Intelligent logistics scheduling method, device and equipment for raw tobacco and storage medium

Abstract

The application discloses a method, device and equipment for intelligent logistics scheduling of raw tobacco and a storage medium, and relates to the technical field of tobacco. The method is used for solving the existing problems in path planning and scheduling of raw tobacco logistics workshop transport vehicles, breaking the excessive dependence on preset parameters and manual experience, establishing a dynamic path planning and scheduling system suitable for workshop operation scenarios, adaptively responding to workshop environment and task demand, improving the stability, executability and scheduling efficiency of path planning, and adapting to the global optimal scheduling demand of the raw tobacco intelligent logistics workshop. In the method, S1 is used for constructing a global environment map to provide an accurate spatial basis; S2 is used for constructing a mixed initial solution space to narrow the optimization search range; S3 is used for realizing dynamic collaborative updating of algorithm parameters to relieve the local optimal problem; S4 is used for generating an enhanced probability calculation model to improve the global search capability; S5 is used for completing path iterative search to accelerate algorithm convergence; and S6 is used for outputting a global optimal scheduling path to guarantee the rationality and optimality of the scheme.

Description

Technical Field

[0001] This application relates to the field of tobacco technology, and in particular to a smart logistics scheduling method, device, equipment and storage medium for raw tobacco. Background Technology

[0002] Raw tobacco is the core raw material for tobacco production, and its logistics scheduling runs through the entire process of receiving, storing, transferring, and distributing raw tobacco. Raw tobacco logistics workshops are typically characterized by irregular operating spaces, complex functional area divisions, and a large number of concurrent transportation tasks, which places high demands on the route planning and task scheduling of transportation vehicles within the workshop.

[0003] With the continuous improvement of intelligent manufacturing in the tobacco industry, how to achieve efficient and accurate path scheduling of raw tobacco transport vehicles in complex workshop environments has become a core research topic in the construction of raw tobacco logistics workshops.

[0004] However, in the process of route planning and scheduling, the transport vehicles in existing raw tobacco logistics workshops generally rely on factory-preset parameters or the established experience of operators to make route decisions.

[0005] The above methods lack the ability to autonomously perceive and dynamically respond to real-time changes in the workshop environment. The path planning results are unstable, have large errors, and are even difficult to execute. As a result, it is difficult to adaptively generate the optimal scheduling path under different workshop layouts, different task loads, and different obstacle distributions. This leads to unstable transportation efficiency and insufficient path planning accuracy, which cannot meet the actual needs of the raw tobacco logistics workshop for global optimal scheduling. Summary of the Invention

[0006] The main objective of this application is to provide a smart logistics scheduling method, device, equipment, and storage medium for raw tobacco, in order to solve the problem in the prior art that it is difficult to adaptively generate the optimal scheduling path under different workshop layouts, different task loads, and different obstacle distributions, resulting in unstable transportation efficiency and insufficient path planning accuracy.

[0007] To achieve the above objectives, this application provides the following technical solution: A smart logistics scheduling method for raw tobacco, wherein the smart logistics scheduling method is applied in a smart logistics workshop for raw tobacco, the smart logistics scheduling method comprising: Step S1: Based on the work space and obstacle distribution data of the raw tobacco smart logistics workshop, a global environment map of raw tobacco logistics operations is constructed by fusing topology modeling and grid modeling. Step S2: Obtain the start and end node data of the raw tobacco transportation task, and initialize the basic parameters of the ant colony algorithm and the goose colony optimization algorithm in combination with the global environment map of the raw tobacco logistics operation to construct a hybrid initial solution space for path search. Step S3: Dynamically adjust the pheromone utilization coefficient and heuristic weight of the ant colony algorithm through the single-leg balance strategy of the goose flock optimization algorithm, and dynamically and collaboratively update the path search rules based on the parameters generated by the path search hybrid initial solution space. Step S4: Incorporate a random perturbation strategy into the path selection mechanism of the ant colony algorithm, and generate an enhanced path transition probability calculation model based on the path search rules. Step S5: The path search space is dynamically compressed by the dimension scaling factor of the goose flock optimization algorithm to perform path iterative search and global pheromone update of the ant colony based on the enhanced path transition probability calculation model. Step S6: Based on the results of the path iterative search and pheromone global update, the globally optimal scheduling path for the raw tobacco transportation task is obtained.

[0008] Beneficial effects of steps S1 to S6: This application addresses existing problems in the route planning and scheduling of transport vehicles in raw tobacco logistics workshops. It eliminates excessive reliance on preset parameters and human experience, establishing a dynamic route planning and scheduling system adapted to workshop operation scenarios. The method enables adaptive responses to workshop environment and task requirements, improves the stability and executability of route planning results, optimizes the ability to generate scheduling routes under different operation scenarios, ensures scheduling efficiency and planning accuracy in raw tobacco logistics transportation, and adapts to the practical application needs of intelligent raw tobacco logistics workshops for globally optimal scheduling. In this process, the global environment map constructed in step S1 provides a precise and unified spatial foundation for path planning, ensuring the adaptability of the planned path to the actual operation scenario in the workshop; the construction of the hybrid initial solution space completed in step S2 provides a reasonable initial foundation for algorithm iteration, narrowing the search range for subsequent optimization; the dynamic collaborative update of algorithm parameters implemented in step S3 can adapt to changes in the search state during the iteration process, alleviating the local optimum problem caused by fixed parameters; the enhanced path transition probability calculation model generated in step S4 can enrich the diversity of path selection and improve the global search capability of the algorithm; the path iterative search and pheromone update completed in step S5 can realize the dynamic optimization of the search space and accelerate the algorithm convergence process; and the globally optimal scheduling path output in step S6 can match the core requirements of the raw tobacco transportation task, ensuring the rationality and optimality of the final scheduling scheme.

[0009] As a further improvement to this application, step S1 involves constructing a global environment map of the raw tobacco logistics operation based on the distribution data of the work space and obstacles in the intelligent logistics workshop, through the fusion of topology modeling and mesh modeling, including: Step S1.1: Collect the three-dimensional point cloud data and two-dimensional planar drawing data of the raw tobacco intelligent logistics workshop, perform voxel filtering and coordinate system alignment on the three-dimensional point cloud data, and orthographically project the processed data onto the two-dimensional planar coordinate system to generate the basic two-dimensional planar data of the workshop. Step S1.2: Extract obstacle contours from the two-dimensional planar basic data of the workshop, perform morphological dilation operation on the obstacle contours to expand the safety margin, and generate workshop space constraint data with obstacle dilation boundaries; Step S1.3: Extract the key node coordinates of the workshop space constraint data, and establish an adjacency relationship table based on the access relationship between nodes to generate the original tobacco logistics workshop topology structure data; Step S1.4: Divide the workshop space constraint data into grids at a fixed resolution, and determine the overlap relationship between each grid and obstacles and mark the passage status to obtain the original tobacco logistics workshop grid map data; Step S1.5: Map the node coordinates of the topology map structure data of the raw tobacco logistics workshop to the corresponding grid positions of the grid map data of the raw tobacco logistics workshop, so as to establish a bidirectional index mapping relationship between topology nodes and grids, and obtain a topology-grid bidirectional index table. Step S1.6: Retrieve the grid accessibility status on the path connecting adjacent nodes of the topology graph structure data according to the topology-grid bidirectional index table, delete the topology edges with inaccessible grids, and obtain the topology-grid fusion map after spatial consistency verification. Step S1.7: Calculate the passage cost weight of each effective topological edge in the topology-grid fusion map based on the number of passable grids crossed and write it into the adjacency table to obtain the weighted topology-grid fusion map. Step S1.8: Overlay and fuse the weighted topology-grid fusion map with the raw tobacco logistics workshop grid map data in a unified coordinate system to obtain a global environment map of raw tobacco logistics operations.

[0010] Beneficial effects of steps S1.1 to S1.8: This series of steps completes the spatial data processing and topology-grid fusion map construction of the raw tobacco intelligent logistics workshop, realizing the accurate representation of workshop operation space and obstacle constraints, solving the problem of insufficient adaptability of single modeling methods, providing a unified and reliable spatial environment foundation for subsequent path planning, and ensuring the matching degree between the planned path and the actual traffic conditions in the workshop. The process includes the following steps: Step S1.1: Preprocessing and establishing a coordinate system for multi-source spatial data to provide standardized two-dimensional planar foundation data for subsequent spatial modeling; Step S1.2: Expanding the safety margin of obstacle boundaries to clarify the traffic constraint boundaries of the workshop space and avoid potential collision risks in path planning; Step S1.3: Building a topological framework for the global path in the workshop, clarifying the traffic adjacency relationships of key nodes, and adapting to the needs of large-scale path planning in the workshop; Step S1.4: Completing the gridded division and traffic status marking of the workshop space to achieve refined spatial representation of the work area; Step S1.5: Establishing a bidirectional index between topological nodes and grids to connect the two types of modeling data and provide data support for fusion modeling; Step S1.6: Verifying the spatial consistency of topological edges, eliminating invalid traffic connections, and ensuring the adaptability of the topological structure to the actual space; Step S1.7: Quantifying and assigning the traffic cost of topological edges to provide a quantifiable decision basis for subsequent path search; and Step S1.8: Integrating and merging topological and grid data to form a global environmental map adapted to the scheduling of raw tobacco logistics.

[0011] As a further improvement to this application, step S2 involves obtaining the start and end node data of the raw tobacco transportation task, and initializing the basic parameters of the ant colony algorithm and the goose colony optimization algorithm in conjunction with the global environment map of the raw tobacco logistics operation, to construct a hybrid initial solution space for path search, including: Step S2.1: Obtain the coordinates of the starting point and ending point of the raw tobacco transportation task to obtain the starting and ending point data of the raw tobacco transportation task. Step S2.2: Extract all feasible path sets of the start and end node data based on the global environment map of the original tobacco logistics operation using the breadth-first search algorithm, and calculate the total path length of each feasible path to obtain the feasible path set and path length set corresponding to the start and end nodes; Step S2.3: Perform Min-Max normalization on the path length set to obtain the normalized path length set for the raw tobacco transportation task; Step S2.4: Initialize the basic parameters of the ant colony algorithm, such as the number of ants, the maximum number of iterations, the initial value of pheromones, the pheromone volatility coefficient, the pheromone utilization coefficient, and the heuristic weights, and define the initial pheromone allocation intensity according to the normalized path length set to obtain the ant colony algorithm initialization parameter set. Step S2.5: Initialize the basic parameters of the goose flock optimization algorithm, such as population size, maximum number of iterations, problem dimension, stage selection probability, decay factor, and random coefficient, to obtain the initial parameter set of the goose flock optimization algorithm. Step S2.6: Based on the ant colony algorithm initialization parameter set and the goose flock optimization algorithm initialization parameter set, and using the start and end node data as constraints, randomly initialize the ant colony positions on the global environment map of the original tobacco logistics operation, calculate the initial path length of each ant, and generate an initial path set. Step S2.7: Calculate the objective function value for each path in the initial path set and sort them. Identify the initial path with the optimal objective function value as the current optimal solution and generate an initial path set with objective function value evaluation. Step S2.8: Integrate the ant colony algorithm initialization parameter set, the goose colony optimization algorithm initialization parameter set, and the initial path set to obtain a hybrid initial solution space for path search.

[0012] Beneficial effects of steps S2.1 to S2.8: This series of steps revolves around the core requirements of raw tobacco transportation tasks, completing the clarification of transportation task boundaries, initialization of dual algorithm parameters, and construction of the initial solution space for path search. It connects the global environment map with the subsequent path optimization iteration process, defines a reasonable initial range for algorithm search, provides a standardized and adapted initial foundation for subsequent path optimization, and helps improve the convergence efficiency and rationality of path planning iterations. The process involves the following steps: Step S2.1 clarifies the start and end node constraints of the raw tobacco transportation task, defining the core start and end boundary for subsequent path planning; Step S2.2 identifies the feasible path range between the start and end nodes, clarifying the feasible domain boundary for path planning; Step S2.3 standardizes the path length, providing a unified quantitative benchmark for initial pheromone allocation; Step S2.4 initializes the basic parameters and allocates the initial pheromone for the ant colony algorithm, establishing the basic rule framework for ant colony path search; Step S2.5 initializes the parameters for the goose colony optimization algorithm, providing a basic configuration for subsequent dynamic adjustment of algorithm parameters; Step S2.6 generates the initial path for the ant colony, constructing an initial path search sample set; Step S2.7 evaluates and sorts the objective function of the initial path, identifying the optimal benchmark in the initial solution and providing a reference anchor for iterative optimization; and Step S2.8 integrates the algorithm parameters and the initial path to form a hybrid initial solution space for path search adapted to this transportation task.

[0013] As a further improvement to this application, step S3 involves dynamically adjusting the pheromone utilization coefficient and heuristic weights of the ant colony algorithm using the single-leg balancing strategy of the goose colony optimization algorithm, and generating dynamically collaboratively updated path search rules based on the path search hybrid initial solution space parameters, including: Step S3.1: Based on the initial stage of the goose flock optimization algorithm, calculate the weight of the random stone, the stone falling time, the sound propagation time, and the average time within the mixed initial solution space of the path search to obtain the time parameter set for the current iteration; Step S3.2: Calculate the stone impact velocity of the current iteration time parameter set, as well as the relationship between the sound propagation distance and the distance between individual guard geese, to obtain the impact velocity set of the current iteration; Step S3.3: Normalize the impact velocity set of the current iteration and calculate the impact velocity adjustment factor as the normalized value of the maximum impact velocity. Step S3.4: Calculate the maximum sound propagation distance of individual geese in the current iteration, and the sound propagation distance adjustment factor is the value of the maximum sound propagation distance; Step S3.5: Based on the current iteration number and the maximum iteration number, calculate the adaptive iteration weight α decay coefficient and β decay coefficient using the exponential decay formula to obtain the parameter decay coefficient set; Step S3.6: Substitute the impact velocity adjustment factor, the sound propagation distance adjustment factor, and the parameter attenuation coefficient set into the dynamic parameter adjustment formula to calculate the dynamic adjustment pheromone utilization coefficient and the dynamic adjustment heuristic weight of the ant colony algorithm, and obtain the parameter set of the ant colony algorithm after dynamic collaborative parameter update. Step S3.7: Redefine the path transition probability rule of the ant colony algorithm according to the parameter set of the ant colony algorithm and the mixed initial solution space of the path search algorithm to obtain the path search rule after dynamic collaborative parameter update. Step S3.8: Integrate the ant colony algorithm parameter set with the path search rules to obtain the path search rules after dynamic collaborative parameter updates.

[0014] Beneficial effects of steps S3.1 to S3.8: This series of steps relies on the single-leg balancing strategy of the goose colony optimization algorithm to achieve dynamic and coordinated adjustment of the core operating parameters of the ant colony algorithm, replacing the traditional fixed parameter configuration mode. It adapts to the balance needs of exploration and development during algorithm iteration, alleviates the problem that the ant colony algorithm is prone to getting trapped in local optima, provides dynamically adaptable rule support for the path search process, and improves the adaptability of the path planning algorithm to different operating scenarios. The process includes the following steps: Step S3.1 calculates the time parameter set for the current iteration, providing a basis for dynamic parameter adjustment within the iteration cycle; Step S3.2 calculates the impact velocity and the propagation distance between individuals, providing core dynamic driving data for parameter adjustment; Step S3.3 normalizes the impact velocity to generate standardized adjustment factors suitable for parameter adjustment; Step S3.4 calculates the adjustment factors related to sound propagation distance, providing a quantitative benchmark for dynamic adjustment of heuristic weights; Step S3.5 calculates the parameter attenuation coefficient for iterative adaptation, adapting to the exploration and development switching needs of the algorithm throughout its entire iteration cycle; Step S3.6 dynamically calculates the core parameters of the ant colony algorithm, generating an algorithm parameter set suitable for the current iteration state; Step S3.7 reconstructs the path transition probability rules based on the dynamically updated parameter set, clarifying the core logic of path search during the iteration process; and Step S3.8 integrates the dynamic parameters and search rules to form dynamic path search rules suitable for the iteration process.

