty container recycling, the supply-demand gap of target logistics containers, and the delivery routes corresponding to the matching relationships. Using heavy cargo delivery tasks and empty container recycling operations as gene nodes, it employs a genetic algorithm's chromosome coding mechanism to arrange these gene nodes, generating an initial gene coding sequence. For this initial gene coding sequence, two parent gene coding sequences are selected according to a preset crossover probability. Fragment exchanges are performed on some gene nodes in the parent gene coding sequences to obtain a crossover recombination sequence. This crossover recombination sequence is used as the input sequence for a mutation operation. One to three empty container recycling operation nodes are randomly selected from the sequence as mutation nodes according to a preset mutation probability. The insertion positions of these mutation nodes in the sequence are adjusted to obtain a mutated optimized sequence. Based on the weight W of the goods in each heavy cargo delivery task in the mutation optimization sequence and the product of the number N of the supply and demand gap of the target logistics containers corresponding to each recycling node and the weight w of a single empty container, the cumulative load capacity of a single vehicle is calculated as L = ΣW + Σ(N×w). If L is less than the maximum load capacity threshold of 8000 kg for vehicles, the mutation optimization sequence is decoded. The sequence is expanded according to the order of gene nodes in the sequence to the access order of the actual delivery location and recycling location. Combined with the delivery route, a complete path including the starting point, each delivery node, each recycling node and the end point is generated, resulting in an optimized vehicle driving path that includes both heavy cargo delivery tasks and empty container recycling operations.

[0065] In one implementation, the system needs to restructure existing delivery routes in response to the aforementioned identified priority recycling nodes for empty container recycling and the target supply-demand gap in logistics containers.

[0066] Specifically, a heuristic algorithm is introduced to perform deep optimization of the initial delivery route corresponding to the matching relationship, so as to achieve efficient coordination between forward delivery and reverse recycling.

[0067] The system employs a genetic algorithm to optimize delivery routes, using natural numbers to construct chromosomes. The delivery center is coded as 0, regular heavy goods delivery nodes are coded from 1 to n according to the delivery order, and recycling nodes that prioritize empty container collection are coded from n+1 to m. The chromosome sequence represents the order in which delivery vehicles visit each node. For example, the chromosome [0,3,7,12,5,0] indicates that the vehicle departs from the delivery center, visits nodes numbered 3, 7, 12, and 5 in sequence, and then returns to the delivery center. The coding length is dynamically determined based on the total number of nodes to be delivered, ensuring coverage of all delivery and recycling task nodes.

[0068] For example, number 0 represents a regional distribution center, numbers 1 to 9 represent community fresh food stores with only heavy goods delivery tasks, and numbers 10 to 15 represent fresh food pre-positioning warehouses marked as priority recycling nodes. A complete chromosome represents the travel path sequence of a transport vehicle. For example, the sequence 0-3-12-5-0 indicates that the vehicle departs from the distribution center, passes through store number 3 to deliver heavy goods, then proceeds to pre-positioning warehouse number 12 to collect empty containers, then proceeds to store number 5 for delivery, and finally returns to the distribution center. This encoding mechanism can intuitively reflect the flow order of vehicles between various nodes, providing a basic data structure for subsequent path optimization. After completing the chromosome encoding of the initial population, the system uses a single-point crossover method to perform crossover operations on the chromosomes, with a crossover probability set to 0.8. That is, two chromosomes are randomly selected in the population, and gene fragments are exchanged at a randomly selected crossover point to generate new path combinations. The mutation operation uses a swap mutation method, with a mutation probability set to 0.05. Two gene positions in the chromosome are randomly selected for exchange, breaking the local optimum of the existing path and exploring a better node access order.

[0069] For example, the crossover and mutation optimization operation includes path node order swapping. Path node order swapping refers to randomly selecting two or more gene positions on the same chromosome or between two different chromosomes and exchanging their corresponding node identifiers to generate a new path sequence. In a logistics scheduling scenario, the system identifies unreasonable detours by calculating the path deviation rate. Specifically, for any three consecutive delivery points A, B, and C in the path, the ratio of the actual path distance Dactual from A to C via B to Dshortest is calculated, i.e., the deviation rate R = Dactual / Dshortest. When R is greater than 1.3, the path segment is determined to have unreasonable detours, and node B is marked as an optimizable node. The system traverses all consecutive node combinations along the entire path, counts all nodes with deviation rates exceeding a threshold, and prioritizes performing crossover and mutation operations on these nodes to reduce the total delivery distance by adjusting the node order.

