Container factory container truck path planning method and system based on Beidou positioning

By using Beidou positioning technology to build a real-life three-dimensional model and dynamic traffic distribution model in the container factory, the accuracy and robustness of the scheduling of the existing system are solved, and efficient and safe scheduling path planning is achieved.

CN119063716APending Publication Date: 2024-12-03WUHAN YUHUIHONG TECH CO LTD

Patent Information

Application Number
CN202411229692.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-03
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

The existing intelligent port scheduling and command system for container trucks cannot meet the accuracy and robustness requirements of the intelligent truck scheduling method, and a single task scheduling optimization cannot guarantee the vehicle's execution of the scheduling scheme and the safety of the operating environment.

Method used

The container factory collector path planning method based on Beidou positioning is adopted, and the road network information is extracted through a real-life three-dimensional model and a dynamic traffic allocation model is constructed to optimize the traffic flow in the container factory and generate the optimal transportation path.

Benefits of technology

It improves the accuracy of card scheduling execution and the efficiency of path planning, ensures the accuracy and safety of path planning, adapts to specific operational tasks, and optimizes traffic flow within the container factory area.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119063716A_ABST
    Figure CN119063716A_ABST
Patent Text Reader

Abstract

The invention provides a container factory container truck path planning method and system based on Beidou positioning, and relates to the technical field of traffic transportation, and the method comprises the steps: collecting geographic space data, dynamic operation data and container truck transportation data, the container truck transportation data comprises real-time position information and real-time state information of a container truck; constructing a live-action three-dimensional model corresponding to the geographic space data based on the geographic space data; extracting road network information from the live-action three-dimensional model to identify each transportation node in a transportation area and a link formed between any two adjacent transportation nodes; constructing a dynamic traffic distribution model according to the real-time position information and the real-time state information; inputting the container truck average density, the container truck average speed and the container truck average flow in each link into the dynamic traffic distribution model to predict state variables of all container trucks in a transportation area in the container factory; and based on the state variables and the dynamic operation data, generating an optimal transportation path for each container truck in combination with a target path optimization function and a time cost function. According to the invention, the accuracy of container truck scheduling execution can be improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of transportation technology, and in particular to a container truck path planning method and system based on Beidou positioning. Background Art

[0002] As an important link between manufacturing and logistics, container factories occupy a key position in the global supply chain system. In recent years, with the vigorous development of international trade and the continuous growth of container demand, the diversification and customization of containers have become increasingly obvious, which has put forward higher requirements for the production efficiency and flexibility of container factories. With the widespread application of intelligent manufacturing technology, the degree of automation of container factories has been significantly improved, and production efficiency and product quality have been greatly improved accordingly. As the core equipment of modern container factories, intelligent production lines undertake the main task of container manufacturing.

[0003] The Chinese patent with the publication number CN116092285A discloses an intelligent port dispatching and commanding system for container trucks, including a database server, a BDS receiver as a base station component, at least one BDS wireless terminal, a video monitoring subsystem, a GPU server with a vehicle path planning subsystem, and a BDS server with a distributed processing subsystem. The present invention also discloses an intelligent port dispatching and commanding method for container trucks, which executes the following steps through the distributed processing subsystem: S51, obtaining the current positioning data of the container truck; S52, obtaining the real-time picture and pushing it to the genetic algorithm software module; S53, establishing the operation path calculation model of the container truck; S54, performing image recognition based on the convolutional neural network; S55, establishing the optimal planning model of the container truck path and sending it to the vehicle path planning subsystem. However, the above scheme does not take into account the execution capability of the container truck, and the dispatching decision is separated from the execution of the container truck, which cannot meet the accuracy and robustness requirements of the intelligent container truck dispatching method, and the single task dispatching optimization cannot guarantee the execution of the dispatching scheme by the vehicle and the safety of the working environment. Therefore, it is very necessary to provide a container factory truck path planning method and system based on Beidou positioning to improve the accuracy of truck scheduling execution. Summary of the invention

[0004] In view of this, the present invention proposes a container truck path planning method and system based on Beidou positioning. By extracting road network information from the real-life 3D model and constructing a dynamic traffic allocation model, it can accurately reflect the real-time traffic conditions in the container factory, optimize the traffic flow in the container factory, and thus improve the accuracy of truck dispatching execution.

[0005] The present invention provides a container truck path planning method based on Beidou positioning, the method comprising:

[0006] Collecting geospatial data, dynamic operation data and container truck transportation data, wherein the container truck transportation data includes real-time location information and real-time status information of container trucks;

[0007] Based on the geospatial data, construct a real-scene three-dimensional model corresponding to the geospatial data;

[0008] Extracting road network information from the real-scene three-dimensional model to identify each transport node in the transport area and a link formed between any two adjacent transport nodes;

[0009] Constructing a dynamic traffic allocation model according to the real-time location information and the real-time status information;

[0010] Inputting the average density of container trucks, the average speed of container trucks and the average flow volume of container trucks in each link into the dynamic traffic assignment model to predict the state variables of all container trucks in the transportation area of ​​the container factory;

[0011] Based on the state variables and the dynamic operation data, an optimal transportation path is generated for each container truck in combination with a target path optimization function and a time cost function.

[0012] On the basis of the above technical solution, preferably, constructing a real-scene three-dimensional model corresponding to the geographic spatial data based on the geographic spatial data specifically includes:

[0013] Preprocessing the satellite images, point cloud data, vector map data and digital elevation model data in the geospatial data respectively to obtain satellite optimized images, point cloud optimized data, vector map optimized data and digital elevation model optimized data;

[0014] Based on the digital elevation model optimization data and the triangulation algorithm, an initial terrain model is constructed, and the orthophoto in the satellite optimization image is mapped onto the initial terrain model as a terrain texture to obtain a three-dimensional terrain model;

[0015] Extracting building outlines and building height information respectively according to the point cloud optimization data and the vector map optimization data to generate an initial building model, and mapping the building image in the satellite optimization image as a building texture to the initial building model to obtain a three-dimensional building model;

[0016] The three-dimensional terrain model and the three-dimensional building model are integrated to construct the real-scene three-dimensional model.