[0015] As a further improvement to this application, step S4 incorporates a random perturbation strategy into the path selection mechanism of the ant colony algorithm, and generates an enhanced path transition probability calculation model based on the path search rules, including: Step S4.1: In the exploration phase of the goose flock optimization algorithm, calculate the dynamic decay factor of the current iteration number based on the path search rule; Step S4.2: Calculate the wake-up intensity coefficient and minimum perturbation threshold in the random perturbation strategy based on the dynamic decay factor to obtain the random perturbation strategy parameter set; Step S4.3: Obtain the obstacle density distribution in the global environment map of the raw tobacco logistics operation, and identify high obstacle density areas and low obstacle density areas to obtain environmental obstacle density distribution data; Step S4.4: Dynamically adjust the wake-up intensity coefficient in the random perturbation strategy parameter set according to the environmental obstacle density distribution data in a proportional manner to obtain an obstacle adaptive perturbation parameter set; Step S4.5: Sample the random numbers from the standard normal distribution and multiply them by the wake-up intensity coefficient in the perturbation parameter set to obtain the random perturbation term; Step S4.6: Add the random perturbation term to the state transition probability formula of the ant colony algorithm to obtain the path transition probability calculation model incorporating the random perturbation strategy; Step S4.7: Based on the path transition probability calculation model and the ant colony algorithm parameter set, obtain the ant path selection result with embedded wake-up mechanism through ant colony path selection; Step S4.8: Integrate the ant path selection results, the ant colony algorithm parameter set, and the path transition probability calculation model to obtain the enhanced path transition probability calculation model.

[0016] Beneficial effects of steps S4.1 to S4.8: This series of steps incorporates a random perturbation strategy into the ant colony algorithm's path selection mechanism, optimizes the perturbation adaptation logic based on workshop environment characteristics, enriches the diversity of the path selection process, alleviates the problem of path homogenization in algorithm iteration, enhances the adaptability of path planning to complex workshop environments, and provides more reasonable probability calculation support for the path iteration search of ant colonies. The process includes the following steps: Step S4.1 calculates the iterative dynamic decay factor to provide an adaptation benchmark within the iteration cycle for adjusting the parameters of the random perturbation strategy; Step S4.2 calculates the core perturbation parameters to clarify the basic configuration boundary of the random perturbation strategy; Step S4.3 analyzes the obstacle density distribution in the workshop environment to provide a spatial basis for the environmental adaptive adjustment of perturbation parameters; Step S4.4 implements the environmental adaptive adjustment of perturbation parameters to improve the adaptability of the perturbation strategy to different complex areas; Step S4.5 generates random perturbation terms adapted to the current iteration to provide diverse support for the optimization of path transition probability; Step S4.6 integrates the calculation logic of random perturbation and path transition probability to construct a probability calculation framework with perturbation enhancement; Step S4.7 verifies the effect of the perturbation strategy in ant path selection to ensure the rationality of the path selection logic; and Step S4.8 integrates the enhanced probability calculation model to provide optimized core calculation basis for subsequent path iteration search.

[0017] As a further improvement to this application, step S5, dynamically compressing the path search space using the dimensionality scaling factor of the goose flock optimization algorithm, to perform path iterative search and global pheromone update of the ant colony based on the enhanced path transition probability calculation model, includes: Step S5.1: Perform path iterative search on the enhanced path transition probability calculation model using the ant population, and calculate the path length of each ant in each iteration to obtain the ant path length set for the current iteration. Step S5.2: Select the path corresponding to the ant with the smallest path length in the ant path length set as the current global optimal path, and update the global optimal path length and global optimal path position to obtain the global optimal path information for the current iteration. Step S5.3: Calculate the pheromone increment released by each ant on its own path according to the pheromone update rule of the ant colony algorithm, and sum up the pheromone increments of all ants to obtain the total pheromone increment set for the current iteration. Step S5.4: Apply the pheromone evaporation coefficients from the ant colony algorithm parameter set after dynamic collaborative parameter update to the pheromones of all ants in the previous iteration, and obtain the pheromone matrix after pheromone evaporation through pheromone evaporation update. Step S5.5: Add the total pheromone increment set of the current iteration to the pheromone matrix to perform a global pheromone update, and obtain the updated pheromone matrix; Step S5.6: Based on the updated pheromone matrix and the current iteration number, the path length calculation formula is dynamically adjusted using the dimension scaling factor and the maximum Euclidean distance to compress and compensate the high-dimensional path, thereby obtaining the path distance calculation formula after dimension scaling compensation. Step S5.7: Recalculate the distance values ​​of all paths in the current population using the path distance calculation formula to update the global optimal path information and obtain the path iteration search results after dimensional compression. Step S5.8: Determine whether the maximum number of iterations has been reached or the convergence criterion has been met. If at least one of them is met, terminate the iteration and obtain the path iteration search results after dimensional compression and the updated global optimal path information.

[0018] Beneficial effects of steps S5.1 to S5.8: This series of steps completes the path iterative search, pheromone global update, and dynamic compression of the search space for the ant colony, forming a closed-loop optimization process for algorithm iteration. It balances the global exploration and local development in the path search process, accelerates the algorithm convergence process, ensures the continuous optimization of path solutions during iteration, and provides core iterative support for the output of the final optimal scheduling path. The process involves the following steps: Step S5.1 calculates the ant path length set within a single iteration, providing basic path data for iterative optimization calculations; Step S5.2 updates the global optimal path information for the current iteration, providing a reference anchor point for subsequent iterations; Step S5.3 calculates the pheromone increment set for a single iteration, providing a quantitative basis for global pheromone updates; Step S5.4 updates historical pheromone evaporation, adapting to dynamically adjusted parameter rules and ensuring the rationality of pheromone iterations; Step S5.5 updates the pheromone matrix globally, achieving positive feedback guidance for the ant colony algorithm and optimizing the path search direction for subsequent iterations; Step S5.6 generates path distance calculation rules after dimensional scaling compensation, achieving dynamic compression of the path search space and adapting to the optimization needs of high-dimensional planning scenarios; Step S5.7 recalculates path distances and updates optimal path information, strengthening the local optimization capability of the iteration process; and Step S5.8 determines the iteration termination condition, closes the loop to control the algorithm iteration process, and outputs the final path iteration search results.

[0019] As a further improvement to this application, step S6, based on the results of the path iterative search and pheromone global update, obtains the globally optimal scheduling path for the raw tobacco transportation task, including: Step S6.1: Based on the results of path iterative search and pheromone global update, extract the global optimal path and global optimal path length in the final iteration process to obtain the optimal path candidate set for the raw tobacco transportation task. Step S6.2: Perform collision detection on each path in the optimal path candidate set to verify whether the path intersects with the obstacle expansion boundary in the global environment map of the original tobacco logistics operation, and obtain the collision detection verification result. Step S6.3: Based on the collision detection verification results, filter all paths that pass the collision detection. If there are multiple paths that pass the collision detection, select the path with the shortest path length as the final optimal path to obtain the global optimal scheduling path for the raw tobacco transportation task. Step S6.4: Remove the transition points in the globally optimal scheduling path to obtain the smoothed optimal scheduling path; Step S6.5: Calculate the total length, expected running time, and energy consumption assessment value of the smoothed optimal scheduling path to obtain the path performance assessment data of the raw tobacco transportation task; Step S6.6: Overlay the smoothed optimal scheduling path with the original tobacco logistics operation global environment map for verification to confirm that the smoothed optimal scheduling path has met all constraints and performance requirements. Step S6.7: After confirmation, the globally optimal scheduling path for the raw tobacco transportation task is obtained.

[0020] Beneficial effects of steps S6.1 to S6.8: This series of steps follows the results of iterative path search and global pheromone update, completing the selection of the optimal path, compliance verification, smoothing optimization, and final confirmation. This ensures that the output path adapts to the spatial constraints and transportation task requirements of the raw tobacco smart logistics workshop, improves the feasibility and rationality of the final scheduling path, and provides a scheduling solution that adapts to the actual operation scenario for raw tobacco logistics transportation. The process includes the following steps: Step S6.1: Extracting the optimal path-related data output during the iteration process and defining the core candidate range for path selection; Step S6.2: Completing collision detection verification of the path to identify potential conflicts between the path and obstacle boundaries, ensuring the safety of the path; Step S6.3: Completing the selection of compliant paths and the optimal path, locking in the core path that meets the optimal length requirement; Step S6.4: Optimizing the transition points of the path to improve path smoothness and adapt to the actual operating characteristics of the transport vehicle; Step S6.5: Calculating the core performance indicators of the path to provide a quantitative reference for path rationality assessment; Step S6.6: Completing the overlay verification of the path and the global environment map to confirm the adaptability of the path to various constraints; and Step S6.7: Completing the final path confirmation output to form the globally optimal scheduling path for the raw tobacco transportation task.

[0021] To achieve the above objectives, this application also provides the following technical solutions: A smart logistics scheduling device for raw tobacco, wherein the smart logistics scheduling device is applied to the smart logistics scheduling method described above, and the smart logistics scheduling device comprises: The raw tobacco logistics operation global environment map construction module is used to construct a raw tobacco logistics operation global environment map based on the operation space and obstacle distribution data of the raw tobacco smart logistics workshop through the fusion of topology modeling and grid modeling. The path search hybrid initial solution space construction module is used to obtain the start and end node data of the raw tobacco transportation task, and to initialize the basic parameters of the ant colony algorithm and the goose colony optimization algorithm in combination with the global environment map of the raw tobacco logistics operation, so as to construct the path search hybrid initial solution space. The path search rule generation module is used to dynamically adjust the pheromone utilization coefficient and heuristic weight of the ant colony algorithm through the single-leg balance strategy of the goose flock optimization algorithm, and generate path search rules with dynamically and collaboratively updated parameters based on the path search hybrid initial solution space. An enhanced path transition probability calculation model generation module is used to incorporate a random perturbation strategy into the path selection mechanism of the ant colony algorithm and generate an enhanced path transition probability calculation model based on the path search rules. The path iterative search and pheromone global update module is used to dynamically compress the path search space through the dimension scaling factor of the goose flock optimization algorithm, so as to perform path iterative search and pheromone global update of the ant population based on the enhanced path transition probability calculation model. The global optimal scheduling path acquisition module is used to obtain the global optimal scheduling path for the raw tobacco transportation task based on the results of the path iterative search and global pheromone update.

[0022] To achieve the above objectives, this application also provides the following technical solutions: An electronic device includes a processor and a memory coupled to the processor, the memory storing program instructions executable by the processor; when the processor executes the program instructions stored in the memory, it implements the intelligent logistics scheduling method for raw tobacco as described above.

[0023] To achieve the above objectives, this application also provides the following technical solutions: A computer-readable storage medium storing program instructions, which, when executed by a processor, enable the implementation of the intelligent logistics scheduling method for raw tobacco as described above. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating the steps of an embodiment of the intelligent logistics scheduling method for raw tobacco according to this application. Figure 2 This is a functional module diagram of one embodiment of a smart logistics scheduling device for raw tobacco according to this application; Figure 3 This is a schematic diagram of the structure of an embodiment of the electronic device of this application; Figure 4 This is a schematic diagram of the structure of one embodiment of the storage medium of this application. Detailed Implementation

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0026] The terms "first," "second," and "third" in this application are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first," "second," or "third" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified. All directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of this application are only used to explain the relative positional relationships and movements between components in a specific orientation (e.g., as shown in the figures). If the specific orientation changes, the directional indications also change accordingly. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0027] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0028] It should be noted that, due to the limited types and number of symbols or letters that can represent specific meanings, for embodiments with many formulas or codes, there may be situations where symbols or letters cannot meet the usage requirements. Therefore, the interpretation of formula symbols in the steps or sub-steps of the embodiments is only valid for the current step or sub-step.

[0029] If the same symbol has different interpretations in different steps or sub-steps, the interpretation in the current step or sub-step shall prevail; if the same symbol appears in different steps or sub-steps, but no interpretation is given in subsequent steps or sub-steps after its first appearance, the interpretation in the first step or sub-step shall be used.

[0030] like Figure 1 As shown, this embodiment provides an example of a smart logistics scheduling method for raw tobacco. In this embodiment, the smart logistics scheduling method is applied to a smart logistics workshop for raw tobacco.

[0031] Specifically, the intelligent logistics scheduling method includes the following steps: Step S1: Based on the distribution data of the working space and obstacles in the raw tobacco smart logistics workshop, a global environment map of raw tobacco logistics operations is constructed by integrating topology modeling and grid modeling.

[0032] Preferably, the core objective of this step is to construct a topology-grid dual-layer fusion global environment map that adapts to the operational characteristics of the tobacco smart logistics workshop, thereby solving the problems of insufficient accuracy of single topology modeling and low computational efficiency of single grid modeling, and providing a unified, accurate, and comprehensive spatial environment benchmark for subsequent path planning.

[0033] Furthermore, step S1 specifically includes the following steps: Step S1.1: Collect 3D point cloud data and 2D planar drawing data of the raw tobacco smart logistics workshop, perform voxel filtering on the 3D point cloud data and align it with the coordinate system, and orthophoto the processed data onto the 2D planar coordinate system to generate the basic 2D planar data of the workshop.

[0034] Preferably, this step is used to achieve standardized processing of multi-source spatial data and unification of spatial benchmarks.

[0035] For data acquisition, 3D point cloud data of the raw tobacco smart logistics workshop can be collected using vehicle-mounted LiDAR and indoor SLAM scanning equipment, while simultaneously collecting 2D CAD floor plan data of the workshop's architecture; the sampling point density of the 3D point cloud data is no less than 100 points / m². 2 It covers the entire workshop operation area with no blind spots.

[0036] For point cloud preprocessing, a voxel filtering algorithm can be used to denoise and downsample the original 3D point cloud. The voxel grid side length is set to 5cm to reduce data redundancy while preserving spatial contour features. The filtering formula is: P filtered =VoxelGrid(P raw ,v size =0.05m). In the formula, P raw P is the original 3D point cloud set. filtered This is the filtered point cloud set.

[0037] For coordinate system alignment, the Iterative Closest Point (ICP) algorithm is used to rigidly register the filtered 3D point cloud coordinate system with the engineering coordinate system of the workshop 2D drawings (the origin is defined as the southwest corner of the workshop wall, the X-axis is along the east-west direction of the workshop, and the Y-axis is along the north-south direction of the workshop). The registration error is controlled within ±2cm. The coordinate transformation formula is: In the formula, R is the rotation matrix, t is the translation vector, (x,y,z) are the original point cloud coordinates, and (x′,y′,z′) are the aligned engineering coordinate system coordinates.

[0038] Specifically, for orthophoto projection and data fusion, the Z-axis elevation information can be removed from the aligned 3D point cloud, while retaining the X and Y plane coordinates. A 2D plane point set can be generated through deduplication. Simultaneously, the CAD drawings are vectorized to extract the contour vectors of static obstacles such as walls, fixed equipment, and columns. These vector vectors are then fused with the point cloud projection data to finally generate the basic 2D plane data of the workshop.