[0070] For example, in the parent chromosome sequence 0 then 2 then 8 then 4 then 0, store number 2 is 15.3 kilometers away from store number 8, while store number 8 is only 4.7 kilometers away from store number 4. The crossover mutation operator swaps the visit order of store number 8 and store number 4, generating the offspring chromosome sequence 0 then 2 then 4 then 8 then 0. After the swap, the total path mileage is reduced from 42.6 kilometers to 36.8 kilometers, a reduction of 5.8 kilometers, resulting in a corresponding decrease in delivery energy consumption of 13.6%. In each iteration, the operator randomly selects two nodes that are not the start or end point and swaps their positions, retaining the sequence with the shorter total mileage as the new generation chromosome. This operation is repeated until no better solution is generated for 20 consecutive generations.

[0071] In one possible implementation, the cross-mutation optimization operation further includes adjusting the insertion position of the recycling node. This operation targets recycling nodes that prioritize empty container recycling, determining the timing of recycling intervention based on the supply-demand gap of the target logistics containers. When the system performs the insertion position adjustment, it first calculates the remaining loading space V_remaining at the current node, where V_remaining equals the vehicle's total capacity minus the space occupied by already loaded containers. Then, it calculates the total demand D_total for all subsequent delivery nodes. When V_remaining is greater than or equal to D_total plus the planned number of empty containers to be recycled, the recycling node is moved forward to the current position. When V_remaining is less than D_total, the recycling node is moved backward until after part of the delivery task is completed. For example, if the vehicle's total capacity is 50 standard container slots, and 30 are already loaded, then V_remaining is 20. If the demands of the next 3 delivery nodes are 5, 8, and 6 respectively, and D_total is 19, and 10 empty containers are planned to be recycled, since 20 is greater than or equal to 19 plus 10, this condition does not hold, so the recycling node is moved backward until at least one delivery node is completed before execution.

[0072] For example, after a transport vehicle completes a partial delivery of heavy goods, it frees up enough space in its cargo compartment to accommodate 50 standard crates. At this point, the supply-demand gap for the target logistics containers is exactly 50. When the system detects that the vehicle's remaining capacity V equals the supply-demand gap Q, it triggers a mutation operation. Specifically, the operation involves extracting the recycling node with the largest container backlog from the sequence of nodes to be visited (in this example, priority recycling node number fourteen), removing it from the end of the path, and inserting it into the next position after the currently visited node. In the adjusted path sequence, the vehicle's next stop after unloading the heavy goods is this recycling node, where it directly loads empty containers and continues subsequent deliveries. This clears the backlog of empty containers at the recycling node and simultaneously fills the supply-demand gap.

[0073] The fitness function F is calculated by weighting three indicators: total path distance D, vehicle load factor L, and timeliness of empty container recovery T. F = 0.5 × D + 0.3 × L + 0.2 × T, where D is the normalized sum of all vehicle mileage, L is the average ratio of actual load to rated load for each vehicle, and T is the deviation of the empty container's dwell time at the customer from the set threshold of 72 hours. The system sorts chromosomes from highest to lowest fitness value, retaining the top 60% as parents and discarding the bottom 40%, converging after 500 iterations. The optimal chromosome output by decoding is the vehicle's travel path, which clearly defines the flow order of vehicles between delivery and recovery nodes, enabling simultaneous delivery of heavy goods and recovery of empty containers in a single transport.