[0017] On the basis of the above technical solution, preferably, the three-dimensional terrain model and the three-dimensional building model are integrated to construct a real-scene three-dimensional model, which specifically includes:

[0018] Converting the coordinate systems of the three-dimensional terrain model and the three-dimensional building model to unify the coordinate systems of the three-dimensional terrain model and the three-dimensional building model;

[0019] Positioning the three-dimensional building model at a corresponding position on the three-dimensional terrain model according to the geographical location information of the three-dimensional building model;

[0020] Performing terrain trimming at the corresponding building position on the three-dimensional terrain model, and unifying rendering parameters of the three-dimensional terrain model and the three-dimensional building model;

[0021] The processed three-dimensional terrain model and the three-dimensional building model are integrated into a preset data structure to construct the real-scene three-dimensional model.

[0022] More preferably, the satellite images, point cloud data, vector map data and digital elevation model data in the geospatial data are preprocessed separately, specifically including:

[0023] Performing geometric correction and radiation correction on the satellite image to obtain the satellite optimized image;

[0024] The point cloud data are filtered and classified in sequence to obtain point cloud optimized data;

[0025] Performing topology check and attribute completion on the vector map data respectively to obtain the vector map optimization data;

[0026] The digital elevation model data is interpolated and smoothed in sequence to obtain the digital elevation model optimized data.

[0027] More preferably, after constructing the real-scene three-dimensional model corresponding to the geospatial data based on the geospatial data, the method further includes:

[0028] rasterizing the real-scene 3D model so as to convert the real-scene 3D model into a 2.5D raster map having location information of container distribution and arrangement and ground elevation value information;

[0029] On the 2.5D grid map, grid cells with excessive height are marked as obstacle cells, and according to the operation requirements of the current transportation operation task, the transportation area and the unloading area of ​​the current operation task are divided on the 2.5D grid map.

[0030] More preferably, the expressions for the average density of container trucks, the average speed of container trucks and the average flow rate of container trucks are:

[0031]

[0032] fr m,i (k) = ρ m,i (k)·v m,i (k) LN m

[0033]

[0034] Where T represents the discrete time step, K represents the time series variable, and ρ m,i (k) represents the average density of container trucks in the ith link segment of the mth link at the kth time step, v m,i (k) represents the average speed of the container truck in the i-th link segment of the m-th link at the k-th time step, fr m,i (k) represents the average flow of container trucks in the ith link segment of the mth link at the kth time step, ρ s,m represents the standard density of container trucks on the mth link, L m Indicates the length of the mth link, LN m represents the number of lanes in the mth link, ξ represents the speed response time constant, κ represents the expected density parameter, and η represents the spatial expected parameter.

[0035] More preferably, the predicting of the state variables of all container trucks in the transport area of ​​the container plant specifically includes:

[0036] Based on the average flow of container trucks when they are located at any link, the maximum flow in the link, and the inflow demand of container trucks, the real-time flow of the container trucks entering the link in the transport area is obtained;

[0037] According to the real-time flow volume and the queue length evolution function, obtaining the length of the container truck queue corresponding to the real-time flow volume;

[0038] Based on the length of the container truck queue, the average density of the container trucks, the average speed of the container trucks and the average flow rate of the container trucks, a state variable corresponding to the length of the container truck queue is predicted.

[0039] More preferably, the expressions of the target path optimization function and the time cost function are respectively:

[0040]

[0041] x(k+1)=f(x(k),u(k c ),d(k))

[0042]

[0043] Where P(k) represents the target path optimization function, θ[] represents the penalty function, x(k) represents the state variable of the container truck in the prediction period corresponding to the kth time step, and K represents the time series variable. represents the performance index function, u(k) represents the control variable corresponding to the prediction period of the kth time step, d(k) represents the system disturbance term corresponding to the prediction period of the kth time step, u(k c ) represents the kth c The control variable of the control cycle, f() represents the state transfer function, TTS(k) represents the time cost function, T represents the discrete time step, K represents the time series variable, ρ m,i (k) represents the average density of container trucks in the ith link segment of the mth link at the kth time step, L m Indicates the length of the mth link, LN m represents the number of lanes in the mth link, M represents a constant, and w o (k) represents the queue length of container trucks waiting to perform transportation tasks at the kth time step, N p Represents the model prediction time.

[0044] More preferably, the method further comprises:

[0045] Set the maximum number of iterations, initial control variables, and step size parameters, and obtain state variables through forward recursive calculations and co-state variables through backward recursive calculations;

[0046] Calculating the gradient of the target path optimization function with respect to the control variable according to the state variable and the co-state variable;

[0047] Based on the current gradient and the gradient of the previous iteration, dynamically adjust the step size parameter to obtain the updated step size parameter, and calculate the updated control variable after iteration according to the control variable iteration function, the updated step size parameter and the current gradient;

[0048] If the number of iterations reaches the preset maximum number of iterations, the iteration is stopped and the optimal control variable is output.

[0049] In a second aspect of the present application, a container plant truck path planning system based on Beidou positioning is provided, wherein the container plant truck path planning system includes a data acquisition management module, a model building optimization module and a path prediction optimization module, wherein:

[0050] The data collection and management module is used to collect geographic space data, dynamic operation data and container truck transportation data, wherein the container truck transportation data includes real-time location information and real-time status information of container trucks;

[0051] The model building optimization module is used to build a real-scene three-dimensional model corresponding to the geospatial data based on the geospatial data, extract road network information from the real-scene three-dimensional model to identify each transportation node in the transportation area and the link formed between any two adjacent transportation nodes, and build a dynamic traffic allocation model according to the real-time location information and the real-time status information;

[0052] The path prediction optimization module is used to input the average density of container trucks, the average speed of container trucks and the average flow of container trucks in each link into the dynamic traffic allocation model to predict the state variables of all container trucks in the transportation area of ​​the container factory, and generate the optimal transportation path for each container truck based on the state variables and the dynamic operation data in combination with the target path optimization function and the time cost function.