[0039] Step S1.2: Extract obstacle contours from the two-dimensional planar basic data of the workshop, perform morphological dilation operation on the obstacle contours to expand the safety margin, and generate workshop space constraint data with obstacle dilation boundaries.

[0040] Preferably, this step is used to define the passable / inaccessible boundaries of the workshop and to allow for safe clearance by transport vehicles.

[0041] For obstacle contour extraction, the Canny edge detection algorithm can be used to extract the pixel-level edges of obstacles based on the two-dimensional planar data of the workshop, and then the closed vector contours of all obstacles can be extracted by the Suzuki85 contour tracking algorithm, covering all static and immovable obstacles such as fixed walls, production equipment, columns, fixed shelves, and fire-fighting facilities in the workshop.

[0042] For safety margin expansion, mathematical morphological expansion can be used to expand the boundary of the extracted obstacle closed contour, reserving a safe avoidance distance for the tractor vehicle; a circular structural element with a radius of 30cm is set, and the expansion operation formula is: A⊕B={z|(B^)z∩A≠∅} where A is the original obstacle contour region, B is the circular structural element, B^ is the image of B, z is the translation amount of the structural element, and ∅ is the empty set.

[0043] Preferably, the expanded obstacle area can be marked as an impassable area, and the remaining area can be marked as an initially passable area, ultimately generating workshop space constraint data with obstacle expansion boundaries.

[0044] Step S1.3: Extract the coordinates of key nodes from the workshop space constraint data, and establish an adjacency table based on the access relationships between nodes to generate the topology structure data of the original tobacco logistics workshop.

[0045] Preferably, this step is used to build a topological framework for the global paths in the workshop, enabling fast indexing and searching of a wide range of paths.

[0046] Specifically, the definition and extraction of topology nodes are as follows: topology nodes are defined as core points in the raw tobacco logistics operation, including path intersections, entrances and exits of raw tobacco storage areas, production feeding ports, workstation entrances and exits, tractor charging stations, elevator entrances, etc. Based on workshop space constraint data, the two-dimensional coordinates of all key nodes are extracted, a unique ID is assigned to each node, and node type attributes are recorded to form a topology node set V={v1,v2,...,v...} n}

[0047] In constructing topological edges and adjacency relationships, traversable direct paths between nodes within the workshop can be defined as topological edges, forming a topological edge set E={e1,e2,...,e...}. m Based on node coordinates, determine whether there is an unobstructed direct path between any two nodes, and establish an adjacency table for each node, recording the ID of adjacent nodes, the length of the corresponding edge, and the direction of travel.

[0048] In the process of generating the topology graph, a directed topology graph G=(V,E) is finally constructed, and a matching adjacency table is output simultaneously to form the topology graph structure data of the raw tobacco logistics workshop.

[0049] Step S1.4: Divide the workshop space constraint data into grids at a fixed resolution, determine the overlap relationship between each grid and obstacles, and mark the passage status to obtain the original tobacco logistics workshop grid map data.

[0050] Preferably, this step is used to achieve a refined spatial representation of the workshop operation area, providing a high-precision spatial basis for local path planning.

[0051] For grid division, based on a unified workshop engineering coordinate system, the entire workshop area is divided into continuous square grids, with a fixed grid resolution of 10cm × 10cm, balancing path planning accuracy and computational efficiency. The conversion relationship between grid row and column indices and world coordinates is established, as shown in the following formula: In the formula, col is the raster column number, row is the raster row number, (x,y) is the world coordinate, and x... min y min Here are the minimum coordinate values ​​in the workshop coordinate system, and resolution = 0.1m is the raster resolution. This is the floor function.

[0052] For the passage status marking, the occupancy grid method can be used to assign a passage status value to each grid. The judgment rule is: if more than 50% of the area of ​​a single grid overlaps with the expanded obstacle area, it is marked as an impassable grid and assigned a value of 1; otherwise, it is marked as a passable grid and assigned a value of 0.

[0053] For raster map generation, after completing the passage status marking of the entire area raster, a binary raster map data of the raw tobacco logistics workshop is generated, which includes raster index, coordinate mapping relationship, and passage status attributes.

[0054] Step S1.5: Map the node coordinates of the original tobacco logistics workshop topology map structure data to the corresponding grid positions of the original tobacco logistics workshop grid map data to establish a bidirectional index mapping relationship between topology nodes and grids, and obtain a topology-grid bidirectional index table.

[0055] Preferably, this step is used to establish a data link between the topology map and the raster map, enabling spatial information exchange between the two types of models.

[0056] For the construction of the forward mapping, all nodes in the topology graph can be traversed. Using the coordinate-raster conversion formula in step S1.4, the raster row number and column number corresponding to the coordinates of each node can be calculated to establish a forward mapping table of "node ID → raster row number, column number, and raster center point coordinates".

[0057] For the construction of the reverse mapping, all rasters in the raster map can be traversed. If the raster range contains topological nodes, the node ID corresponding to the raster index is recorded, and a reverse mapping table of "raster row number + column number → node ID list" is established.

[0058] Specifically, for the generation of bidirectional index tables, the forward mapping table and the reverse mapping table are integrated to form a topology-grid bidirectional index table, which enables fast bidirectional positioning and data retrieval of topology nodes and grid cells.

[0059] Step S1.6: Retrieve the grid accessibility status on the path connecting adjacent nodes of the topology graph structure data according to the topology-grid bidirectional index table, delete the topology edges containing inaccessible grids, and obtain the topology-grid fusion map after spatial consistency verification.

[0060] Preferably, this step is used to verify the actual drivability of topological edges, eliminate spatial conflicts between the topological map and the raster map, and ensure that the topological path matches the actual drivable area.

[0061] For topological edge path grid retrieval, each topological edge in the topological graph can be traversed to obtain the coordinates of the topological nodes corresponding to the start and end points of the edge. The Bresenham line algorithm is used to calculate the row and column indices of all grids traversed by the direct path from the start point to the end point.

[0062] Specifically, for the accessibility verification and invalid edge removal, the accessibility status of each grid cell traversed by the path is verified based on the grid map data. If any grid cell in the path is impassable, the topological edge is determined to be invalid and is simultaneously deleted from the edge set and adjacency table of the topological graph. If all traversable grid cells are accessible, the edge is retained and marked as a valid topological edge.

[0063] For the generation of the fused map, after verifying all topological edges, the edge set and adjacency table of the topological graph are updated, and finally the topological-grid fused map after spatial consistency verification is obtained.

[0064] Step S1.7: Calculate the passage cost weight of each effective topological edge in the topology-grid fusion map based on the number of passable grids crossed, and write it into the adjacency table to obtain the weighted topology-grid fusion map.

[0065] Preferably, this step is used to quantify the passage cost of the topological edges, providing a quantifiable decision basis for subsequent path search algorithms.

[0066] For the calculation of the topological edge travel cost, for each valid topological edge, the number of passable grids it passes through is obtained using the Bresenham algorithm. The actual Euclidean length of the edge is then calculated based on the grid resolution. The travel cost weight is determined based on this weight, and the calculation formula is: cost base =n×resolution. Where n is the number of passable grid cells traversed by the topological edge, and cost is... base This represents the basic travel cost of the topological edge.

[0067] Specifically, for updating the adjacency table, the calculated passage cost weight can be written into the adjacency table corresponding to the topological edge, and the attribute fields of the edge can be updated to convert the topological graph into a weighted directed graph.

[0068] For the generation of weighted fused maps, after assigning costs to all valid topological edges and updating the adjacency list, the weighted topology-grid fused map is finally obtained.

[0069] Step S1.8: Overlay and fuse the weighted topology-raster fusion map with the original tobacco logistics workshop raster map data in a unified coordinate system to obtain the global environment map of the original tobacco logistics operation.

[0070] Preferably, this step is used to complete the final integration of the two-layer map to form a unified environmental benchmark adapted to the raw tobacco logistics scheduling.

[0071] For spatial overlay and fusion, under a unified workshop engineering coordinate system, the node, weighted edge, and adjacency relationship data of the weighted topology-grid fusion map are spatially aligned and integrated with the original tobacco logistics workshop grid map data to form a two-layer fusion map of "global topology layer + local grid layer".

[0072] Among them, for map attribute integration, all spatial information such as topology-grid bidirectional index table, obstacle expansion boundary, passable area range, and node attributes can be integrated into the fused map to form a complete environmental data system.

[0073] Preferably, in the step of generating a global environment map of raw tobacco logistics operations, the global topology layer is used for rapid global path planning for the large-scale raw tobacco transportation task, and the local grid layer is used for fine-grained path adjustment and obstacle avoidance in densely populated areas, providing a complete and accurate spatial environment foundation for subsequent path planning algorithms.

[0074] Beneficial effects of steps S1.1 to S1.8: This series of steps completes the spatial data processing and topology-grid fusion map construction of the raw tobacco intelligent logistics workshop, realizing the accurate representation of workshop operation space and obstacle constraints, solving the problem of insufficient adaptability of single modeling methods, providing a unified and reliable spatial environment foundation for subsequent path planning, and ensuring the matching degree between the planned path and the actual traffic conditions in the workshop. The process includes the following steps: Step S1.1: Preprocessing and establishing a coordinate system for multi-source spatial data to provide standardized two-dimensional planar foundation data for subsequent spatial modeling; Step S1.2: Expanding the safety margin of obstacle boundaries to clarify the traffic constraint boundaries of the workshop space and avoid potential collision risks in path planning; Step S1.3: Building a topological framework for the global path in the workshop, clarifying the traffic adjacency relationships of key nodes, and adapting to the needs of large-scale path planning in the workshop; Step S1.4: Completing the gridded division and traffic status marking of the workshop space to achieve refined spatial representation of the work area; Step S1.5: Establishing a bidirectional index between topological nodes and grids to connect the two types of modeling data and provide data support for fusion modeling; Step S1.6: Verifying the spatial consistency of topological edges, eliminating invalid traffic connections, and ensuring the adaptability of the topological structure to the actual space; Step S1.7: Quantifying and assigning the traffic cost of topological edges to provide a quantifiable decision basis for subsequent path search; and Step S1.8: Integrating and merging topological and grid data to form a global environmental map adapted to the scheduling of raw tobacco logistics.

[0075] Step S2: Obtain the start and end node data of the raw tobacco transportation task, and initialize the basic parameters of the ant colony algorithm and the goose colony optimization algorithm in combination with the global environment map of the raw tobacco logistics operation to construct a hybrid initial solution space for path search.

[0076] Preferably, the core objective of this step is to anchor the core constraint boundary of the raw tobacco transportation task, complete the dual algorithm initialization configuration of Ant Colony Optimization (ACO) and Goose Group Optimization (GOA), construct a hybrid initial solution space for path search adapted to this transportation task, provide a standardized initial benchmark for the subsequent dynamic parameter optimization and path iterative search of the algorithm, eliminate the blindness of iterative search, and improve the convergence efficiency and initial rationality of the solution.

[0077] Furthermore, step S2 specifically includes the following steps: Step S2.1: Obtain the coordinates of the starting point and ending point of the raw tobacco transportation task to obtain the starting and ending point data of the raw tobacco transportation task.

[0078] Preferably, this step is used to clarify the core boundary constraints of path planning.

[0079] For task data traceability, the raw tobacco outbound storage location number and target feeding station number can be extracted from the raw tobacco transportation work order issued by the workshop production execution system (MES) as the starting and ending anchor points of the transportation task.

[0080] Specifically, for coordinate mapping extraction, based on the global environment map generated in step S1, the two-dimensional coordinates (x, y, y) of the starting and ending topology nodes in the workshop's unified engineering coordinate system are extracted through the pre-mapping relationship between workstation / storage location numbers and topology node IDs. start ,y start ), (x goal ,y goal ).

[0081] Preferably, it can verify whether the start and end coordinates fall within the passable grid range of the global environment map; if the coordinates fall within the impassable area, it automatically matches the nearest passable topology node as the corrected start and end node, and finally generates compliant start and end node data for the raw tobacco transportation task.

[0082] Step S2.2: Extract all feasible path sets based on the global environment map of the original tobacco logistics operation using the breadth-first search algorithm for the start and end node data, and calculate the total path length of each feasible path to obtain the feasible path set and path length set corresponding to the start and end nodes.

[0083] Preferably, this step is used to define the feasible domain of path planning, providing basic data for subsequent initial pheromone allocation.

[0084] For breadth-first search (BFS) execution, based on the topological graph structure of the weighted topology-grid fusion map generated in step S1, the starting node is used as the root node and the ending node as the target node, and the BFS algorithm is executed. During the search process, only valid topological edges that have been verified by spatial consistency are traversed, repeated visits to nodes are prohibited, and a maximum search depth threshold of 20 is set to avoid invalid long path traversals. The search terminates when the ending node is reached or all reachable nodes have been traversed.

[0085] Specifically, for the calculation of feasible paths and lengths, for all acyclic feasible paths obtained through the search, the corresponding travel cost weights are accumulated according to the adjacency order of the topological edges, the total path length of a single path is calculated, and the final output is a set of feasible paths containing all path node sequences, as well as a set of path lengths corresponding to the total length of each path.

[0086] Step S2.3: Perform Min-Max normalization on the path length set to obtain the normalized path length set for the raw tobacco transportation task.

[0087] This step is used to eliminate the influence of the dimension of path length and provide a unified quantitative benchmark for the differentiated allocation of initial pheromones.

[0088] For the Min-Max normalization method, all original length values ​​in the path length set are linearly normalized, mapping the values ​​to the [0,1] interval. The calculation formula is as follows: L norm (i)=(L(i)-L min ) / (L max -L min In the formula, L(i) is the original total length of the i-th feasible path, L... min L is the minimum value in the set of path lengths. max L is the maximum value in the set of path lengths. norm (i) represents the normalized length of the i-th path.

[0089] Step S2.4: Initialize the basic parameters of the ant colony algorithm, such as the number of ants, maximum number of iterations, initial pheromone value, pheromone evaporation coefficient, pheromone utilization coefficient, and heuristic weights. Define the initial pheromone allocation intensity based on the normalized path length set to obtain the ant colony algorithm initialization parameter set.

[0090] Preferably, this step is used to build the basic rule framework for ant colony path search, and to complete the initial assignment of core parameters and the differentiated allocation of initial pheromones.

[0091] Specifically, for the initialization of basic parameters, based on the topology of the original tobacco logistics workshop and the tractor scheduling scenario, the following fixed initial parameters are set: Ant count m: Set to 30, taking 10% to 15% of the total number of nodes in the workshop topology, balancing the diversity of path search and computational efficiency.

[0092] Maximum number of iterations (Max) iter Setting it to 100 balances the convergence effect of the algorithm with the computation time.

[0093] The initial pheromone baseline value τ0 is set to 1.0, which is the initial pheromone baseline for all topological edges.

[0094] Initial pheromone evaporation coefficient ρ0: set to 0.2 to provide an initial baseline for subsequent dynamic adjustments.

[0095] Initial pheromone utilization coefficient α0 (pheromone heuristic factor): set to 1.0 to provide an initial baseline for subsequent dynamic adjustments.

[0096] Initial heuristic weight β0 (expected heuristic factor): set to 5.0 to provide an initial baseline for subsequent dynamic adjustments.

[0097] The total amount of pheromone released in a single iteration, Q, is set to 100 to provide a fixed baseline for pheromone updates.