[0074] Step S107: Obtain the total expected container turnover quantity corresponding to the optimized vehicle driving route. The total expected container turnover quantity includes the number of new delivery target logistics containers and the number of empty containers to be recycled. If the total expected container turnover quantity is greater than the supply-demand gap of the target logistics containers, then generate a container scheduling execution instruction containing delivery time nodes and recycling time nodes according to the optimized vehicle driving route. The delivery time nodes and recycling time nodes are sorted according to the priority labels in the target time axis sequence. The task nodes with smaller priority values ​​are executed first. Specifically, the task nodes with priority label 1 are arranged at the beginning of the time axis, and the task nodes with priority labels 2 and 3 are sequentially postponed to ensure that key delivery and recycling tasks are completed first. At the same time, the priority labels are associated with the execution order field of each time node to form an executable scheduling instruction sequence.

[0075] Optimize vehicle travel routes by obtaining the route data structure, which includes the latitude and longitude coordinates of each station. Extract the spatial coordinate sequence of the stations and calculate the physical distance set between adjacent stations using the coordinate sequence. Obtain the physical distance set D={d1,d2,...,dn}, where di represents the distance from the i-th station to the (i+1)-th station. Using a preset vehicle speed v=60 km / h, calculate the basic travel time sequence T={t1,t2,...,tn} between adjacent stations using ti=di / v. Obtain the basic travel time sequence, adding 30 minutes of loading and unloading time to the delivery time node for delivery stations and adding 20 minutes of loading and unloading time to the recycling time node for recycling stations. Obtain the delivery time node set P={p1,p2,...,pm} and the recycling time node set R={r1,r2,...,rk}, and construct the initial time axis sequence S={s1,s2,...,sq} by chronological order. Each sequence element contains three pieces of information: timestamp, station identifier, and job type. Obtain the number of new target logistics containers and the number of empty containers to be recycled, and sum them to obtain the total expected container turnover C. Obtain the total expected container turnover C and the supply-demand gap G of target logistics containers. If C is greater than G, it indicates that the vehicle capacity is limited and cannot complete all tasks. In this case, the timeline sequence needs to be prioritized. A decision tree algorithm is used to classify the initial timeline sequence. The input features of the decision tree include the importance score of the site, the urgency score of the container demand, and the remoteness score of the geographical location. The classification rule calculates a comprehensive score based on the feature weights and outputs three categories of labels: high priority, medium priority, and low priority, resulting in a target timeline sequence with priority labels. Obtain the target timeline sequence with priority labels, prioritize the scheduling of delivery time nodes and recycling time nodes corresponding to high priority labels, and generate container scheduling execution instructions containing site order, arrival time, and job type.

[0076] The system first reads the determined optimized vehicle driving route data. The order parsing algorithm receives the route data as input, extracts the demand details of each business node by parsing the attribute fields of the route nodes, and outputs a structured list containing node type, container demand quantity, and geographic coordinates. Specifically, the system identifies the number of new delivery target logistics containers for Store 1 and Store 3 in the route as 120 and 80 respectively, and obtains the number of empty containers to be recycled in Store 2 as 150 through RFID reading and writing devices deployed in each front warehouse. The system performs an accumulation calculation to obtain the total expected container turnover quantity N for this route as 350. The system calls the inventory management module to obtain the target logistics container supply and demand gap M of the current regional distribution center, which is 300. The system performs a judgment that N is greater than M, confirming that 350 is greater than 300, thus meeting the trigger condition. Based on this judgment result, the system starts the time window calculation module, which receives the route node coordinate sequence, the standard vehicle driving speed of 45.5 km / h, and the loading and unloading operation time of 0.5 minutes per container as input parameters. The system is set to depart at 08:00. Based on the route distance, the journey to store number one is 22.75 kilometers, taking 30 minutes. Delivering 120 containers takes 60 minutes, and the delivery completion timestamp for this node is set at 08:30. The arrival time of the second pre-positioning warehouse is calculated at 09:15, and the time to retrieve 150 containers is 75 minutes, with a retrieval completion timestamp of 09:45. The arrival time of store number three is 10:25, and the delivery of 80 containers takes 40 minutes, with a delivery completion timestamp of 10:50. Based on the priority labels output by the decision tree algorithm, the system sets a tolerance range of ±5 minutes for the timestamp of store number one (high priority) and ±10 minutes for the timestamps of pre-positioning warehouses two and three (medium priority). The system encapsulates the vehicle's unique identifier, node coordinates, timestamp, and priority tolerance into a JSON format data packet and sends it to the task receiving module of the vehicle terminal via the 4G network, generating a container scheduling execution instruction containing delivery and retrieval time nodes.