[0053] The container truck path planning method and system based on Beidou positioning provided by the present invention has the following beneficial effects compared with the prior art:

[0054] (1) By collecting multi-dimensional data and constructing a real-life 3D model, an accurate digital expression of the container factory environment is achieved. At the same time, the road network information is extracted from the real-life 3D model and a dynamic traffic allocation model is constructed, which can accurately reflect the real-time traffic conditions in the container factory. The dynamic traffic allocation model takes into account multiple factors such as target path optimization and time cost. Combined with the predicted state variables, the optimal transportation path can be dynamically generated, and the efficiency of path planning can be improved. The speed of solving the paths of multiple container trucks can be improved, and effective anti-collision path planning can be performed according to the actual transportation task scenario, optimizing the traffic flow in the container factory, thereby improving the accuracy of container truck scheduling execution;

[0055] (2) Rasterization is used to simplify the complex three-dimensional model into an easy-to-process 2.5D grid map, which greatly improves the data processing efficiency and the speed of path planning. The 2.5D grid map retains the container distribution location and ground elevation information, providing more accurate terrain data support for subsequent path planning. At the same time, grid cells with excessive height are marked as obstacle cells, which effectively avoids planning inaccessible paths and improves the accuracy and safety of path planning. The transportation area and unloading area are divided according to the operation requirements, making the path planning more targeted and better meeting the needs of specific operation tasks. This processing method takes into account both two-dimensional plane and height information, and considers the impact of terrain undulations on path planning while ensuring calculation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0057] Figure 1 A schematic diagram of a process flow of a container truck path planning method in a container factory area based on Beidou positioning provided by the present invention;

[0058] Figure 2 A schematic diagram of the framework of a container factory truck path planning system based on Beidou positioning provided by the present invention.

[0059] Explanation of the accompanying drawings: 1. Container factory truck path planning system; 11. Data collection and management module; 12. Model building and optimization module; 13. Path prediction and optimization module. DETAILED DESCRIPTION

[0060] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0061] The present invention discloses a container truck path planning method based on Beidou positioning, referring to Figure 1 The method includes steps S1 to S6.

[0062] Step S1, collecting geographic space data, dynamic operation data and container truck transportation data, wherein the container truck transportation data includes real-time location information and real-time status information of container trucks.

[0063] In this step, geospatial data includes terrain data, such as the height and slope of the factory area; road layout, such as the road network, road width, turning radius, etc. within the factory area; building location, such as the location and size of warehouses, office buildings and other buildings; container yard layout, such as the location and capacity of each yard. Dynamic operation data includes loading and unloading plans, such as the loading and unloading time and location of each container; yard usage, such as the real-time occupancy rate of each yard; equipment status, such as the working status and efficiency of cranes and loading and unloading equipment; and personnel scheduling, such as the work arrangement and location of operators.

[0064] Furthermore, each truck is equipped with a Beidou-compatible receiver and communication module. The receiver can receive signals from Beidou satellites, and the communication module is responsible for data transmission. The real-time location information obtained by the Beidou satellite positioning system includes the latitude and longitude coordinates, altitude, driving direction (azimuth) and velocity vector of the truck. The real-time status information obtained by the Beidou satellite positioning system includes the current driving speed, load status (through weight sensor), fuel status, vehicle diagnostic information (engine status, tire pressure, etc.) and task execution status of the truck. The short message communication function of the Beidou satellite positioning system is used to realize real-time two-way transmission of data. In areas with weak signals in the container plant, other communication networks (such as 4G / 5G) can be combined to ensure the continuity of data transmission. The container plant truck path planning system receives real-time data from all trucks, combines the precise time information provided by the Beidou system, realizes the time and space synchronization of data, and processes and analyzes the collected data in real time through big data analysis technology.

[0065] Step S2: constructing a real-scene three-dimensional model corresponding to the geographic spatial data based on the geographic spatial data.

[0066] This step also includes steps S21 to S24.

[0067] Step S21 , pre-processing the satellite images, point cloud data, vector map data and digital elevation model data in the geospatial data respectively to obtain satellite optimized images, point cloud optimized data, vector map optimized data and digital elevation model optimized data.

[0068] In this step, the satellite image is geometrically corrected and radiometrically corrected to obtain the satellite optimized image; the point cloud data is filtered and classified in turn to obtain the point cloud optimized data; the vector map data is topologically checked and attributed to obtain the vector map optimized data; the digital elevation model data is interpolated and smoothed in turn to obtain the digital elevation model optimized data.

[0069] Step S22, constructing an initial terrain model based on the digital elevation model optimization data and the triangulation algorithm, and mapping the orthophoto in the satellite optimization image as a terrain texture onto the initial terrain model to obtain a three-dimensional terrain model.

[0070] In this step, the original elevation data collected by the Beidou receiver is filtered and denoised, and the cross-validation results in the quality control step are used to correct outliers. The triangulated irregular network (TIN) method is used to process elevation data, which is particularly suitable for representing complex terrain to retain the high-precision feature point information collected by the Beidou system. The high-resolution DEM data collected by the Beidou system is combined with drones to increase terrain details, and terrain refinement algorithms such as fractal algorithms are applied to further enhance the sense of reality. The orthophotos obtained by drone photogrammetry are used as textures to ensure accurate alignment with Beidou positioning data, and texture fusion technology is applied to eliminate seams and color differences. The real-time nature of the Beidou system is used to establish a dynamic update mechanism for the terrain model, and an automated periodic update process is set up for areas with variable terrain.

[0071] Step S23, extracting building outlines and building height information respectively according to the point cloud optimization data and the vector map optimization data to generate an initial building model, and mapping the building image in the satellite optimization image as a building texture to the initial building model to obtain a three-dimensional building model.

[0072] In this step, a Beidou-compatible high-precision receiver (such as a survey-grade GNSS receiver) is equipped in the container factory area, and the receiver supports Beidou B1, B2, B3 and other multi-frequency signals. Control points are set at key locations in the container factory area, and the receiver is used for long-term static observation (usually 2-4 hours) to collect multi-frequency Beidou signals. At the same time, other satellite system signals such as GPS and GLONASS are received to improve accuracy, and observation data, including original observation values ​​and ephemeris data, are recorded. Use a vehicle-mounted Beidou receiver to drive along the container factory road, use the real-time dynamic (RTK) measurement mode to achieve centimeter-level accuracy, and continuously record information such as the centerline coordinates and width of the road. Use a handheld Beidou receiver to walk around the building, stay at the corner of the building for a long time to ensure accurate coordinates, and record additional information such as the building height and number of floors to obtain the building outline information. Sample and measure the terrain change points and elevation feature points, and use the receiver's tilt measurement function to measure points that are not easy to reach directly to achieve terrain feature point measurement. Install Beidou terminals on containers or transport vehicles, use Beidou short message communication function to transmit location information in real time, and establish a real-time update mechanism for container locations. Combine Beidou receivers with drones to conduct aerial photogrammetry, use Beidou satellite positioning system to provide accurate location and attitude information for drones, and generate high-precision digital elevation models (DEMs) through photogrammetry methods.