[0098] Specifically, for the initial pheromone differential allocation, based on the normalized path length set, the initial pheromone enhancement allocation is performed on the topological edges covered by feasible paths between the start and end nodes. The calculation formula is as follows: τ ij (0)=τ0+k·(1-L norm (i)) In the formula, τ ij (0) represents the initial pheromone concentration of the topological edge (i,j), k is the enhancement coefficient, set to 0.5, L norm (i) is the normalized length value of the shortest feasible path containing this topological edge. In this way, the edges corresponding to shorter initial feasible paths have higher initial pheromone concentrations, guiding the initial search direction of the ant colony.

[0099] Step S2.5: Initialize the basic parameters of the goose flock optimization algorithm, such as population size, maximum number of iterations, problem dimension, stage selection probability, decay factor, and random coefficient, to obtain the initial parameter set of the goose flock optimization algorithm.

[0100] Preferably, this step is used to complete the basic parameter configuration of the goose colony optimization algorithm, providing a configuration benchmark for the dynamic optimization of the core parameters of the ant colony algorithm. The specific implementation is as follows: Combine the iteration cycle and optimization goal of the ant colony algorithm to complete the parameter initialization assignment of the goose colony optimization algorithm.

[0101] The population size N is set to 20 to match the size of the ant colony algorithm, balancing the efficiency of parameter optimization with the computational cost; the maximum number of iterations Max...t The maximum number of iterations is set to 100, consistent with the ant colony algorithm, to synchronize the iteration cycles of the two algorithms; the problem dimension dim is set to 3, corresponding to the three core parameters of the ant colony algorithm to be optimized: pheromone utilization coefficient α, heuristic weight β, and pheromone evaporation coefficient ρ; the stage selection probability threshold m threshold Set to 0.5 for determining the switch between the development and exploration phases; the formula selects a threshold p. threshold The initial attenuation factor γ0 is set to 0.2 for selecting and determining the position update strategy during the development phase; the initial attenuation factor γ0 is set to 2.0 to provide an initial benchmark for dynamic attenuation during the exploration phase; the random coefficient c is set to 0.1 to control the step size of the position update; the stone weight range is set to [5g, 25g] to match the parameter range of the single-leg balance strategy; the sound speed constant Ss is set to 343.2m / s to provide a fixed constant for calculating the sound propagation distance. After completing the initialization assignment of all parameters, the initialization parameter set for the goose flock optimization algorithm is generated.

[0102] Step S2.6: Initialize the parameter set based on the ant colony algorithm and the goose flock optimization algorithm. Using the start and end node data as constraints, randomly initialize the ant colony positions on the global environment map of the original tobacco logistics operation, calculate the initial path length of each ant, and generate the initial path set.

[0103] Preferably, this step is used to generate initial path search samples for the ant colony, providing basic data for initial solution evaluation.

[0104] In this way, the initial position of the ants is uniformly set, and the initial position of all ants is anchored to the starting node of the transportation task, ensuring that all ants start the path search from the same starting point.

[0105] For the initial path search, the initial pheromone matrix and initial state transition probability formula based on the ant colony algorithm are used to initialize the parameter set, allowing each ant to perform the path search independently. During the search, ants only select walkable adjacent nodes in the topology graph, prohibiting repeated visits to already visited nodes to avoid generating loop paths; when an ant reaches the destination node, the ant's path search is terminated, and its path node sequence is recorded.

[0106] Preferably, after completing the initial path search for all ants, the Euclidean distance between adjacent nodes is accumulated for the node sequence of each path, the total length of a single path is calculated, and finally an initial path set containing the initial path node sequence of all ants and the corresponding path length is generated.

[0107] Step S2.7: Calculate the objective function value for each path in the initial path set and sort them. Identify the initial path with the optimal objective function value as the current optimal solution and generate an initial path set with objective function value evaluation.

[0108] Preferably, this step is used to complete the quantitative evaluation and ranking of the initial paths, and to clarify the initial optimal benchmark for iterative optimization.

[0109] The objective function is defined with minimizing the total path length as the core optimization objective. A steering penalty term is also introduced to adapt to the operating characteristics of the tractor unit. The objective function is defined as follows: f(L k )=ω1·L k +ω2·N turn,k In the formula, f(L) k Let L be the objective function value of the k-th ant path. k N is the total length of the path. turn,k The objective function represents the number of turns on the path (a turn is defined as a change in direction of more than 30° between three adjacent path segments). ω1 is the length weight, set to 0.9, and ω2 is the turning penalty weight, set to 0.1. The smaller the objective function value, the better the overall performance of the path.

[0110] For evaluation and ranking, for each path in the initial path set, the corresponding objective function value is calculated, all paths are sorted in ascending order of objective function value, the path with the smallest objective function value is extracted and marked as the current global optimal initial path, and its path node sequence, path length and objective function value are recorded.

[0111] Step S2.8: Integrate the ant colony algorithm initialization parameter set, the goose flock optimization algorithm initialization parameter set, and the initial path set to obtain a hybrid initial solution space for path search.

[0112] Preferably, this step is used to complete the structured integration of all initial data, forming a unified input benchmark for subsequent iterative optimization.

[0113] Specifically, the three core data sets—ant colony optimization algorithm initialization parameter set, goose colony optimization algorithm initialization parameter set, and initial path set with evaluation—are structurally integrated with task constraint data, global environment map space constraints, and iterative initial state data to form a hybrid initial solution space for path search containing four dimensions: (1) Task constraint dimensions: coordinates of the start and end nodes of the raw tobacco transportation task, range of feasible paths, and space access constraints in the workshop.

[0114] (2) Algorithm configuration dimension: dual algorithm initialization parameter set, initial pheromone matrix, and iteration threshold configuration.

[0115] (3) Initial solution dimension: initial path set with objective function evaluation, initial state of ant population, and current global optimal initial solution.

[0116] (4) Iteration baseline dimension: initial value of iteration number, initial baseline of dynamic parameters, and algorithm stage switching rules.

[0117] Beneficial effects of steps S2.1 to S2.8: This series of steps revolves around the core requirements of raw tobacco transportation tasks, completing the clarification of transportation task boundaries, initialization of dual algorithm parameters, and construction of the initial solution space for path search. It connects the global environment map with the subsequent path optimization iteration process, defines a reasonable initial range for algorithm search, provides a standardized and adapted initial foundation for subsequent path optimization, and helps improve the convergence efficiency and rationality of path planning iterations. The process involves the following steps: Step S2.1 clarifies the start and end node constraints of the raw tobacco transportation task, defining the core start and end boundary for subsequent path planning; Step S2.2 identifies the feasible path range between the start and end nodes, clarifying the feasible domain boundary for path planning; Step S2.3 standardizes the path length, providing a unified quantitative benchmark for initial pheromone allocation; Step S2.4 initializes the basic parameters and allocates the initial pheromone for the ant colony algorithm, establishing the basic rule framework for ant colony path search; Step S2.5 initializes the parameters for the goose colony optimization algorithm, providing a basic configuration for subsequent dynamic adjustment of algorithm parameters; Step S2.6 generates the initial path for the ant colony, constructing an initial path search sample set; Step S2.7 evaluates and sorts the objective function of the initial path, identifying the optimal benchmark in the initial solution and providing a reference anchor for iterative optimization; and Step S2.8 integrates the algorithm parameters and the initial path to form a hybrid initial solution space for path search adapted to this transportation task.

[0118] Step S3: The pheromone utilization coefficient and heuristic weight of the ant colony algorithm are dynamically adjusted by the single-leg balance strategy of the goose optimization algorithm, and the path search rules are dynamically and collaboratively updated based on the path search hybrid initial solution space.

[0119] Preferably, the core objective of this step is to leverage the Single-Leg Balancing Strategy (SLBS) of the Goose Colony Optimization Algorithm (GOA) to achieve iterative dynamic collaborative adjustment of the core operating parameters of the Ant Colony Algorithm (ACO), replacing the fixed parameter configuration mode of the traditional ACO. This precisely balances the global exploration and local development capabilities during the algorithm iteration process, fundamentally alleviating the problems of premature convergence and poor adaptability to dynamic scenarios in the traditional ACO, and providing core rule support for iterative adaptation in subsequent path search.

[0120] Furthermore, step S3 specifically includes the following steps: Step S3.1: Based on the initial stage of the goose flock optimization algorithm, calculate the weight of the random stone, the stone falling time, the sound propagation time, and the average time in the mixed initial solution space of the path search to obtain the time parameter set of the current iteration.

[0121] Preferably, this step is a pre-calculation step for the GOA single-leg balance strategy, generating core time base data for dynamic parameter adjustment.

[0122] Specifically, for the iteration triggering logic, in each iteration cycle of the algorithm, a random number m∈[0,1] following a uniform distribution is first generated. When m≥0.5, the GOA development stage is entered, and the calculation process of this step is started. The iteration cycle is completely synchronized with the ant colony algorithm iteration cycle, and a single iteration corresponds to one parameter update.

[0123] Specifically, for the generation of core random parameters, random parameters are generated independently for each goose flock search agent based on the preset range of the GOA initialization parameter set: (1) Weight S of the stone W Generates uniformly distributed random values ​​in the interval [5,25], with units of g.

[0124] (2) The time T for the stone to fall Oit Generate a random integer in the interval [1, dm], where dm=3 is the problem dimension of GOA, corresponding to the 3 core parameters of ACO to be optimized.

[0125] (3) Sound propagation time T Sit Generate random integers within the interval [1, dm].

[0126] Specifically, for the calculation of time feature values: based on the generated random parameters, the total time and average time of a single iteration are calculated using the following formulas: In the formula, T T T is the average total time of a single iteration. A The average time provides a time reference for subsequent calculations of impact velocity and position updates.

[0127] Step S3.2: Calculate the stone impact velocity of the current iteration time parameter set, as well as the relationship between the sound propagation distance and the distance between individual guard geese, to obtain the impact velocity set of the current iteration.

[0128] Preferably, this step calculates the core driving parameters of the single-leg balancing strategy based on the time parameter set, providing a quantitative basis for the subsequent dynamic adjustment of ACO parameters.

[0129] Specifically, for calculating the impact velocity of the stone, the velocity of the stone hitting the ground is calculated for each individual goose in the flock, based on the stone's weight and fall time. The calculation formula is as follows: In the formula, g = 9.81 m / s 2 F is the constant of gravitational acceleration. S The impact velocity of the stone reflects the degree of deviation between the current solution and the optimal region; a larger value indicates a higher degree of deviation.

[0130] Specifically, the calculation of sound propagation distance and inter-individual distance is based on the constant speed of sound and sound propagation time. The formula for calculating the sound propagation distance and the distance between the warning goose and the individual is as follows: In the formula, S S =343.2 m / s is the speed of sound constant under standard atmospheric pressure, D Tit D is the total distance the sound travels. Git This value, representing the equivalent distance between geese and individuals, reflects the spatial dispersion of the flock; a larger value indicates a more dispersed population distribution.

[0131] Step S3.3: Normalize the impact velocity set of the current iteration and calculate the impact velocity adjustment factor as the normalized value of the maximum impact velocity.

[0132] Preferably, this step standardizes the impact velocity data to generate a normalized adjustment factor adapted to the ACO pheromone heuristic factor adjustment.

[0133] For the impact velocity normalization process, the Min-Max normalization method is used to map the impact velocities of all individuals in the current iteration to the [0,1] interval. The principle is the same as the Min-Max normalization mentioned above, and will not be repeated here.

[0134] Specifically, for the calculation of the impact velocity adjustment factor, the maximum value of the normalized impact velocity within the current iteration is taken as the core adjustment factor θ, i.e., θ = max(F S_norm (i)), the value range of this factor is fixed at [0,1], and it is used for the dynamic adjustment of the subsequent pheromone utilization coefficient α.

[0135] Step S3.4: Calculate the maximum sound propagation distance of individual geese in the current iteration, and the sound propagation distance adjustment factor is the value of the maximum sound propagation distance.

[0136] Preferably, this step standardizes the individual distance data to generate a normalized adjustment factor that adapts to the adjustment of the ACO expected heuristic factor.

[0137] In the process of normalizing the propagation distance, the Min-Max normalization method is also used to map the guard goose-individual distance of all individuals in the current iteration to the interval [0,1].

[0138] Specifically, for the calculation of the sound propagation distance adjustment factor, the maximum value of the normalized spacing within the current iteration is taken as the core adjustment factor δ, i.e., δ = max(D G_norm (i)), the value range of this factor is fixed at [0,1], and it is used for the dynamic adjustment of the subsequent heuristic weight β.

[0139] Step S3.5: Based on the current iteration number and the maximum iteration number, calculate the adaptive iteration weight α decay coefficient and β decay coefficient using the exponential decay formula to obtain the parameter decay coefficient set.

[0140] Preferably, this step is based on the iterative process, calculating the decay coefficient of the adaptive iterative weight to achieve a smooth switch between exploration and development throughout the entire iterative cycle.

[0141] Specifically, the decay coefficient calculation logic uses the current iteration number and the maximum iteration number to calculate the decay coefficient γθ of the pheromone factor and the decay coefficient γδ of the heuristic factor using an exponential decay function. This enables adaptive adjustment of parameter weights in the early and later stages of the iteration. The calculation formula is as follows: In the formula, t is the current iteration number, and Max is the maximum iteration number. iter =100 is the preset maximum number of iterations, θ0=1.0 and δ0=1.0 are the initial baseline values ​​of the adjustment factors; after preliminary experimental calibration, γ is set... θ =2.0、γ δ =1.5, balancing the exploration and development needs throughout the entire iteration cycle.

[0142] It is worth noting that in the early stage of iteration, the value of t is small, θ(t)≈θ0, δ(t)≈0, which strengthens the dominant role of pheromone weight and accelerates the algorithm to converge to the potential optimal region; in the later stage of iteration, the value of t is close to the maximum number of iterations, θ(t)→0, δ(t)≈δ0, which strengthens the dominant role of heuristic information weight, enhances the global exploration ability, and avoids the algorithm from getting trapped in local optima.

[0143] Step S3.6: Substitute the impact velocity adjustment factor, sound propagation distance adjustment factor, and parameter attenuation coefficient set into the dynamic parameter adjustment formula to calculate the dynamic adjustment pheromone utilization coefficient and dynamic adjustment heuristic weight of the ant colony algorithm, and obtain the parameter set of the ant colony algorithm after dynamic collaborative parameter update.

[0144] Preferably, this step, based on the aforementioned adjustment factor and decay coefficient, completes the dynamic calculation of the ACO core parameters, achieving iterative parameter collaborative updates.

[0145] Specifically, for the dynamic update of the pheromone evaporation coefficient, the pheromone evaporation coefficient ρ(t) is dynamically adjusted based on the global optimal path fitness of the current iteration, and the calculation formula is as follows: In the formula, ρ0 = 0.2 is the initial volatility coefficient, and fit B (t) represents the objective function value of the optimal path in the current iteration, fit maxThis is the historical maximum objective function value since the iteration began. This formula can reduce the evaporation coefficient and strengthen the accumulation of pheromones on the optimal path when the path optimization effect is significant; and increase the evaporation coefficient to weaken the constraint of historical pheromones and enhance the exploration ability when the algorithm stagnates.

[0146] Among them, for the dynamic update of the pheromone utilization coefficient, the pheromone heuristic factor α(t) is dynamically adjusted by combining the initial baseline value, the impact velocity adjustment factor, and the attenuation coefficient. The calculation formula is as follows: α(t) = α0 + θ(t)·F S_max_norm In the formula, α0 = 1.0 is the initial pheromone utilization coefficient, and F S_max_norm α(t) represents the maximum impact velocity after normalization in the current iteration. When the impact velocity is large, α(t) increases synchronously, strengthening the positive feedback effect of pheromones and accelerating the convergence of the algorithm.