[0077] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from the spirit or essential characteristics of this application. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of this application is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within this application. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A smart scheduling method for intelligent fruit and vegetable eco-friendly baskets, characterized in that, include: Acquire the material delivery task demand of each circulation node within the target time period and the structural damage monitoring data of the smart fruit and vegetable environmental protection basket; The structural integrity health level of the smart fruit and vegetable eco-friendly basket is determined by the structural loss monitoring data; the real-time inventory quantity of available smart fruit and vegetable eco-friendly baskets at each circulation node is obtained according to the material distribution task demand. The maximum material carrying capacity of the smart fruit and vegetable eco-friendly basket is determined based on the structural integrity health level. A quantity difference is compared between the real-time inventory of available smart fruit and vegetable eco-friendly baskets and the material delivery task demand to obtain the supply-demand gap of smart fruit and vegetable eco-friendly baskets at each circulation node during different delivery time periods. Based on the supply-demand gap of smart fruit and vegetable eco-friendly baskets and the maximum material carrying capacity, a predictive model is used to process the historical vehicle circulation trajectory data corresponding to the supply-demand gap of smart fruit and vegetable eco-friendly baskets to obtain the average circulation turnover speed of smart fruit and vegetable eco-friendly baskets in the current delivery network environment. Based on the average circulation turnover speed and the maximum material carrying capacity, feasibility screening is performed on heavy-load delivery tasks exceeding a preset quality threshold included in the material delivery task demand to determine the carrying capacity matching relationship between the smart fruit and vegetable eco-friendly baskets and the heavy-load delivery tasks. Based on the carrying capacity matching relationship, the statistical value of the number of unrecovered empty vehicles remaining at each end node within the historical delivery cycle is obtained. If the statistical value of the number of unrecovered empty vehicles remaining is greater than the preset retention threshold, the corresponding end node is determined as the priority recycling node for empty vehicle recycling. For the recycling nodes that prioritize empty vehicle recovery and the supply-demand gap of the smart fruit and vegetable eco-friendly baskets, an optimization algorithm is used to perform cross-mutation optimization on the delivery routes corresponding to the load matching relationship to obtain an optimized equipment travel path that simultaneously includes heavy-load delivery tasks and empty vehicle recovery tasks; the total expected number of vehicles traveling along the optimized equipment travel path is obtained; if the total expected number of vehicles traveling is greater than the supply-demand gap of the smart fruit and vegetable eco-friendly baskets, a vehicle scheduling execution instruction including delivery time nodes and recycling time nodes is generated based on the optimized equipment travel path.

2. The intelligent scheduling method for the intelligent fruit and vegetable eco-friendly basket as described in claim 1, characterized in that, The process of determining the structural integrity health level of the smart fruit and vegetable eco-friendly basket using the structural loss monitoring data includes: acquiring the initial inventory quantity of each circulation node, as well as the node distance and load distribution data between circulation nodes; obtaining a loss degree parameter by weighting and summing the carrier frame deformation and load-bearing plate wear depth included in the structural loss monitoring data with preset weights; determining the structural integrity health level of the smart fruit and vegetable eco-friendly basket based on the loss degree parameter; if the structural integrity health level is severely damaged, removing the corresponding carrier from the initial inventory quantity to obtain the real-time inventory quantity of available smart fruit and vegetable eco-friendly baskets at each circulation node; determining the replenishment quantity for each circulation node by the difference between the material delivery task demand and the real-time inventory quantity of available smart fruit and vegetable eco-friendly baskets; classifying and mapping the replenishment quantity and node distance using a pre-trained support vector machine model to obtain the allocation priority between each circulation node; constructing an allocation matrix based on the allocation priority; extracting the corresponding transportation capacity matching value based on the allocation matrix; and determining the final delivery route by combining the transportation capacity matching value with the load distribution data.