[0073] The building corner coordinates collected by BeiDou receivers are used to accurately draw the building outline, and the outline information is verified and improved by combining GIS data. The tilt measurement function of BeiDou receivers is used to obtain the building height and roof shape. For complex buildings, drone photogrammetry data is combined for modeling. The basic geometric shape of the building is constructed based on the precise size data collected by the BeiDou system. A parametric modeling method is used to quickly generate a large number of similar buildings. High-resolution photos taken on site are used for precise texture mapping to ensure that the texture corresponds accurately to the building geometry located by BeiDou. The continuous monitoring capability of the BeiDou system is used to achieve dynamic updates of the building model.

[0074] Use the continuous coordinate data collected by the vehicle-mounted Beidou receiver to accurately construct the road centerline, use the centimeter-level accuracy data obtained by the RTK measurement mode to define the road width and lane information, use spline curves to generate smooth road curves based on the high-density point data collected by the Beidou system, apply the correct road cross-section profile to ensure consistency with the actual road conditions, use field survey data to apply appropriate road surface textures and materials, and apply more detailed textures for special sections such as intersections or turns. According to the precise location information of Beidou positioning, add traffic lights, road signs, guardrails and other facilities to ensure that the locations of these facilities in the model are completely consistent with the actual ones. Use the location and size data of key facilities such as container yards and warehouses collected by the Beidou system for modeling, and establish an update mechanism based on Beidou real-time data for dynamically changing facilities such as container locations.

[0075] Step S24, integrating the three-dimensional terrain model and the three-dimensional building model to construct a real-scene three-dimensional model.

[0076] Furthermore, the coordinate systems of the three-dimensional terrain model and the three-dimensional building model are converted to unify the coordinate systems of the three-dimensional terrain model and the three-dimensional building model; the three-dimensional building model is positioned at a corresponding position on the three-dimensional terrain model according to the geographical location information of the three-dimensional building model; the terrain is trimmed at the corresponding building position on the three-dimensional terrain model, and the rendering parameters of the three-dimensional terrain model and the three-dimensional building model are unified; the processed three-dimensional terrain model and the three-dimensional building model are integrated into a preset data structure to construct a real-life three-dimensional model.

[0077] Align the georeference and coordinate system, it is understood that the coordinate system suitable for the project area is selected, such as WGS84 or local coordinate system, the original data collected by the Beidou system (usually CGCS2000 coordinate system) is converted to the selected project coordinate system, and high-precision coordinate transformation is performed. Use the control points collected by static data to establish a high-precision control network, perform network adjustment calculations, ensure the coordinate consistency of the entire container factory area, use the control point network as a benchmark, accurately align all collected data, perform trajectory fitting and smoothing on dynamically collected data (such as road measurement), convert the ellipsoid height measured by the Beidou system to normal height (or other required elevation system), use an accurate geoid model for elevation conversion, align the data collected by the Beidou system with data from other sources (such as LiDAR point cloud, aerial imagery), use feature point matching, ICP algorithm and other methods to ensure accurate alignment between different data sources, use independent checkpoints to verify the accuracy of coordinate system conversion and alignment, analyze and record residuals to ensure that accuracy requirements are met.

[0078] It is understandable that Beidou terminals are installed on dynamic elements such as containers and transport vehicles, Beidou short message communication functions are configured to achieve real-time location information transmission, a central data processing center is established to receive real-time location information of all dynamic elements, and real-time data processing algorithms such as Kalman filtering are used to optimize location accuracy. Interfaces are reserved for dynamic elements in the three-dimensional model, and data structures are designed to support real-time updates of information such as the position, speed, and direction of dynamic elements. By introducing a real-time rendering engine in the central data processing center, the three-dimensional model can update the position and state of dynamic elements according to the received Beidou positioning data, and realize smooth movement and rotation of dynamic elements. Based on real-time location data, collision detection between dynamic elements is realized, and path planning algorithms are integrated to provide real-time navigation support for dynamic elements. Geographic fences are set to trigger alarms when dynamic elements exceed the predetermined area, monitor abnormal behaviors (such as long-term stillness, speeding), and alarm in time.

[0079] This step also includes steps S25 to S26.

[0080] Step S25, rasterizing the real-scene 3D model so as to convert the real-scene 3D model into a 2.5D raster map having the location information of the container distribution arrangement and the ground elevation value information.

[0081] In this step, according to the size of the container and the operating accuracy of the transport vehicle, select an appropriate grid size, usually 0.5m-1m. Project the three-dimensional model onto a two-dimensional plane, usually using the orthographic projection method. For each grid cell, extract its corresponding ground elevation value, using digital elevation model (DEM) data or obtaining elevation information by sampling the three-dimensional model surface. Analyze the distribution of containers in the three-dimensional model, map their location information to the corresponding grid cells, and record whether each grid cell is occupied by a container. Assign attribute values ​​to each grid cell, including the ground elevation value, the container occupancy status (occupied / unoccupied), and the container height (if occupied). Special treatment is performed on the boundary area of ​​the model to ensure the integrity of the rasterized map.

[0082] Step S26, marking the grid cells with excessive height as obstacle cells on the 2.5D grid map, and dividing the transportation area and the unloading area of ​​the current operation task on the 2.5D grid map according to the operation requirements of the current transportation operation task.

[0083] In this step, a height threshold is set, usually based on the maximum passing height of the transport vehicle. Traverse all grid cells and mark cells above the threshold as obstacle cells. Consider the stacking height of containers and mark the grid cells that exceed the safe height as obstacles. Add a safety buffer around the obstacle cell, usually 1.5 times the width of the vehicle, and mark the grid cells in the buffer as obstacles to ensure transportation safety. Determine the area range where transportation operations are required based on the current task requirements. Mark the boundaries of the transportation area on the grid map, while considering the road width and turning radius to ensure the accessibility of the transportation area. Mark the designated unloading area on the grid map based on the job requirements. Design a dynamic update mechanism to adapt to the real-time changes in the container position and regularly update the container occupancy status in the grid map. After marking the obstacles and areas, perform path planning preprocessing, such as building a graph structure of the passable area to prepare for the subsequent path planning algorithm. Develop visualization tools to intuitively display the rasterized map, obstacle cells, and different areas, using different colors or patterns to represent obstacles, transportation areas, and unloading areas.