[0147] Among them, for the dynamic update of heuristic weights, the expected heuristic factor β(t) is dynamically adjusted by combining the initial benchmark value, the propagation distance adjustment factor, and the attenuation coefficient. The calculation formula is as follows: β(t) = β0 + δ(t)·D G_max_norm In the formula, β0=5.0 is the initial heuristic weight, and D G_max_norm This represents the maximum spacing value after normalization in the current iteration; when the individual distribution is dispersed, β(t) increases synchronously, strengthening the guiding role of heuristic information and enhancing the diversity of path search.

[0148] Preferably, the dynamically updated ρ(t), α(t), β(t), and the corresponding iteration parameters and adjustment factors can be integrated to generate a parameter set for the ant colony algorithm with dynamic collaborative parameter updates. The parameter set is updated once per iteration, synchronously covering the core computation rules of the ant colony algorithm.

[0149] Step S3.7: Redefine the path transition probability rule of the ant colony algorithm based on the parameter set of the ant colony algorithm and the mixed initial solution space of path search, and obtain the path search rule after dynamic collaborative parameter update.

[0150] Preferably, this step reconstructs the core path transition probability rules of the ant colony algorithm based on the dynamically updated parameter set of the ant colony algorithm, providing iterative adaptation computational logic for the path selection of the ant colony.

[0151] Specifically, for the reconstruction of the state transition probability formula, the traditional ACO state transition formula with fixed parameters is replaced with an iterative adaptation formula based on dynamic parameters α(t) and β(t). The reconstructed path transition probability formula is as follows: In the formula, allowed k Let τ be the set of neighboring nodes that the k-th ant can currently access. ij(t) represents the current pheromone concentration from node i to node j, η ij (t)=1 / d ij For heuristic functions, d ij Let be the Euclidean distance from node i to j.

[0152] Preferably, the core change in the reconstructed path search rule is that α(t) and β(t) in each iteration are real-time values ​​dynamically optimized by GOA, rather than fixed constants. This allows the bias in path selection to be adaptively adjusted according to the iteration process and the distribution of solutions, balancing global exploration and local development.

[0153] Step S3.8: Integrate the ant colony algorithm parameter set with the path search rules to obtain the path search rules after dynamic collaborative parameter updates.

[0154] Preferably, this step completes the structured integration of dynamic parameters and path search rules, forming a complete path search rule system that can be directly invoked in subsequent iterations.

[0155] Preferably, the dynamic ant colony algorithm parameter set generated in step S3.6 can be deeply integrated with the dynamic path transition probability rules generated in step S3.7, and supplemented with supporting content such as iterative synchronization logic, parameter update triggering conditions, node access constraints, and taboo table rules, to form a complete path search rule specific to each iteration.

[0156] Preferably, the range of values ​​for the dynamic parameters can be verified, limiting α(t)∈[0.5,3.0], β(t)∈[2.0,8.0], and ρ(t)∈[0.05,0.5], to avoid the parameters deviating excessively from the reasonable range, which would cause the algorithm's search logic to fail.

[0157] Beneficial effects of steps S3.1 to S3.8: This series of steps relies on the single-leg balancing strategy of the goose colony optimization algorithm to achieve dynamic and coordinated adjustment of the core operating parameters of the ant colony algorithm, replacing the traditional fixed parameter configuration mode. It adapts to the balance needs of exploration and development during algorithm iteration, alleviates the problem that the ant colony algorithm is prone to getting trapped in local optima, provides dynamically adaptable rule support for the path search process, and improves the adaptability of the path planning algorithm to different operating scenarios. The process includes the following steps: Step S3.1 calculates the time parameter set for the current iteration, providing a basis for dynamic parameter adjustment within the iteration cycle; Step S3.2 calculates the impact velocity and the propagation distance between individuals, providing core dynamic driving data for parameter adjustment; Step S3.3 normalizes the impact velocity to generate standardized adjustment factors suitable for parameter adjustment; Step S3.4 calculates the adjustment factors related to sound propagation distance, providing a quantitative benchmark for dynamic adjustment of heuristic weights; Step S3.5 calculates the parameter attenuation coefficient for iterative adaptation, adapting to the exploration and development switching needs of the algorithm throughout its entire iteration cycle; Step S3.6 dynamically calculates the core parameters of the ant colony algorithm, generating an algorithm parameter set suitable for the current iteration state; Step S3.7 reconstructs the path transition probability rules based on the dynamically updated parameter set, clarifying the core logic of path search during the iteration process; and Step S3.8 integrates the dynamic parameters and search rules to form dynamic path search rules suitable for the iteration process.

[0158] Step S4: Incorporate a random perturbation strategy into the path selection mechanism of the ant colony algorithm, and generate an enhanced path transition probability calculation model based on the path search rules.

[0159] Preferably, the core objective of this step is to incorporate an adaptive random perturbation strategy into the path selection mechanism of the Ant Colony Algorithm (ACO), and to enhance and optimize the state transition probability model of the traditional ACO by combining the iterative stage characteristics of the Goose Group Optimization (GOA) algorithm with the obstacle distribution characteristics of the workshop environment. This breaks the path convergence problem caused by pheromone homogenization during iteration, alleviates the premature convergence defect of the algorithm, enhances the global exploration capability and adaptability to complex environments of path search, and provides a more robust probability calculation core model for subsequent iterative path search of ant colonies.

[0160] Furthermore, step S4 specifically includes the following steps: Step S4.1: In the exploration phase of the goose flock optimization algorithm, calculate the dynamic decay factor of the current iteration number based on the path search rules.

[0161] Preferably, this step provides an attenuation benchmark for the random perturbation strategy that adapts to the iteration period, thereby enabling the perturbation intensity to be adaptively and smoothly adjusted as the algorithm iterates.

[0162] Specifically, for the iteration phase triggering logic, in each iteration cycle of the algorithm, a random number m∈[0,1] following a uniform distribution is first generated. When m<0.5, the GOA exploration phase is entered, and the calculation process of this step is started. The iteration cycle is completely synchronized with the iteration cycle of the ant colony algorithm and GOA. A single iteration corresponds to one decay factor update.

[0163] For the calculation of the dynamic decay factor, based on the current iteration number and the maximum iteration number, an exponential decay model is used to calculate the dynamic decay factor λ(t). This ensures that the perturbation intensity is greater in the early stage of iteration to encourage global exploration; the perturbation intensity gradually decreases in the later stage of iteration to ensure the convergence stability of the algorithm. The calculation formula is as follows: In the formula, t is the current iteration number, and Max is the maximum iteration number. iter =100 is the preset maximum number of iterations (consistent with the initialization parameters in step S2); λ0=1.0 is the initial baseline value of the decay factor; μ=1.2 is the decay rate coefficient, which is calibrated through pre-experimentation in the raw tobacco logistics scheduling scenario to balance the exploration and convergence requirements of the entire iteration cycle.

[0164] Step S4.2: Calculate the wake-up intensity coefficient and minimum perturbation threshold in the random perturbation strategy based on the dynamic decay factor to obtain the random perturbation strategy parameter set.

[0165] Preferably, this step is based on a dynamic attenuation factor to calculate the core control parameters of the random disturbance strategy and to clarify the amplitude boundary and minimum threshold of the disturbance.

[0166] Specifically, the wake-up intensity coefficient is used to control the overall amplitude of random perturbations. It is iteratively adapted based on a dynamic attenuation factor, and the calculation formula is: ω(t)=ω max ·λ(t). Where ω max =0.8 is the maximum baseline value of wake-up intensity, ensuring that the perturbation amplitude will not cover the core guiding role of the original pheromone and heuristic terms; ω(t) is the wake-up intensity coefficient of the current iteration, and its value range gradually decreases from 0.8 to close to 0 as the iteration progresses.

[0167] In calculating the minimum perturbation threshold, to prevent the perturbation strength from approaching 0 in the later stages of iteration, which would cause the algorithm to completely lose its ability to escape local optima, a minimum perturbation threshold ω is set as a safety net. min The calculation formula is: ω min =0.1·ω max That is, the minimum perturbation threshold is fixed at 0.08, and when the iteratively calculated ω(t) < ω min When forced to take ω(t) = ω min It retains the basic ability to explore disturbances.

[0168] Step S4.3: Obtain the obstacle density distribution in the global environment map of the original tobacco logistics operation, and identify high obstacle density areas and low obstacle density areas to obtain environmental obstacle density distribution data.

[0169] Preferably, this step is based on a global environment map to quantify the density of obstacles in different areas of the workshop, providing a spatial basis for the environmental adaptive adjustment of the disturbance strategy.

[0170] For grid-level density calculation, based on the grid map data of the original tobacco logistics workshop generated in step S1, the sliding window method is used to calculate the obstacle density of the entire area. The sliding window size is set to 30×30 grids, corresponding to the actual workshop size of 3m×3m, matching the minimum turning radius and safe passage range of the tractor. The window sliding step is one grid. The formula for calculating the obstacle density within a single window is: ρ obs (x,y)=N total / N block In the formula, ρ obs (x,y) represents the obstacle density corresponding to the center grid cell (x,y) of the window, N block N represents the number of impassable grid cells within the window. total The total number of grid cells in the window (fixed at 900), with a density range of [0,1].

[0171] For regional density grading, regional grading is performed based on the calculated full-grid density data, and the grading rules are as follows: (1) High obstacle density region: ρ obs ≥0.3 means that at least 30% of the grids within the window are impassable, which are mostly areas with limited space for path selection, such as densely populated equipment areas, workstations, and narrow passages. (2) Region with medium obstacle density: 0.1 < ρ obs <0.3; (3) Low obstacle density area: ρ obs ≤0.1 means that the proportion of non-passable grids within the window is no more than 10%, which are mostly areas with sufficient path selection space, such as main roads in workshops and open passage areas.

[0172] Specifically, for node-density mapping, based on the topology-grid bidirectional index table in step S1, the obstacle density value and density classification label of the location of each topology node are matched to establish a mapping relationship of "topology node ID → obstacle density".

[0173] Step S4.4: Based on the environmental obstacle density distribution data, dynamically adjust the wake-up intensity coefficient in the random perturbation strategy parameter set in proportion to obtain the obstacle adaptive perturbation parameter set.

[0174] Preferably, this step adaptively adjusts the wake-up intensity coefficient based on the obstacle density distribution in the workshop environment to achieve differentiated control that enhances disturbance in complex and confined areas and achieves stable convergence in open areas.

[0175] Specifically, for the density adaptive adjustment logic, a proportional adjustment rule is adopted. Regions with higher obstacle density correspondingly increase the wake-up intensity coefficient. This is because high-density regions have fewer feasible paths, and ant path selection is easily constrained by local pheromones, requiring greater perturbation intensity to expand the search range. Low-density regions have sufficient feasible paths, eliminating the need for excessive perturbation and avoiding compromise of algorithm convergence. The adjustment calculation formula is as follows: ω′(t,i)=ω(t)·(1+k·ρ obs (i)). In the formula, ω′(t,i) is the adaptive wake-up intensity coefficient corresponding to node i in the t-th iteration; ω(t) is the iterative baseline wake-up intensity coefficient generated in step S4.2; k=1.5 is the density gain coefficient, which is calibrated by the scene; ρ obs (i) represents the obstacle density value corresponding to node i.

[0176] Regarding the value range constraint, to avoid the adjusted wake-up strength coefficient being too large and causing the original path search rules to become invalid, the upper limit of the coefficient is set to ω. max =0.8, that is, when ω′(t,i)>0.8, it is forced to take 0.8; the lower limit is maintained at ω set in step S4.2. min =0.08, ensuring the reasonableness of the disturbance.

[0177] Step S4.5: Sample the random numbers from the standard normal distribution and multiply them by the wake-up intensity coefficient in the perturbation parameter set to obtain the random perturbation term.

[0178] Preferably, this step generates a random perturbation term that conforms to statistical characteristics based on an adaptive perturbation parameter set, providing core variables with enhanced diversity for the path transition probability model.

[0179] For sampling from a standard normally distributed random number, the Box-Muller transform algorithm is used to generate random numbers ε that follow a standard normally distributed N(0,1). This algorithm can quickly generate samples that conform to a normal distribution using uniformly distributed random numbers. The sampling formula is as follows: In the formula, U1 and U2 are independent random numbers uniformly distributed in the interval (0,1), and the generated ε1 and ε2 are standard normal random numbers that follow N(0,1).

[0180] For the calculation of the random perturbation term, a random perturbation term is independently generated for each of the ant's optional neighboring nodes at the current node, and the calculation formula is as follows: In the formula, In the t-th iteration, the k-th ant selects the random perturbation term corresponding to node j from node i; ω′(t,i) is the adaptive wake-up intensity coefficient corresponding to node i; ε is the standard normal random number obtained by sampling.

[0181] Specifically, regarding the constraint on the disturbance term, to avoid extreme random values ​​causing abnormal probability calculations, the random disturbance term is truncated and restricted. Values ​​outside the range are forcibly replaced with interval boundary values ​​to ensure that disturbances do not deviate excessively from a reasonable range.

[0182] Step S4.6: Add the random perturbation term to the state transition probability formula of the ant colony algorithm to obtain the path transition probability calculation model incorporating the random perturbation strategy.

[0183] Preferably, this step incorporates the random perturbation term into the core state transition probability formula of the ant colony algorithm, reconstructs the path selection calculation logic, and forms a probability calculation framework with perturbation enhancement.

[0184] Specifically, for the reconstruction of the state transition probability formula, a random perturbation term is incorporated into the path transition probability formula with dynamic parameter adaptation generated in step S3. The reconstructed path transition probability calculation formula is as follows: In the formula, each core parameter is completely consistent with the preceding steps: (1) allowed k It is the set of accessible neighboring nodes corresponding to the current node of the k-th ant (excluding visited nodes and impassable nodes).

[0185] (2) τ ij (t) represents the current pheromone concentration from node i to node j.

[0186] (3) η ij (t)=1 / d ij For heuristic functions, d ij Let be the Euclidean distance from node i to j.

[0187] (4) α(t) and β(t) are the pheromone utilization coefficients and heuristic weights that are dynamically updated in step S3.

[0188] (5) This refers to the random perturbation term generated in the preceding steps.

[0189] It is worth noting that the reconstructed formula incorporates a random perturbation term through multiplication, which not only retains the core guiding role of the original pheromone and heuristic terms, but also achieves differentiated adjustment of path selection probability through the perturbation term, breaking the path convergence problem caused by pheromone homogenization; at the same time, the perturbation term is adaptively adjusted with the iteration process and the density of environmental obstacles, taking into account both exploration capability and convergence stability.

[0190] Preferably, for the current node of a single ant, after calculating the transition probability of all possible nodes, the sum of all probability values ​​can be automatically verified to ensure that the sum is 1. If there is a deviation caused by floating-point error, it can be normalized and corrected proportionally to avoid the failure of the roulette wheel selection logic.

[0191] Step S4.7: Based on the path transition probability calculation model and the ant colony algorithm parameter set, the ant path selection result with embedded wake-up mechanism is obtained through ant colony path selection.

[0192] Preferably, this step, based on the reconstructed path transition probability calculation model, completes the single-step node selection and full path traversal of the ant population, verifying the actual execution effect of the perturbation strategy.

[0193] Specifically, the ant path search execution process involves executing a complete path search independently for each ant in the population. The specific process is as follows: (1) Initialization: Set the starting position of the ant as the starting node of the transportation task, initialize the tabu list, add the starting node to the tabu list, and prohibit repeated access.