3. The intelligent scheduling method for the intelligent fruit and vegetable eco-friendly basket as described in claim 1, characterized in that, The step of comparing the real-time inventory of available smart fruit and vegetable eco-friendly baskets with the required quantity for the material delivery task to obtain the supply-demand gap of smart fruit and vegetable eco-friendly baskets at each circulation node during different delivery periods includes: extracting the corresponding gap severity level based on the number of supply-demand gaps; if the gap severity level is higher than a preset level threshold, using a random forest algorithm to extract features from the number of supply-demand gaps of smart fruit and vegetable eco-friendly baskets and the different delivery periods to obtain the emergency replenishment quantity for each circulation node; constructing a spatial distance matrix by combining the emergency replenishment quantity with the geographical coordinate information of each circulation node; calculating the transfer cost between each circulation node based on the spatial distance matrix; obtaining the number of spare vehicles at each circulation node; determining the vehicle scheduling priority for each circulation node by performing a weighted summation based on the transfer cost and the number of spare vehicles; generating target scheduling instructions for each circulation node based on the vehicle scheduling priority; and controlling the transportation equipment to perform vehicle transfer operations according to the target scheduling instructions.

4. The intelligent scheduling method for the intelligent fruit and vegetable eco-friendly basket as described in any one of claims 1-3, characterized in that, The step of using a predictive model to process historical vehicle circulation trajectory data corresponding to the supply-demand gap of the smart fruit and vegetable eco-friendly baskets, and obtaining the average circulation speed of the smart fruit and vegetable eco-friendly baskets in the current delivery network environment, includes: obtaining the supply-demand gap quantity and the maximum material carrying capacity of the smart fruit and vegetable eco-friendly baskets; filtering out historical vehicle circulation trajectory data that meets the conditions based on the supply-demand gap quantity and the maximum material carrying capacity, the historical vehicle circulation trajectory data including the round-trip circulation time and the transit dwell time from the circulation node to the terminal node; extracting the round-trip circulation time and the transit dwell time from the historical vehicle circulation trajectory data; calculating the average circulation time of each circulation node based on the round-trip circulation time and the transit dwell time; if the average circulation time is greater than a preset time threshold, obtaining real-time traffic data in the current delivery network environment; A random forest regression prediction model is used to extract and fuse features from the average turnover time and the real-time road condition data to obtain the dynamic turnover characteristics of the smart fruit and vegetable eco-friendly basket. The average turnover speed of the smart fruit and vegetable eco-friendly basket in the current delivery network environment is determined by weighted calculation based on the dynamic turnover characteristics and the maximum material carrying capacity.

5. The intelligent scheduling method for the intelligent fruit and vegetable eco-friendly basket as described in any one of claims 1-3, characterized in that, The process of determining the load-bearing matching relationship between the smart fruit and vegetable eco-friendly basket and the heavy-load delivery task includes: extracting heavy-load delivery tasks exceeding a preset quality threshold from the material delivery task demand; obtaining the material space volume and material stacking limit of the heavy-load delivery task; calculating the number of heavy-load orders that a single smart fruit and vegetable eco-friendly basket can carry based on the maximum material carrying capacity, the material space volume, and the material stacking limit; obtaining the node loading and unloading time of the heavy-load delivery task; calculating the delivery flow efficiency of the heavy-load delivery task based on the average turnover speed and the node loading and unloading time; if the delivery flow efficiency is greater than a preset efficiency threshold, calculating the total number of smart fruit and vegetable eco-friendly baskets required based on the number of heavy-load orders that a single smart fruit and vegetable eco-friendly basket can carry and the total number of orders included in the heavy-load delivery task; and determining the load-bearing matching relationship between the smart fruit and vegetable eco-friendly basket and the heavy-load delivery task based on the number of heavy-load orders that a single smart fruit and vegetable eco-friendly basket can carry and the total number of smart fruit and vegetable eco-friendly baskets required.