[0084] Setting up rasterization processing simplifies complex three-dimensional models into easy-to-process 2.5D raster maps, greatly improving data processing efficiency and path planning speed. The 2.5D raster map retains the container distribution location and ground elevation information, providing more accurate terrain data support for subsequent path planning. At the same time, the grid cells with excessive height are marked as obstacle cells, effectively avoiding the planning of inaccessible paths, improving the accuracy and safety of path planning, and dividing the transportation area and unloading area according to the operation requirements, making the path planning more targeted and better meeting the needs of specific operation tasks. This processing method takes into account both two-dimensional plane and height information, while ensuring calculation efficiency, it also considers the impact of terrain undulations on path planning.

[0085] Step S3, extracting road network information from the real-scene three-dimensional model to identify each transport node in the transport area and the link formed between any two adjacent transport nodes.

[0086] In this step, image processing techniques such as edge detection and morphological operations are used to identify the road outline from the top view, and a thinning algorithm (such as the Zhang-Suen algorithm) is applied to extract the road centerline. For wide areas (such as storage yards), the Voronoi diagram method is used to generate drivable paths. Road intersections are identified as main nodes, key locations (such as loading and unloading area entrances, parking spots) are marked as special nodes, and auxiliary nodes are added at fixed intervals on long straight roads to improve the flexibility of path planning. Adjacent nodes are connected to form initial links, and a unique identifier is assigned to each link. The geometric characteristics (length, curvature, etc.) and attribute information (width, type, etc.) of the link are recorded. A node-link relationship data structure, such as an adjacency list or adjacency matrix, is established to ensure the connectivity of the network and handle possible breaks or isolated points. The capacity of each link is evaluated, and factors such as road width and slope are considered. One-way and two-way links are marked, and special sections such as height and weight restrictions are identified and marked. Possible turning restrictions are handled at intersection nodes, and the allowed turning directions and prohibited turning actions are recorded. Assign estimated travel time and cost to each link, consider factors such as road type and road conditions, set speed limits, and mark special areas such as deceleration zones and danger zones. For large sites, build a hierarchical road network, including main roads and secondary roads, and realize the mapping relationship between different levels of road networks. Simplify complex intersections to improve path planning efficiency, while merging close parallel links to reduce redundancy.

[0087] Step S4, constructing a dynamic traffic allocation model based on the real-time location information and real-time status information.

[0088] In this step, the Beidou satellite positioning system receiver is used to collect vehicle location data, usually at a frequency of 1-5 seconds / time, RFID or Bluetooth beacon technology is used to collect container location information, and the location and status of fixed facilities (such as cranes) are collected through IoT devices. The vehicle terminal collects vehicle status (speed, load, fuel, etc.), the road sensor network collects road condition information (traffic flow, average speed, etc.), the camera system uses computer vision technology to identify congestion, and the loading and unloading area management system provides operation status and queue information.

[0089] In one example, the A* algorithm is used to generate a basic path set, and the Yen's algorithm is applied to generate K shortest paths. The road grade and turning restrictions are considered to generate realistic paths, and the genetic algorithm is used to generate a diverse path set. The simulated annealing algorithm is applied to explore unconventional path options. The expected travel time is calculated (considering current traffic conditions and historical data), fuel consumption and carbon emissions are estimated, the reliability of the path is evaluated (considering historical volatility), the complexity of the path (such as the number of turns, changes in road types, etc.), and the path is evaluated based on the above data. The analytic hierarchy process is used to determine the weight of each indicator, and the TOPSIS method is used for multi-criteria decision analysis. Check whether the path meets vehicle restrictions (such as height and weight restrictions) and verify whether the path complies with current traffic control regulations.

[0090] Step S5, inputting the average density of container trucks, the average speed of container trucks and the average flow volume of container trucks in each link into the dynamic traffic allocation model to predict the state variables of all container trucks in the transportation area of ​​the container factory.

[0091] In this step, the expressions for the average density, average speed and average flow of trucks are:

[0092]

[0093] fr m,i (k) = ρ m,i (k)·v m,i (k) LN m

[0094]

[0095] Where T represents the discrete time step, K represents the time series variable, and ρ m,i (k) represents the average density of container trucks in the ith link segment of the mth link at the kth time step, v m,i (k) represents the average speed of the container truck in the i-th link segment of the m-th link at the k-th time step, fr m,i (k) represents the average flow of container trucks in the ith link segment of the mth link at the kth time step, V[ρm,i (k)] represents the average density of container trucks as ρ m,i The expected speed of the container truck at time (k), v fm represents the free speed of the container truck on the mth link, ρ s,m represents the standard density of container trucks on the mth link, L m Indicates the length of the mth link, LN m represents the number of lanes of the mth link, α m represents the density sensitivity parameter, ξ represents the velocity response time constant, κ represents the expected density parameter, and η represents the spatial expectation parameter.

[0096] In an example, the total flow of container trucks at the nth transport node can be expressed as:

[0097]

[0098]

[0099] Among them, Q n (k) represents the total flow of container trucks when entering the nth transport node, n represents the transport node formed between two or more adjacent links, fr jNj (k) represents the flow of container trucks in the last segment of link j at the kth time step, represents the proportion of the mth link leaving the nth node, ρ j,1 represents the traffic density of the first link segment in the jth link at the kth time step, O n represents the starting point set of the card set, v j,Nj (k) The speed of the last segment (i.e., the exit segment) of the jth link at the kth time step, N m+1 represents the total number of segments in the jth link, I n represents the set of all links entering the nth node, v m,0 represents the speed of the mth link entering the first link segment, v m,0 Indicates the speed of the mth link entering the first link segment, fr m,0 Represents the average flow of the mth link entering the first link segment.

[0100] This step also includes steps S51 to S53.