[0194] (2) Single-step node selection: Based on the current node of the ant, obtain the set of accessible neighboring nodes allowedk, calculate the transition probability of each optional node through the probability model in step S4.6, and use the roulette wheel selection method to randomly select the next access node based on the probability distribution.

[0195] (3) State update: Add the selected next node to the ant's path sequence and taboo list, and update the current node to the newly selected node.

[0196] (4) Termination Judgment: If the current node of the ant is the destination node of the transportation task, terminate the path search of the ant; if allowed k If the set is empty and the destination has not been reached, the ant path search is considered to have failed, and the search is restarted.

[0197] Preferably, during the path selection process, when the ant is in a high obstacle density area or the path has not been optimized in multiple consecutive iterations, the random perturbation term will automatically increase the diversity of path selection, awaken the ant to explore feasible paths guided by non-optimal pheromones, avoid getting trapped in local optima, and record the changes in path selection under the action of this mechanism.

[0198] Step S4.8: Integrate the ant path selection results, the ant colony algorithm parameter set, and the path transition probability calculation model to obtain the enhanced path transition probability calculation model.

[0199] Preferably, this step completes the structured integration of all core elements, forming a complete enhanced probability calculation model that can be directly used for subsequent iterative searches.

[0200] Among these, the integration of multi-dimensional elements involves the deep integration of four core elements to form a complete model system: (1) Core computing layer: The path transition probability calculation formula and normalization verification rule reconstructed in step S4.6 with the integration of random perturbation strategy.

[0201] (2) Dynamic parameter layer: The ant colony algorithm parameter set after dynamic collaborative updating of parameters generated in step S3, including iteratively adapted α(t), β(t), and ρ(t).

[0202] (3) Disturbance adaptation layer: iterative dynamic decay rule, obstacle density adaptive adjustment rule, random disturbance term generation and constraint rule.

[0203] (4) Execution rule layer: Ant path selection process, taboo list rules, roulette wheel selection rules, path search termination rules.

[0204] Preferably, after integration, the model is iteratively adapted and verified to ensure that the model can automatically adjust the core parameters and perturbation intensity according to the number of iterations, environmental regions, and algorithm stages, without logical conflicts or parameter out-of-bounds issues, and can stably support the full-cycle path iterative search of ant colonies.

[0205] Beneficial effects of steps S4.1 to S4.8: This series of steps incorporates a random perturbation strategy into the ant colony algorithm's path selection mechanism, optimizes the perturbation adaptation logic based on workshop environment characteristics, enriches the diversity of the path selection process, alleviates the problem of path homogenization in algorithm iteration, enhances the adaptability of path planning to complex workshop environments, and provides more reasonable probability calculation support for the path iteration search of ant colonies. The process includes the following steps: Step S4.1 calculates the iterative dynamic decay factor to provide an adaptation benchmark within the iteration cycle for adjusting the parameters of the random perturbation strategy; Step S4.2 calculates the core perturbation parameters to clarify the basic configuration boundary of the random perturbation strategy; Step S4.3 analyzes the obstacle density distribution in the workshop environment to provide a spatial basis for the environmental adaptive adjustment of perturbation parameters; Step S4.4 implements the environmental adaptive adjustment of perturbation parameters to improve the adaptability of the perturbation strategy to different complex areas; Step S4.5 generates random perturbation terms adapted to the current iteration to provide diverse support for the optimization of path transition probability; Step S4.6 integrates the calculation logic of random perturbation and path transition probability to construct a probability calculation framework with perturbation enhancement; Step S4.7 verifies the effect of the perturbation strategy in ant path selection to ensure the rationality of the path selection logic; and Step S4.8 integrates the enhanced probability calculation model to provide optimized core calculation basis for subsequent path iteration search.

[0206] Step S5: The path search space is dynamically compressed by the dimensionality scaling factor of the goose flock optimization algorithm to perform path iterative search and global pheromone update of the ant population based on the enhanced path transition probability calculation model.

[0207] Preferably, the core objective of this step is to achieve iterative dynamic compression of the path search space by relying on the standard dimensional scaling strategy of the Goose Group Optimization (GOA) algorithm. Combined with the enhanced path transition probability calculation model generated in step S4, the path search, global pheromone update, and optimal solution iterative optimization of the ant population throughout the entire iterative cycle are completed, forming a closed-loop optimization process for algorithm iteration. This balances global exploration and local development capabilities, accelerates the algorithm's convergence speed, avoids invalid searches during the iteration process, and provides core iterative support for the final optimal path output.

[0208] Furthermore, step S5 specifically includes the following steps: Step S5.1: Perform path iterative search on the enhanced path transition probability calculation model using the ant population. Calculate the path length of each ant in each iteration to obtain the ant path length set for the current iteration.

[0209] Preferably, this step provides basic path data for iterative optimization, and the core is to complete the path search and length calculation of the entire ant population in a single iteration based on the enhanced probability model.

[0210] Specifically, for iterative input loading, when a single iteration starts, the core input elements of the current iteration are loaded synchronously, including the enhanced path transition probability calculation model generated in step S4, the parameter set of the current iteration dynamic ant colony algorithm generated in step S3, the global environment map of the original tobacco logistics operation constructed in step S1, and the transportation task start and end node data determined in step S2. The iteration cycle is completely synchronized with the iteration cycle of the ant colony algorithm and GOA.

[0211] In the process of performing a full-population path search, for the 30 ants initialized in step S2, each ant is independently assigned a random number seed, and the path traversal is performed in strict accordance with the node selection rules of the enhanced path transition probability calculation model. Starting from the task starting node, the ants successively determine the next node to visit by using the roulette wheel selection method, and the taboo table is updated simultaneously to prevent repeated visits to traversed nodes, until the task ending node is reached, thus completing the construction of a single complete path and recording the path node sequence of each ant.

[0212] For path length quantification, for each ant's generated complete path node sequence, the total length of a single path is calculated by summing the straight-line distances between adjacent nodes based on the Euclidean distance of the topological edges of the global environment map. The calculation formula is as follows: Where k is the ant number, v i For the i-th node on the path, d(v i ,v i+1 ) represents the Euclidean distance between adjacent nodes, and n represents the total number of nodes in the path.

[0213] Step S5.2: Select the path corresponding to the ant with the smallest path length in the ant path length set as the current global optimal path, and update the global optimal path length and global optimal path position to obtain the global optimal path information for the current iteration.

[0214] Preferably, this step is based on the path search results of a single iteration to complete the iterative update of the global optimal solution, providing a reference anchor point for subsequent optimization.

[0215] Specifically, for the selection of the optimal path in the current iteration, based on the ant path length set generated in step S5.1, the path length and corresponding node sequence of all ants are traversed, the ant individual with the smallest path length in the current iteration is selected, its complete path node sequence and path length are extracted, and it is marked as the optimal path in the current iteration.

[0216] Specifically, for the iterative update of the global optimal path: the current iterative optimal path is compared with the global optimal path of the previous iteration using the objective function defined in step S2.7. The comparison rule is as follows: if the objective function value of the current iterative optimal path is less than the objective function value of the historical global optimal path, then the global optimal path is updated to the current iterative optimal path, and the length of the global optimal path, the path node sequence, and the objective function value are updated simultaneously; if the current iterative optimal path has no performance advantage, then the original global optimal path is retained without modification.

[0217] Specifically, for the optimal path spatial anchoring, based on the topology-grid bidirectional index table generated in step S1, all nodes of the updated global optimal path are mapped to the corresponding positions on the grid map, the complete spatial coordinate sequence of the path is recorded, and marked as the global optimal path position.

[0218] Step S5.3: Calculate the pheromone increment released by each ant on its respective path according to the pheromone update rule of the ant colony algorithm, and sum the pheromone increments of all ants to obtain the total pheromone increment set for the current iteration.

[0219] Preferably, this step is based on the standard ant cycle model of the ant colony algorithm to calculate the pheromone increment in a single iteration, providing a quantitative basis for the global update of pheromones.

[0220] Specifically, for calculating the pheromone increment of a single ant, for all ants that have completed a full path traversal in the current iteration, the ant-week model is used to calculate the pheromone increment released by a single ant on the corresponding path topological edge. The calculation formula is as follows: In the formula Let L be the pheromone increment released by the k-th ant on the topological edge (i,j) in the t-th iteration, Q=100 be the total pheromone release in a single iteration initialized in step S2, and L be the pheromone increment. k This represents the total path length of the k-th ant; if the ant does not pass through this topological edge, the corresponding increment is assigned a value of 0.

[0221] Specifically, for the calculation of the total increment of topological edges, for each valid topological edge in the global topological graph, the increment of pheromone released by all ants on that edge in the current iteration is accumulated, and the calculation formula is as follows: In the formula, Δτ ij (t) represents the total pheromone increment of the topological edge (i,j) in the current iteration.

[0222] Step S5.4: Apply the pheromone evaporation coefficients from the ant colony algorithm parameter set after dynamic collaborative parameter updates to the pheromones of all ants in the previous iteration, and obtain the pheromone matrix after pheromone evaporation through pheromone evaporation update.

[0223] Preferably, this step is based on the dynamically updated volatility coefficient to complete the historical pheromone volatility update, balancing the positive feedback strength and exploration capability of the algorithm.

[0224] Among them, for the core parameter loading, the pheromone evaporation coefficient ρ(t) generated in the current iteration dynamic ant colony algorithm parameter set generated in step S3, and the global pheromone matrix τ(t-1) generated after the (t-1)th iteration; at the beginning of the iteration (t=1), τ(0) is the initial pheromone matrix generated in step S2.4.

[0225] Specifically, for the pheromone evaporation update calculation, the pheromone concentration of each topological edge in the global pheromone matrix is ​​updated by performing an evaporation update operation to simulate the natural decay process of pheromones. The calculation formula is as follows: ,in Let be the pheromone concentration of the topological edge (i,j) after evaporation and updating.

[0226] Specifically, regarding the lower limit constraint on concentration, a lower limit τ for pheromone concentration is set to prevent excessive pheromone volatilization from causing the guiding effect to fail. min =0.01, if the concentration after evaporation and renewal is lower than this lower limit, then it is forcibly assigned the value τ. min This ensures the algorithm's basic search guidance capabilities.

[0227] Step S5.5: Add the total pheromone increment set of the current iteration to the pheromone matrix to perform a global pheromone update, and obtain the updated pheromone matrix.

[0228] Preferably, this step completes the global update of pheromones, realizing positive feedback guidance for the ant colony algorithm and providing a core basis for path selection in the next iteration.

[0229] Specifically, for core input loading, the pheromone matrix τ generated in step S5.4 after pheromone evaporation is loaded. ∗ (t), and the total pheromone increment set Δτ generated in step S5.3 for the current iteration. ij (t).

[0230] Specifically, for the global pheromone update calculation, for each valid topological edge in the global topology graph, the total pheromone increment of the current iteration is added to the pheromone concentration after evaporation to complete the global pheromone update. The calculation formula is as follows: , where τ ij (t) represents the final pheromone concentration of the topological edge (i,j) after global update.

[0231] Specifically, regarding the upper limit constraint on pheromone concentration, an upper limit τ for pheromone concentration is set to prevent premature convergence of the algorithm due to excessively high pheromone concentration. max =10.0, if the updated concentration is higher than this upper limit, then force the value to be τ.max The ability to explore and develop balancing algorithms.

[0232] Step S5.6: Based on the updated pheromone matrix and the current iteration number, the path length calculation formula is dynamically adjusted using the dimension scaling factor and the maximum Euclidean distance to compress and compensate the high-dimensional path, thus obtaining the path distance calculation formula after dimension scaling compensation.

[0233] Preferably, this step is based on the GOA standard dimensional scaling strategy to achieve dynamic compression and adaptation of the path search space and correct the distance calculation deviation of high-dimensional paths.

[0234] Specifically, for core input loading, the updated pheromone matrix generated in loading step S5.5, the current iteration number t, and the maximum iteration number Max are included. iter =100, the GOA problem dimension dim initialized in step S2 is 3, and the current iteration dynamic decay factor λ(t) generated in step S4.1 is λ(t).

[0235] For the calculation of the dimension scaling factor, the standard dimension scaling strategy based on the goose flocking optimization algorithm is used, combined with the iterative process to calculate the dimension scaling factor σ(t), and the calculation formula is as follows: , where σ max =1.0 is the initial maximum value of the dimension scaling factor; this formula ensures that the search space is kept intact in the early stage of iteration to encourage global exploration, and gradually compresses the search space in the later stage of iteration to focus on the local optimum region.

[0236] Specifically, for the calculation of the maximum Euclidean distance, based on the updated pheromone matrix and the current globally optimal path node sequence, the maximum Euclidean distance between adjacent nodes in the path is calculated, denoted as D. max , serving as a dimensional reference for path distance calculation.

[0237] Specifically, for the dimensionality compensation of the path distance formula, a dimensionality scaling factor and the maximum Euclidean distance are incorporated into the original path length calculation formula to compress and compensate for high-dimensional long paths, resulting in the path distance calculation formula after dimensionality scaling compensation: , where Lk′ is the compensated path distance value and Lk is the original total path length; this formula can enhance the distance advantage of short paths while avoiding excessive suppression of long paths.

[0238] Step S5.7: Recalculate the distance values ​​of all paths in the current population using the path distance calculation formula to update the global optimal path information and obtain the path iteration search results after dimensional compression.

[0239] Preferably, this step completes the path performance re-evaluation based on the compensated distance formula, realizes dynamic compression of the search space, and completes the closed-loop optimization of a single iteration.

[0240] Among them, the core input loading includes the path distance calculation formula after dimensional scaling compensation generated in step S5.6, the current iteration full ant path node sequence and original path length generated in step S5.1, and the current iteration global optimal path information generated in step S5.2.

[0241] Specifically, for the recalculation of path distances across the entire population, the compensated path distance calculation formula is used to recalculate the compensated path distance value Lk′ for each path of all ants in the current iteration, forming the compensated path distance set for the current iteration.

[0242] Specifically, for updating the global optimal path information, based on the compensated path distance set, the path with the smallest compensated distance value in the current iteration is selected and compared with the current global optimal path in terms of objective function performance. If the new path has better performance, the node sequence, path length, compensated distance value, and spatial location information of the global optimal path are updated; if there is no better path, the original global optimal path information is retained.

[0243] Specifically, for the search space compression constraint, based on the dimension scaling factor σ(t), the range of optional ant nodes in the next iteration is constrained, retaining only those with Euclidean distances to the current globally optimal path nodes within σ(t)·D. max The topological nodes within the range are used as the set of optional nodes for the next iteration, and invalid search areas are eliminated.

[0244] Step S5.8: Determine whether the maximum number of iterations has been reached or the convergence criterion has been met. If at least one of them is met, terminate the iteration and obtain the path iteration search results after dimensional compression and the updated global optimal path information.

[0245] Preferably, this step completes the dual judgment of the iteration termination condition, the closed-loop control algorithm iteration process, and outputs the final iterative optimization result.

[0246] Specifically, the iteration termination condition involves a dual-judgment process. After each iteration, a termination condition check is performed, and the algorithm terminates if either condition is met. The specific rules are as follows: First, the maximum number of iterations is checked; if the current iteration number t ≥ 100, the termination condition is met. Second, the convergence criterion is checked; if the objective function value of the globally optimal path changes by less than 10 in 15 consecutive iterations, the termination condition is met. -4 That is, |f(L) t )-f(L t-15 )|<10 -4 The algorithm is considered convergent when the termination condition is met.