6. The intelligent scheduling method for the intelligent fruit and vegetable eco-friendly basket as described in any one of claims 1-3, characterized in that, If the statistical value of the number of unrecovered empty vehicles is greater than a preset threshold, then the corresponding end node is determined as the priority recovery node for empty vehicle recovery, including: extracting the number of empty vehicles remaining at each end node in the historical delivery cycle based on the load matching relationship; calculating the statistical value of the number of empty vehicles remaining; if the statistical value is greater than the preset threshold, then determining the corresponding end node as the priority recovery node for empty vehicle recovery; obtaining the geographical coordinates of the recovery node; using a hierarchical clustering algorithm to cluster the geographical coordinates to obtain multiple recovery point clusters; for each recovery point cluster, obtaining road network data between each recovery node within the cluster; extracting the distance and time values ​​between adjacent recovery nodes based on the road network data; using the Dijkstra algorithm to calculate the shortest traversal path within the recovery point cluster based on the distance and time values, generating an initial route map for the recovery point cluster; and obtaining the maximum loading capacity of the transportation equipment performing the recovery task. If the total number of empty vehicles remaining in the cluster of recycling points is less than the maximum loading capacity, a corresponding recycling dispatch order is generated based on the initial route map.

7. The intelligent scheduling method for the intelligent fruit and vegetable eco-friendly basket as described in any one of claims 1-3, characterized in that, The optimization algorithm is used to perform cross-mutation optimization on the delivery routes corresponding to the load matching relationship to obtain an optimized equipment driving path that simultaneously includes heavy-load delivery tasks and empty vehicle recovery tasks. This includes: obtaining heavy-load delivery tasks and recovery tasks as gene nodes; arranging the gene nodes using the chromosome encoding mechanism of a genetic algorithm to generate an initial gene coding sequence; selecting two parent gene coding sequences for the initial gene coding sequence according to a preset crossover probability; exchanging fragments of some gene nodes in the parent gene coding sequences to obtain a crossover recombination sequence; randomly selecting mutated nodes for the crossover recombination sequence according to a preset mutation probability; adjusting the insertion position of the recovery node corresponding to the mutated node to obtain a mutated optimized sequence; calculating the cumulative load per vehicle corresponding to the mutated optimized sequence based on the heavy-load delivery tasks in the mutated optimized sequence and the supply-demand gap of the smart fruit and vegetable environmental protection baskets; if the cumulative load per vehicle is less than a preset maximum equipment load threshold, then decoding the mutated optimized sequence to obtain an optimized equipment driving path that simultaneously includes heavy-load delivery tasks and empty vehicle recovery tasks.

8. The intelligent scheduling method for the intelligent fruit and vegetable eco-friendly basket as described in any one of claims 1-3, characterized in that, The step of generating vehicle scheduling execution instructions containing delivery time nodes and collection time nodes based on the optimized equipment travel path includes: extracting spatial coordinate sequences of each station based on the optimized equipment travel path; calculating the physical distance set between adjacent stations using the spatial coordinate sequences; calculating the basic travel time sequence between adjacent stations based on the physical distance set and a preset equipment travel speed threshold; accumulating the delivery time node and collection time node corresponding to the optimized equipment travel path based on the basic travel time sequence and a preset station loading and unloading operation time; constructing an initial time axis sequence based on the delivery time node and collection time node; obtaining the number of newly delivered smart fruit and vegetable environmental protection baskets and the number of empty collection vehicles; summing the number of newly delivered smart fruit and vegetable environmental protection baskets and the number of empty collection vehicles to obtain the expected total number of vehicle circulations; if the expected total number of vehicle circulations is greater than the supply-demand gap of smart fruit and vegetable environmental protection baskets, then using a decision tree algorithm to classify the initial time axis sequence to obtain a target time axis sequence with priority labels; and generating vehicle scheduling execution instructions containing delivery time nodes and collection time nodes based on the target time axis sequence with priority labels.

9. Intelligent fruit and vegetable eco-friendly basket, characterized in that, include: The basket includes a bottom and side walls surrounding the bottom, the side walls and the bottom enclosing a storage space for holding fruits and vegetables; A reinforcing structure is provided on the bottom and / or the side wall to improve the load-bearing capacity of the basket. A structural loss monitoring tag is set on the basket body to provide structural loss monitoring data of the smart fruit and vegetable environmental protection basket, so as to determine the structural integrity health level and maximum material load capacity of the smart fruit and vegetable environmental protection basket. The intelligent fruit and vegetable eco-friendly basket is scheduled using the intelligent scheduling method for the intelligent fruit and vegetable eco-friendly basket as described in any one of claims 1-8.