[0101] Step S51, based on the average flow of container trucks when they are located at any link, the maximum flow in the link, and the inflow demand of container trucks, obtain the real-time flow of container trucks entering the link in the transportation area.

[0102] Step S52, according to the real-time flow volume and the queue length evolution function, obtain the truck queue length corresponding to the real-time flow volume.

[0103] The expression of the queue length evolution function is:

[0104]

[0105] w o (k+1)=w o (k)+T[d o (k)-I o (k)]

[0106] Among them, w o (k) represents the queue length of container trucks waiting to perform transportation tasks at the kth time step, d o (k) represents the inflow demand of container trucks at the starting point at the kth time step, I o (k) represents the real-time flow of container trucks entering the link in the transport area at the kth time step, Ep represents the maximum flow of container trucks entering the link in the transport area, and ρ max represents the congestion density of container trucks in the transport area, ρ j,1 Represents the traffic density of the first link segment in the j-th link at the k-th time step.

[0107] Step S53, predicting the state variable corresponding to the length of the container truck queue based on the length of the container truck queue, the average density of the container trucks, the average speed of the container trucks and the average flow of the container trucks.

[0108] Step S6, based on the state variables and dynamic operation data, the optimal transportation path is generated for each container truck in combination with the target path optimization function and the time cost function.

[0109] In this step, the expression of the target path optimization function is:

[0110]

[0111] x(k+1)=f(x(k),u(k c ),d(k))

[0112] Mk≤k c ≤M(k+1)

[0113] Where P(k) represents the target path optimization function, θ[] represents the penalty function, x(k) represents the state variable of the container truck in the prediction period corresponding to the kth time step, and K represents the time series variable. represents the performance index function, u(k) represents the control variable corresponding to the prediction period of the kth time step, d(k) represents the system disturbance term corresponding to the prediction period of the kth time step, u(k c ) represents the kth c is the control variable of a control cycle, f() represents the state transfer function, and M represents a constant.

[0114] In one example, a database connection (such as MongoDB or PostgreSQL) is used to read the latest truck status data, which includes vehicle ID, current location (latitude and longitude), load status, remaining fuel / power, and current speed. Real-time operation data is obtained through API calls or message queues (such as Kafka). Real-time operation data includes current task information, time window requirements, priority settings, and special transportation requirements. Set the global parameters of path planning, including the maximum calculation time (such as 30 seconds), the number of candidate paths K (such as K=5), and the initial weights of various cost factors (w1, w2,..., w5). Adjust the strategy of subsequent path search according to the priority of the task. High-priority tasks may allow higher computational complexity or more aggressive optimization.

[0115] Load the road network topology, use spatial query to select all transportation nodes and links in the planning area, create a graph object (such as using the NetworkX library), add all nodes and edges to the graph, and for each edge, calculate the initial weight weight = f (length, current traffic, speed limit, road type).

[0116] Furthermore, the A* algorithm is used as the main search method, and the expression of the heuristic function is: h(n) = w1*straight-line distance (n, target) + w2*time estimate (n, target) + w2*traffic factor. Among them, the straight-line distance can use the Haversine formula to calculate the spherical distance, the time estimate can be based on the historical average speed, and the traffic factor needs to consider the congestion of the current road section. The heuristic function weight is dynamically adjusted according to the search progress, and the traffic information is updated during the search process. K candidate paths are generated by Yen's algorithm, usually K = 5 to 10, and the path similarity metric (such as the Jaccard coefficient) is used to enforce sufficient differences between the generated paths. At the same time, the time window constraint is considered in the generation process, special road section restrictions are handled, and different candidate paths are generated in parallel using multi-threading, and a preliminary score is calculated for each generated path. The scoring factors include total length, estimated time, reliability, etc.

[0117] This step also includes steps S61 to S64.

[0118] Step S61, setting the maximum number of iterations, initial control variables and step size parameters, and obtaining state variables through forward recursive calculation, and obtaining co-state variables through reverse recursive calculation.

[0119] In this step, the state variables can be expressed as:

[0120] x(k)=[ρ1(k),v1(k),fr1(k),...,ρ m (k),v m (k),fr m (k),w1(k),...,w o (k)] T

[0121] The Hamiltonian function is taken for the performance index function in the target path optimization function. The Hamiltonian function can be expressed as:

[0122] H[x(k),u(k),λ(k+1),k]=Φ[x(k),u(k),λ(k+1),k]

[0123] +λ T (k+1)·f[x(k),u(k),λ(k+1),k]

[0124]

[0125] Co-variables can be expressed as:

[0126]

[0127] Step S62, calculating the gradient of the target path optimization function with respect to the control variable according to the state variable and the co-state variable.

[0128] The gradient of the control variable can be expressed as:

[0129]

[0130] Step S63, dynamically adjust the step size parameter based on the current gradient and the gradient of the previous iteration to obtain the updated step size parameter, and calculate the updated control variable after the iteration according to the control variable iteration function, the updated step size parameter and the current gradient.

[0131] In this step, the expression of the control variable iteration function is:

[0132]

[0133] u τ+1 (k c )=sat(u τ (k c )+Δu τ(k c ))

[0134] Among them, γ represents the step size parameter of the new control variable, sat() represents the saturation vector function when the learning rate is γ, and γ max represents the upper bound of the step size parameter γ, γ min represents the lower bound of the step size parameter γ, k c Indicates the kth c control cycle, u τ (k c ) indicates that after the τth iteration, the kth c The control variable value of a control cycle, Δu τ (k c ) indicates that after the τth iteration, the kth c The control variable increment of the control cycle, sign() represents the sign function, Indicates that in the τth iteration, the c The partial gradient of the objective function of the p-th control cycle with respect to the p-th control variable, γ + represents the acceleration factor, γ - Represents the deceleration factor.

[0135] Step S64: if the number of iterations reaches the preset maximum number of iterations, the iteration is stopped and the optimal control variable is output.

[0136] In this embodiment, the expression of the time cost function is:

[0137]

[0138] Where TTS(k) represents the time cost function, T represents the discrete time step, K represents the time series variable, and ρ m,i (k) represents the average density of container trucks in the ith link segment of the mth link at the kth time step, L m Indicates the length of the mth link, LN m represents the number of lanes in the mth link, M represents a constant, and w o (k) represents the queue length of container trucks waiting to perform transportation tasks at the kth time step, N p Represents the model prediction time.