[0247] Specifically, for the iterative loop logic, if any termination condition is not met, return to step S3, start the parameter dynamic update process for the next round of iteration, and continue to execute the iterative loop; if the termination condition is met, terminate the algorithm iterative loop immediately.

[0248] Preferably, after the iteration terminates, the final dimension-compressed path iteration search result is output, as well as the final global optimal path information updated throughout the iteration process, including the node sequence, path length, objective function value, and spatial coordinate sequence of the final global optimal path.

[0249] Beneficial effects of steps S5.1 to S5.8: This series of steps completes the path iterative search, pheromone global update, and dynamic compression of the search space for the ant colony, forming a closed-loop optimization process for algorithm iteration. It balances the global exploration and local development in the path search process, accelerates the algorithm convergence process, ensures the continuous optimization of path solutions during iteration, and provides core iterative support for the output of the final optimal scheduling path. The process involves the following steps: Step S5.1 calculates the ant path length set within a single iteration, providing basic path data for iterative optimization calculations; Step S5.2 updates the global optimal path information for the current iteration, providing a reference anchor point for subsequent iterations; Step S5.3 calculates the pheromone increment set for a single iteration, providing a quantitative basis for global pheromone updates; Step S5.4 updates historical pheromone evaporation, adapting to dynamically adjusted parameter rules and ensuring the rationality of pheromone iterations; Step S5.5 updates the pheromone matrix globally, achieving positive feedback guidance for the ant colony algorithm and optimizing the path search direction for subsequent iterations; Step S5.6 generates path distance calculation rules after dimensional scaling compensation, achieving dynamic compression of the path search space and adapting to the optimization needs of high-dimensional planning scenarios; Step S5.7 recalculates path distances and updates optimal path information, strengthening the local optimization capability of the iteration process; and Step S5.8 determines the iteration termination condition, closes the loop to control the algorithm iteration process, and outputs the final path iteration search results.

[0250] Step S6: Based on the results of path iterative search and global pheromone update, the globally optimal scheduling path for the raw tobacco transportation task is obtained.

[0251] Preferably, the core purpose of this step is to take the final results of the path iterative search and pheromone global update output in step S5, complete the construction of the candidate set of the optimal path, compliance verification, smoothing optimization, performance quantification and implementation verification, and output the globally optimal scheduling path that is adapted to the spatial constraints, transportation task requirements and tractor operation characteristics of the tobacco smart logistics workshop. This solves the problem of the actual executability of the algorithm iterative output path and provides a standardized path solution that can be directly issued and executed for the workshop transportation vehicle scheduling system.

[0252] Furthermore, step S6 specifically includes the following steps: Step S6.1: Based on the results of path iterative search and pheromone global update, extract the global optimal path and global optimal path length in the final iteration process to obtain the optimal path candidate set for the raw tobacco transportation task.

[0253] Preferably, this step is used to define the core scope of path screening, extract high-potential paths from the optimization results of the entire iteration cycle, and construct a standardized candidate set.

[0254] Among them, for core input loading, the final path iteration search result output in step S5, the global optimal path information for the entire iteration cycle, the final global pheromone matrix updated after the entire iteration process, and the raw tobacco transportation task start and end node constraint data determined in step S2 are loaded synchronously.

[0255] Among them, for the extraction of high-potential paths, three types of core paths were extracted and grouped. The first type is the global optimal path that is finally locked after iterative convergence. The second type is the top 10 paths with objective function values ​​in each iteration of the whole iteration cycle. The third type is the acyclic feasible path composed of the top 5 topological edges with the highest global pheromone concentration.

[0256] Preferably, all extracted paths are deduplicated by removing duplicate paths with identical node sequences, and each retained path is matched with corresponding attribute fields such as path node sequence, spatial coordinate sequence, total path length, objective function value, and number of turns.

[0257] Step S6.2: Perform collision detection on each path in the optimal path candidate set to verify whether the path intersects with the obstacle expansion boundary in the original tobacco logistics operation global environment map, and obtain the collision detection verification result.

[0258] Preferably, this step is used to verify the traffic safety of candidate paths and eliminate invalid paths that pose a risk of collision with obstacles in the workshop.

[0259] Among them, for core input loading, the optimal path candidate set generated in step S6.1, the obstacle expansion boundary vector data in the global environment map of the raw tobacco logistics operation constructed in step S1, and the vehicle body outline dimension parameters of the workshop tractor are loaded synchronously.

[0260] In the path segmentation process, for each path in the candidate set, the continuous sequence of nodes is converted into a set of path segments. The spatial coordinates of every two adjacent nodes constitute an independent straight path segment, which completely covers the entire path.

[0261] For collision detection, the Separated Axis Theorem (SAT) two-dimensional collision detection algorithm can be used to perform intersection checks on each path segment and the closed polygons of all obstacle expansion boundaries. The check rule is: if the path segment intersects with the obstacle expansion boundary polygon, or if the path segment enters the interior of the polygon, then the path is determined to have a collision risk.

[0262] Specifically, for result collection and labeling, for each path in the candidate set, the collision detection result is output and labeled as "pass" or "fail". The collision position and corresponding obstacle information of the failed path are recorded simultaneously, and finally a complete collision detection verification result is generated.

[0263] Step S6.3: Based on the collision detection verification results, filter all paths that pass the collision detection. If there are multiple paths that pass the collision detection, select the path with the shortest path length as the final optimal path to obtain the globally optimal scheduling path for the raw tobacco transportation task.

[0264] Preferably, this step is used to complete the screening of compliant paths and the locking of the optimal path, and output the initial optimal path that satisfies the spatial constraints.

[0265] Specifically, for core input loading, the collision detection verification results generated in step S6.2 and the optimal path candidate set generated in step S6.1 are loaded synchronously.

[0266] For the selection of compliant paths, based on the collision detection verification results, all paths marked as "passed" are extracted to form a subset of compliant feasible paths; if the subset of compliant feasible paths is empty, return to step S5, relax the iteration convergence threshold, and re-execute the path iteration search and optimization.

[0267] Specifically, for the optimal path sorting and selection, all paths within the compliant feasible path subset are sorted in ascending order by total path length. If there are paths with the same length, they are sorted in ascending order by the number of path turns. The first path after sorting is extracted and marked as the initial globally optimal scheduling path.

[0268] Step S6.4: Remove the transition points in the globally optimal scheduling path to obtain the smoothed optimal scheduling path.

[0269] Preferably, this step is used to remove redundant turning points in the path, optimize the smoothness of the path, and adapt to the actual operating characteristics of the tractor.

[0270] Among them, for core input loading, the node sequence of the globally optimal scheduling path generated in step S6.3 and the passable area data of the global environment map generated in step S1 are loaded synchronously.

[0271] Among them, the definition of the transition point determination rule is that an intermediate node that meets the following two conditions is defined as a transition point: first, the turning angle of the path formed by three adjacent nodes is less than 15°; second, the vertical distance from the intermediate node to the line connecting the two preceding and following nodes is less than 20cm. Such nodes have no substantial impact on the path direction and are considered removable redundant nodes.

[0272] For path smoothing and simplification, the Douglas-Peucker algorithm is used to simplify the path node sequence. The distance threshold is set to 20cm, which matches the transition turning point determination rule. Under the premise of preserving the core path direction and collision-free constraints, redundant transition turning points are automatically removed.

[0273] Preferably, the collision detection process in step S6.2 can be re-executed for the simplified path node sequence to ensure that the smoothed path is free of collision risk throughout; if a collision occurs after simplification, the process can be reverted to the previous version of the node sequence, and only the transition points that are not affected by collisions can be removed.

[0274] Step S6.5: Calculate the total length, expected running time, and energy consumption assessment value of the smoothed optimal scheduling path to obtain the path performance assessment data of the raw tobacco transportation task.

[0275] Preferably, this step is used to perform multi-dimensional quantitative evaluation of the smoothed optimal path, providing data support for verifying the rationality of the path.

[0276] Among them, for the core input loading, the smoothed optimal scheduling path, path basic attribute data, and rated operating parameters of the tractor in the raw tobacco logistics workshop generated in step S6.4 are loaded synchronously. The rated parameters of the tractor include: safe driving speed in the workshop 1.5m / s, turning speed 0.5m / s, rated time for a single turn 2s, energy consumption per unit driving distance 0.02kWh / m, and energy consumption per turn 0.005kWh / turn.

[0277] Specifically, for the calculation of core performance indicators, the quantitative calculation of three core indicators is completed. The first is the total path length, which is obtained by accumulating and smoothing the Euclidean distances between adjacent nodes of the path to obtain the accurate total length L. total The unit is meters (m); the second is the expected running time, calculated using the formula T. total =L total / 1.5+N turn ×2, where N turn The first is the number of turns on the smoothed path, measured in seconds; the second is the energy consumption assessment value, calculated using the formula E. total =L total ×0.02+N turn ×0.005, in kWh.

[0278] Step S6.6: Overlay the smoothed optimal scheduling path with the original global environment map of tobacco logistics operations to verify that the smoothed optimal scheduling path has met all constraints and performance requirements.

[0279] Preferably, this step is used to complete the full-dimensional compliance verification of the final path, ensuring that the path meets all preset constraints and performance requirements.

[0280] Among them, the core input loading synchronously loads the smoothed optimal scheduling path generated in step S6.4, the path performance evaluation data generated in step S6.5, the global environment map of the raw tobacco logistics operation constructed in step S1, and the transportation task constraints determined in step S2.

[0281] For multi-dimensional overlay verification, full verification of four categories of constraints is performed: First, spatial constraint verification, which overlays the spatial coordinate sequence of the path onto the global environment map to verify that the entire path is within the passable grid range and that the minimum distance to all obstacles is no less than a preset safety margin of 30cm, with no risk of collision; Second, task constraint verification, which verifies that the start and end points of the path completely match the start and end nodes of the transportation task, with no positional offset; Third, topological constraint verification, which verifies that all adjacent nodes of the path correspond to valid topological edges in the global environment map, with no invalid cross-node passage segments; Fourth, performance requirement verification, which verifies that the total path length and expected running time meet the timeliness requirements of the workshop transportation work order, with no cases exceeding the rated threshold.

[0282] Preferably, if all verification items pass, the overall verification is considered successful; if any verification item fails, the corresponding preceding step is returned for correction. For example, if the spatial constraint fails, the path is re-optimized in step S6.4, and if the performance fails, the path is re-iterated in step S5.

[0283] Step S6.7: After confirmation, the globally optimal scheduling path for the raw tobacco transportation task is obtained.

[0284] Preferably, the final confirmed globally optimal scheduling path is standardized and encapsulated, including a unique transportation task ID, start and end node information, complete path node sequence, spatial coordinate sequence, total path length, number of turns, expected running time, energy consumption assessment value, corresponding global environment map version number, and verification pass flag, among other attribute fields. This ultimately generates the globally optimal scheduling path for the raw tobacco transportation task, which can be directly issued to the transportation vehicle scheduling system of the raw tobacco smart logistics workshop to drive the tractor to execute the corresponding raw tobacco transportation task.

[0285] Beneficial effects of steps S6.1 to S6.8: This series of steps follows the results of iterative path search and global pheromone update, completing the selection of the optimal path, compliance verification, smoothing optimization, and final confirmation. This ensures that the output path adapts to the spatial constraints and transportation task requirements of the raw tobacco smart logistics workshop, improves the feasibility and rationality of the final scheduling path, and provides a scheduling solution that adapts to the actual operation scenario for raw tobacco logistics transportation. The process includes the following steps: Step S6.1: Extracting the optimal path-related data output during the iteration process and defining the core candidate range for path selection; Step S6.2: Completing collision detection verification of the path to identify potential conflicts between the path and obstacle boundaries, ensuring the safety of the path; Step S6.3: Completing the selection of compliant paths and the optimal path, locking in the core path that meets the optimal length requirement; Step S6.4: Optimizing the transition points of the path to improve path smoothness and adapt to the actual operating characteristics of the transport vehicle; Step S6.5: Calculating the core performance indicators of the path to provide a quantitative reference for path rationality assessment; Step S6.6: Completing the overlay verification of the path and the global environment map to confirm the adaptability of the path to various constraints; and Step S6.7: Completing the final path confirmation output to form the globally optimal scheduling path for the raw tobacco transportation task.

[0286] In summary, the overall beneficial effects of steps S1 to S6 are as follows: This application addresses existing problems in the route planning and scheduling of transport vehicles in raw tobacco logistics workshops. It eliminates excessive reliance on preset parameters and human experience, establishing a dynamic route planning and scheduling system adapted to workshop operation scenarios. The method enables adaptive responses to workshop environment and task requirements, improves the stability and executability of route planning results, optimizes the ability to generate scheduling routes under different operation scenarios, ensures scheduling efficiency and planning accuracy in raw tobacco logistics transportation, and adapts to the practical application needs of intelligent raw tobacco logistics workshops for globally optimal scheduling. In this process, the global environment map constructed in step S1 provides a precise and unified spatial foundation for path planning, ensuring the adaptability of the planned path to the actual operation scenario in the workshop; the construction of the hybrid initial solution space completed in step S2 provides a reasonable initial foundation for algorithm iteration, narrowing the search range for subsequent optimization; the dynamic collaborative update of algorithm parameters implemented in step S3 can adapt to changes in the search state during the iteration process, alleviating the local optimum problem caused by fixed parameters; the enhanced path transition probability calculation model generated in step S4 can enrich the diversity of path selection and improve the global search capability of the algorithm; the path iterative search and pheromone update completed in step S5 can realize the dynamic optimization of the search space and accelerate the algorithm convergence process; and the globally optimal scheduling path output in step S6 can match the core requirements of the raw tobacco transportation task, ensuring the rationality and optimality of the final scheduling scheme.

[0287] like Figure 2As shown, this embodiment provides an example of a smart logistics scheduling device for raw tobacco. In this embodiment, the smart logistics scheduling device is applied to the smart logistics scheduling method as described in the above embodiment.

[0288] Specifically, the intelligent logistics scheduling device includes a global environment map construction module 1 for raw tobacco logistics operations, a hybrid initial solution space construction module 2 for path search, a path search rule generation module 3, an enhanced path transition probability calculation model generation module 4, a path iterative search and pheromone global update module 5, and a global optimal scheduling path acquisition module 6, which are connected electrically or through communication in sequence.

[0289] The module 1, which constructs a global environment map for raw tobacco logistics operations, is used to build a global environment map for raw tobacco logistics operations based on the operational space and obstacle distribution data of the intelligent logistics workshop for raw tobacco, through the fusion of topology modeling and mesh modeling. The module 2, which constructs a hybrid initial solution space for path search, is used to obtain the start and end node data of raw tobacco transportation tasks and, in conjunction with the global environment map for raw tobacco logistics operations, initialize the basic parameters of the ant colony algorithm and the goose colony optimization algorithm to construct a hybrid initial solution space for path search. The module 3, which generates path search rules, is used to dynamically adjust the pheromone utilization coefficient and heuristic weight of the ant colony algorithm based on the single-leg balancing strategy of the goose colony optimization algorithm, and generates a hybrid initial solution space for path search. The system generates path search rules with dynamically and collaboratively updated parameters; the enhanced path transition probability calculation model generation module 4 incorporates a random perturbation strategy into the path selection mechanism of the ant colony algorithm, and generates an enhanced path transition probability calculation model based on the path search rules; the path iterative search and pheromone global update module 5 dynamically compresses the path search space through the dimensional scaling factor of the goose colony optimization algorithm, so as to perform path iterative search and pheromone global update of the ant colony based on the enhanced path transition probability calculation model; and the globally optimal scheduling path acquisition module 6 obtains the globally optimal scheduling path for the raw tobacco transportation task based on the results of the path iterative search and pheromone global update.