[0139] By collecting multi-dimensional data and constructing a real-life 3D model, an accurate digital expression of the container factory environment is achieved. At the same time, the road network information is extracted from the real-life 3D model and a dynamic traffic distribution model is constructed, which can accurately reflect the real-time traffic conditions in the container factory. The dynamic traffic distribution model takes into account multiple factors such as target path optimization and time cost. Combined with the predicted state variables, it can dynamically generate the optimal transportation path, improve the efficiency of path planning, and increase the speed of solving the paths of multiple container trucks. Effective anti-collision path planning is performed according to the actual transportation task scenarios, which optimizes the traffic flow in the container factory and thereby improves the accuracy of container truck scheduling execution.

[0140] Based on the above method, the embodiment of the present application discloses a container factory area container truck path planning system based on Beidou positioning, referring to Figure 2 The container truck path planning system 1 includes a data collection management module 11, a model building optimization module 12 and a path prediction optimization module 13, wherein:

[0141] The data collection management module 11 is used to collect geographic space data, dynamic operation data and container truck transportation data, wherein the container truck transportation data includes the real-time location information and real-time status information of the container truck;

[0142] The model building optimization module 12 is used to build a real-scene three-dimensional model corresponding to the geospatial data based on the geospatial data, extract road network information from the real-scene three-dimensional model to identify each transport node in the transport area and the link formed between any two adjacent transport nodes, and build a dynamic traffic allocation model based on the real-time location information and real-time status information;

[0143] The path prediction optimization module 13 is used to input the average density of container trucks, the average speed of container trucks and the average flow of container trucks in each link into the dynamic traffic allocation model to predict the state variables of all container trucks in the transportation area of ​​the container factory, and generate the optimal transportation path for each container truck based on the state variables and dynamic operation data in combination with the target path optimization function and the time cost function.

[0144] In one example, the model building optimization module 12 is used to pre-process the satellite images, point cloud data, vector map data and digital elevation model data in the geospatial data to obtain satellite optimized images, point cloud optimized data, vector map optimized data and digital elevation model optimized data; based on the digital elevation model optimization data and the triangulation algorithm, an initial terrain model is constructed, and the orthophoto in the satellite optimized image is mapped as a terrain texture to the initial terrain model to obtain a three-dimensional terrain model; the building outline and building height information are extracted according to the point cloud optimization data and the vector map optimization data to generate an initial building model, and the building image in the satellite optimized image is mapped as a building texture to the initial building model to obtain a three-dimensional building model; the three-dimensional terrain model and the three-dimensional building model are integrated to construct a real-life three-dimensional model.

[0145] In one example, the model construction optimization module 12 is used to convert the coordinate systems of the three-dimensional terrain model and the three-dimensional building model so as to unify the coordinate systems of the three-dimensional terrain model and the three-dimensional building model; locate the three-dimensional building model at the corresponding position on the three-dimensional terrain model according to the geographical location information of the three-dimensional building model; perform terrain trimming at the corresponding building position on the three-dimensional terrain model, and unify the rendering parameters of the three-dimensional terrain model and the three-dimensional building model; integrate the processed three-dimensional terrain model and the three-dimensional building model into a preset data structure to construct a real-life three-dimensional model.

[0146] In one example, the model building optimization module 12 is used to perform geometric correction and radiation correction on satellite images to obtain satellite optimized images; filter and classify point cloud data in turn to obtain point cloud optimized data; perform topology check and attribute completion on vector map data to obtain vector map optimized data; interpolate and smooth digital elevation model data in turn to obtain digital elevation model optimized data.

[0147] In one example, the model building optimization module 12 is used to rasterize the real-life three-dimensional model so that the real-life three-dimensional model is converted into a 2.5D grid map with location information of container distribution and ground elevation value information; grid cells with excessive height are marked as obstacle cells on the 2.5D grid map, and the transportation area and unloading area of ​​the current operation task are divided on the 2.5D grid map according to the operation requirements of the current transportation operation task.

[0148] In one example, the expressions for the average density of trucks, the average speed of trucks, and the average flow of trucks are:

[0149]

[0150] fr m,i(k) = ρ m,i (k)·v m,i (k) LN m

[0151]

[0152] Where T represents the discrete time step, K represents the time series variable, and ρ m,i (k) represents the average density of container trucks in the ith link segment of the mth link at the kth time step, v m,i (k) represents the average speed of the container truck in the i-th link segment of the m-th link at the k-th time step, fr m,i (k) represents the average flow of container trucks in the ith link segment of the mth link at the kth time step, ρ s,m represents the standard density of container trucks on the mth link, L m Indicates the length of the mth link, LN m represents the number of lanes in the mth link, ξ represents the speed response time constant, κ represents the expected density parameter, and η represents the spatial expected parameter.

[0153] In one example, the path prediction optimization module 13 is used to obtain the real-time flow of container trucks entering the link in the transportation area based on the average flow of container trucks when the container trucks are located at any link, the maximum flow in the link, and the inflow demand of container trucks; obtain the length of the container truck queue corresponding to the real-time flow according to the real-time flow and queue length evolution function; and predict the state variables corresponding to the length of the container truck queue based on the length of the container truck queue, the average density of the container trucks, the average speed of the container trucks, and the average flow of the container trucks.

[0154] In an example, the expressions of the target path optimization function and the time cost function are:

[0155]

[0156] x(k+1)=f(x(k),u(k c ),d(k))

[0157]

[0158] Where P(k) represents the target path optimization function, θ[] represents the penalty function, x(k) represents the state variable of the container truck in the prediction period corresponding to the kth time step, and K represents the time series variable. represents the performance index function, u(k) represents the control variable corresponding to the prediction period of the kth time step, d(k) represents the system disturbance term corresponding to the prediction period of the kth time step, u(k c ) represents the kthc The control variable of the control cycle, f() represents the state transfer function, TTS(k) represents the time cost function, T represents the discrete time step, K represents the time series variable, ρ m,i (k) represents the average density of container trucks in the ith link segment of the mth link at the kth time step, L m Indicates the length of the mth link, LN m represents the number of lanes in the mth link, M represents a constant, and w o (k) represents the queue length of container trucks waiting to perform transportation tasks at the kth time step, N p Represents the model prediction time.