[0290] Figure 3 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Figure 3 As shown, the electronic device 7 includes a processor 71 and a memory 72 coupled to the processor 71.

[0291] The memory 72 stores program instructions for implementing the intelligent logistics scheduling method for raw tobacco in any of the above embodiments.

[0292] The processor 71 is used to execute program instructions stored in the memory 72 for intelligent logistics scheduling of raw tobacco.

[0293] The processor 71 can also be referred to as a CPU (Central Processing Unit). The processor 71 may be an integrated circuit chip with signal processing capabilities. The processor 71 can also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. A general-purpose processor can be a microprocessor or any conventional processor.

[0294] Furthermore, Figure 4 This is a schematic diagram of the structure of a storage medium according to an embodiment of this application. See also: Figure 4 The storage medium 8 in this embodiment stores program instructions 81 capable of implementing all the above methods. These program instructions 81 can be stored in the storage medium as a software product, including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods in each embodiment of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks, or terminal devices such as computers, servers, mobile phones, and tablets.

[0295] In the several embodiments provided in this application, it should be understood that the disclosed apparatus, devices, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, signal, or other forms.

[0296] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units described above can be implemented in hardware or as software functional units. The above are merely embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made based on the description and drawings of this application, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A smart logistics scheduling method for raw tobacco, wherein the smart logistics scheduling method is applied in a smart logistics workshop for raw tobacco, characterized in that, The intelligent logistics scheduling method includes: Step S1: Based on the work space and obstacle distribution data of the raw tobacco smart logistics workshop, a global environment map of raw tobacco logistics operations is constructed by fusing topology modeling and grid modeling. Step S2: Obtain the start and end node data of the raw tobacco transportation task, and initialize the basic parameters of the ant colony algorithm and the goose colony optimization algorithm in combination with the global environment map of the raw tobacco logistics operation to construct a hybrid initial solution space for path search. Step S3: Dynamically adjust the pheromone utilization coefficient and heuristic weight of the ant colony algorithm through the single-leg balance strategy of the goose flock optimization algorithm, and dynamically and collaboratively update the path search rules based on the parameters generated by the path search hybrid initial solution space. Step S4: Incorporate a random perturbation strategy into the path selection mechanism of the ant colony algorithm, and generate an enhanced path transition probability calculation model based on the path search rules. Step S5: The path search space is dynamically compressed by the dimension scaling factor of the goose flock optimization algorithm to perform path iterative search and global pheromone update of the ant colony based on the enhanced path transition probability calculation model. Step S6: Based on the results of the path iterative search and pheromone global update, the globally optimal scheduling path for the raw tobacco transportation task is obtained.

2. The intelligent logistics scheduling method according to claim 1, characterized in that, Step S1: Based on the operational space and obstacle distribution data of the raw tobacco intelligent logistics workshop, a global environment map of the raw tobacco logistics operation is constructed by fusing topology modeling and mesh modeling, including: Step S1.1: Collect the three-dimensional point cloud data and two-dimensional planar drawing data of the raw tobacco intelligent logistics workshop, perform voxel filtering and coordinate system alignment on the three-dimensional point cloud data, and orthographically project the processed data onto the two-dimensional planar coordinate system to generate the basic two-dimensional planar data of the workshop. Step S1.2: Extract obstacle contours from the two-dimensional planar basic data of the workshop, perform morphological dilation operation on the obstacle contours to expand the safety margin, and generate workshop space constraint data with obstacle dilation boundaries; Step S1.3: Extract the key node coordinates of the workshop space constraint data, and establish an adjacency relationship table based on the access relationship between nodes to generate the original tobacco logistics workshop topology structure data; Step S1.4: Divide the workshop space constraint data into grids at a fixed resolution, and determine the overlap relationship between each grid and obstacles and mark the passage status to obtain the original tobacco logistics workshop grid map data; Step S1.5: Map the node coordinates of the topology map structure data of the raw tobacco logistics workshop to the corresponding grid positions of the grid map data of the raw tobacco logistics workshop, so as to establish a bidirectional index mapping relationship between topology nodes and grids, and obtain a topology-grid bidirectional index table. Step S1.6: Retrieve the grid accessibility status on the path connecting adjacent nodes of the topology graph structure data according to the topology-grid bidirectional index table, delete the topology edges with inaccessible grids, and obtain the topology-grid fusion map after spatial consistency verification. Step S1.7: Calculate the passage cost weight of each effective topological edge in the topology-grid fusion map based on the number of passable grids crossed and write it into the adjacency table to obtain the weighted topology-grid fusion map. Step S1.8: Overlay and fuse the weighted topology-grid fusion map with the raw tobacco logistics workshop grid map data in a unified coordinate system to obtain a global environment map of raw tobacco logistics operations.

3. The intelligent logistics scheduling method according to claim 1, characterized in that, Step S2: Obtain the start and end node data of the raw tobacco transportation task, and initialize the basic parameters of the ant colony algorithm and the goose colony optimization algorithm in conjunction with the global environment map of the raw tobacco logistics operation to construct a hybrid initial solution space for path search, including: Step S2.1: Obtain the coordinates of the starting point and ending point of the raw tobacco transportation task to obtain the starting and ending point data of the raw tobacco transportation task. Step S2.2: Extract all feasible path sets of the start and end node data based on the global environment map of the original tobacco logistics operation using the breadth-first search algorithm, and calculate the total path length of each feasible path to obtain the feasible path set and path length set corresponding to the start and end nodes; Step S2.3: Perform Min-Max normalization on the path length set to obtain the normalized path length set for the raw tobacco transportation task; Step S2.4: Initialize the basic parameters of the ant colony algorithm, such as the number of ants, the maximum number of iterations, the initial value of pheromones, the pheromone volatility coefficient, the pheromone utilization coefficient, and the heuristic weights, and define the initial pheromone allocation intensity according to the normalized path length set to obtain the ant colony algorithm initialization parameter set. Step S2.5: Initialize the basic parameters of the goose flock optimization algorithm, such as population size, maximum number of iterations, problem dimension, stage selection probability, decay factor, and random coefficient, to obtain the initial parameter set of the goose flock optimization algorithm. Step S2.6: Based on the ant colony algorithm initialization parameter set and the goose flock optimization algorithm initialization parameter set, and using the start and end node data as constraints, randomly initialize the ant colony positions on the global environment map of the original tobacco logistics operation, calculate the initial path length of each ant, and generate an initial path set. Step S2.7: Calculate the objective function value for each path in the initial path set and sort them. Identify the initial path with the optimal objective function value as the current optimal solution and generate an initial path set with objective function value evaluation. Step S2.8: Integrate the ant colony algorithm initialization parameter set, the goose colony optimization algorithm initialization parameter set, and the initial path set to obtain a hybrid initial solution space for path search.

4. The intelligent logistics scheduling method according to claim 1, characterized in that, Step S3 involves dynamically adjusting the pheromone utilization coefficient and heuristic weights of the ant colony algorithm using the single-leg balancing strategy of the goose colony optimization algorithm, and generating dynamically collaboratively updated path search rules based on the path search hybrid initial solution space parameters, including: Step S3.1: Based on the initial stage of the goose flock optimization algorithm, calculate the weight of the random stone, the stone falling time, the sound propagation time, and the average time within the mixed initial solution space of the path search to obtain the time parameter set for the current iteration; Step S3.2: Calculate the stone impact velocity of the current iteration time parameter set, as well as the relationship between the sound propagation distance and the distance between individual guard geese, to obtain the impact velocity set of the current iteration; Step S3.3: Normalize the impact velocity set of the current iteration and calculate the impact velocity adjustment factor as the normalized value of the maximum impact velocity. Step S3.4: Calculate the maximum sound propagation distance of individual geese in the current iteration, and the sound propagation distance adjustment factor is the value of the maximum sound propagation distance; Step S3.5: Based on the current iteration number and the maximum iteration number, calculate the adaptive iteration weight α decay coefficient and β decay coefficient using the exponential decay formula to obtain the parameter decay coefficient set; Step S3.6: Substitute the impact velocity adjustment factor, the sound propagation distance adjustment factor, and the parameter attenuation coefficient set into the dynamic parameter adjustment formula to calculate the dynamic adjustment pheromone utilization coefficient and the dynamic adjustment heuristic weight of the ant colony algorithm, and obtain the parameter set of the ant colony algorithm after dynamic collaborative parameter update. Step S3.7: Redefine the path transition probability rule of the ant colony algorithm according to the parameter set of the ant colony algorithm and the mixed initial solution space of the path search algorithm to obtain the path search rule after dynamic collaborative parameter update. Step S3.8: Integrate the ant colony algorithm parameter set with the path search rules to obtain the path search rules after dynamic collaborative parameter updates.

5. The intelligent logistics scheduling method according to claim 1, characterized in that, Step S4: Incorporate a random perturbation strategy into the path selection mechanism of the ant colony algorithm, and generate an enhanced path transition probability calculation model based on the path search rules, including: Step S4.1: In the exploration phase of the goose flock optimization algorithm, calculate the dynamic decay factor of the current iteration number based on the path search rule; Step S4.2: Calculate the wake-up intensity coefficient and minimum perturbation threshold in the random perturbation strategy based on the dynamic decay factor to obtain the random perturbation strategy parameter set; Step S4.3: Obtain the obstacle density distribution in the global environment map of the raw tobacco logistics operation, and identify high obstacle density areas and low obstacle density areas to obtain environmental obstacle density distribution data; Step S4.4: Dynamically adjust the wake-up intensity coefficient in the random perturbation strategy parameter set according to the environmental obstacle density distribution data in a proportional manner to obtain an obstacle adaptive perturbation parameter set; Step S4.5: Sample the random numbers from the standard normal distribution and multiply them by the wake-up intensity coefficient in the perturbation parameter set to obtain the random perturbation term; Step S4.6: Add the random perturbation term to the state transition probability formula of the ant colony algorithm to obtain the path transition probability calculation model incorporating the random perturbation strategy; Step S4.7: Based on the path transition probability calculation model and the ant colony algorithm parameter set, obtain the ant path selection result with embedded wake-up mechanism through ant colony path selection; Step S4.8: Integrate the ant path selection results, the ant colony algorithm parameter set, and the path transition probability calculation model to obtain the enhanced path transition probability calculation model.

6. The intelligent logistics scheduling method according to claim 1, characterized in that, Step S5 involves dynamically compressing the path search space using the dimensionality scaling factor of the goose flock optimization algorithm to perform iterative path search and global pheromone update for the ant colony based on the enhanced path transition probability calculation model, including: Step S5.1: Perform path iterative search on the enhanced path transition probability calculation model using the ant population, and calculate the path length of each ant in each iteration to obtain the ant path length set for the current iteration. Step S5.2: Select the path corresponding to the ant with the smallest path length in the ant path length set as the current global optimal path, and update the global optimal path length and global optimal path position to obtain the global optimal path information for the current iteration. Step S5.3: Calculate the pheromone increment released by each ant on its own path according to the pheromone update rule of the ant colony algorithm, and sum up the pheromone increments of all ants to obtain the total pheromone increment set for the current iteration. Step S5.4: Apply the pheromone evaporation coefficients from the ant colony algorithm parameter set after dynamic collaborative parameter update to the pheromones of all ants in the previous iteration, and obtain the pheromone matrix after pheromone evaporation through pheromone evaporation update. Step S5.5: Add the total pheromone increment set of the current iteration to the pheromone matrix to perform a global pheromone update, and obtain the updated pheromone matrix; Step S5.6: Based on the updated pheromone matrix and the current iteration number, the path length calculation formula is dynamically adjusted using the dimension scaling factor and the maximum Euclidean distance to compress and compensate the high-dimensional path, thereby obtaining the path distance calculation formula after dimension scaling compensation. Step S5.7: Recalculate the distance values ​​of all paths in the current population using the path distance calculation formula to update the global optimal path information and obtain the path iteration search results after dimensional compression. Step S5.8: Determine whether the maximum number of iterations has been reached or the convergence criterion has been met. If at least one of them is met, terminate the iteration and obtain the path iteration search results after dimensional compression and the updated global optimal path information.

7. The intelligent logistics scheduling method according to claim 1, characterized in that, Step S6: Based on the results of the path iterative search and pheromone global update, obtain the globally optimal scheduling path for the raw tobacco transportation task, including: Step S6.1: Based on the results of path iterative search and pheromone global update, extract the global optimal path and global optimal path length in the final iteration process to obtain the optimal path candidate set for the raw tobacco transportation task. Step S6.2: Perform collision detection on each path in the optimal path candidate set to verify whether the path intersects with the obstacle expansion boundary in the global environment map of the original tobacco logistics operation, and obtain the collision detection verification result. Step S6.3: Based on the collision detection verification results, filter all paths that pass the collision detection. If there are multiple paths that pass the collision detection, select the path with the shortest path length as the final optimal path to obtain the global optimal scheduling path for the raw tobacco transportation task. Step S6.4: Remove the transition points in the globally optimal scheduling path to obtain the smoothed optimal scheduling path; Step S6.5: Calculate the total length, expected running time, and energy consumption assessment value of the smoothed optimal scheduling path to obtain the path performance assessment data of the raw tobacco transportation task; Step S6.6: Overlay the smoothed optimal scheduling path with the original tobacco logistics operation global environment map for verification to confirm that the smoothed optimal scheduling path has met all constraints and performance requirements. Step S6.8: After confirmation, the globally optimal scheduling path for the raw tobacco transportation task is obtained.

8. A smart logistics scheduling device for raw tobacco, wherein the smart logistics scheduling device is applied to the smart logistics scheduling method as described in any one of claims 1 to 7, characterized in that, The intelligent logistics dispatching device includes: The raw tobacco logistics operation global environment map construction module is used to construct a raw tobacco logistics operation global environment map based on the operation space and obstacle distribution data of the raw tobacco smart logistics workshop through the fusion of topology modeling and grid modeling. The path search hybrid initial solution space construction module is used to obtain the start and end node data of the raw tobacco transportation task, and to initialize the basic parameters of the ant colony algorithm and the goose colony optimization algorithm in combination with the global environment map of the raw tobacco logistics operation, so as to construct the path search hybrid initial solution space. The path search rule generation module is used to dynamically adjust the pheromone utilization coefficient and heuristic weight of the ant colony algorithm through the single-leg balance strategy of the goose flock optimization algorithm, and generate path search rules with dynamically and collaboratively updated parameters based on the path search hybrid initial solution space. An enhanced path transition probability calculation model generation module is used to incorporate a random perturbation strategy into the path selection mechanism of the ant colony algorithm and generate an enhanced path transition probability calculation model based on the path search rules. The path iterative search and pheromone global update module is used to dynamically compress the path search space through the dimension scaling factor of the goose flock optimization algorithm, so as to perform path iterative search and pheromone global update of the ant population based on the enhanced path transition probability calculation model. The global optimal scheduling path acquisition module is used to obtain the global optimal scheduling path for the raw tobacco transportation task based on the results of the path iterative search and global pheromone update.

9. An electronic device, characterized in that, The system includes a processor and a memory coupled to the processor, the memory storing program instructions executable by the processor; when the processor executes the program instructions stored in the memory, it implements the intelligent logistics scheduling method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores program instructions that, when executed by a processor, enable the intelligent logistics scheduling method as described in any one of claims 1 to 7.

Images