[0159] In one example, the path prediction optimization module 13 is used to set the maximum number of iterations, the initial control variables and the step size parameters, and obtain the state variables through forward recursive calculation, and obtain the co-state variables through reverse recursive calculation; calculate the gradient of the target path optimization function with respect to the control variables based on the state variables and the co-state variables; dynamically adjust the step size parameters based on the current gradient and the gradient of the previous iteration to obtain the updated step size parameters, and calculate the updated control variables after iteration based on the control variable iteration function, the updated step size parameters and the current gradient; if the number of iterations reaches the preset maximum number of iterations, stop the iteration and output the optimal control variables.

[0160] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A container truck path planning method based on Beidou positioning. It is characterized in that The method comprises: Collecting geospatial data, dynamic operation data and container truck transportation data, wherein the container truck transportation data includes real-time location information and real-time status information of container trucks; Based on the geospatial data, construct a real-scene three-dimensional model corresponding to the geospatial data; Extracting road network information from the real-scene three-dimensional model to identify each transport node in the transport area and a link formed between any two adjacent transport nodes; Constructing a dynamic traffic allocation model according to the real-time location information and the real-time status information; Inputting the average density of container trucks, the average speed of container trucks and the average flow volume of container trucks in each link into the dynamic traffic assignment model to predict the state variables of all container trucks in the transportation area of ​​the container factory; Based on the state variables and the dynamic operation data, an optimal transportation path is generated for each container truck in combination with a target path optimization function and a time cost function.

2. The method according to claim 1, It is characterized in that The constructing of a real-scene three-dimensional model corresponding to the geospatial data based on the geospatial data specifically includes: Preprocessing the satellite images, point cloud data, vector map data and digital elevation model data in the geospatial data respectively to obtain satellite optimized images, point cloud optimized data, vector map optimized data and digital elevation model optimized data; Based on the digital elevation model optimization data and the triangulation algorithm, an initial terrain model is constructed, and the orthophoto in the satellite optimization image is mapped onto the initial terrain model as a terrain texture to obtain a three-dimensional terrain model; Extracting building outlines and building height information respectively according to the point cloud optimization data and the vector map optimization data to generate an initial building model, and mapping the building image in the satellite optimization image as a building texture to the initial building model to obtain a three-dimensional building model; The three-dimensional terrain model and the three-dimensional building model are integrated to construct the real-scene three-dimensional model.

3. The method according to claim 2, It is characterized in that The step of integrating the three-dimensional terrain model and the three-dimensional building model to construct a real-scene three-dimensional model specifically includes: Converting the coordinate systems of the three-dimensional terrain model and the three-dimensional building model to unify the coordinate systems of the three-dimensional terrain model and the three-dimensional building model; Positioning the three-dimensional building model at a corresponding position on the three-dimensional terrain model according to the geographical location information of the three-dimensional building model; Performing terrain trimming at the corresponding building position on the three-dimensional terrain model, and unifying rendering parameters of the three-dimensional terrain model and the three-dimensional building model; The processed three-dimensional terrain model and the three-dimensional building model are integrated into a preset data structure to construct the real-scene three-dimensional model.

4. The method according to claim 2, It is characterized in that The satellite images, point cloud data, vector map data and digital elevation model data in the geospatial data are preprocessed respectively, specifically including: Performing geometric correction and radiation correction on the satellite image to obtain the satellite optimized image; The point cloud data are filtered and classified in sequence to obtain point cloud optimized data; Performing topology check and attribute completion on the vector map data respectively to obtain the vector map optimization data; The digital elevation model data is interpolated and smoothed in sequence to obtain the digital elevation model optimized data.

5. The method according to claim 1, It is characterized in that After constructing the real-scene three-dimensional model corresponding to the geospatial data based on the geospatial data, the method further includes: rasterizing the real-scene 3D model so as to convert the real-scene 3D model into a 2.5D raster map having location information of container distribution and arrangement and ground elevation value information; On the 2.5D grid map, grid cells with excessive height are marked as obstacle cells, and according to the operation requirements of the current transportation operation task, the transportation area and the unloading area of ​​the current operation task are divided on the 2.5D grid map.

6. The method according to claim 1, It is characterized in that The expressions for the average density, average speed and average flow rate of container trucks are: fr m,i (k)=ρ m,i (k)·v m,i (k)·LN m Where T represents the discrete time step, K represents the time series variable, and ρ m,i (k) represents the average density of container trucks in the ith link segment of the mth link at the kth time step, v m,i (k) represents the average speed of the container truck in the i-th link segment of the m-th link at the k-th time step, fr m,i (k) represents the average flow of container trucks in the ith link segment of the mth link at the kth time step, ρ s,m represents the standard density of container trucks on the mth link, L m Indicates the length of the mth link, LN m represents the number of lanes of the mth link, ξ represents the speed response time constant, κ represents the expected density parameter, and η represents the spatial expected parameter.

7. The method according to claim 1, It is characterized in that The state variables of all container trucks in the transport area of ​​the container factory area are predicted to include: Based on the average flow of container trucks when they are located at any link, the maximum flow in the link, and the inflow demand of container trucks, the real-time flow of the container trucks entering the link in the transport area is obtained; According to the real-time flow volume and the queue length evolution function, obtaining the length of the container truck queue corresponding to the real-time flow volume; Based on the length of the container truck queue, the average density of the container trucks, the average speed of the container trucks and the average flow rate of the container trucks, a state variable corresponding to the length of the container truck queue is predicted.

8. The method according to claim 1, It is characterized in that The expressions of the target path optimization function and the time cost function are respectively: x(k+1)=f(x(k),u(k c ),d(k)) Where P(k) represents the target path optimization function, θ[] represents the penalty function, x(k) represents the state variable of the container truck in the prediction period corresponding to the kth time step, and K represents the time series variable. represents the performance index function, u(k) represents the control variable corresponding to the prediction period of the kth time step, d(k) represents the system disturbance term corresponding to the prediction period of the kth time step, u(k c ) represents the kth c The control variable of the control cycle, f() represents the state transfer function, TTS(k) represents the time cost function, T represents the discrete time step, K represents the time series variable, ρ m,i (k) represents the average density of container trucks in the ith link segment of the mth link at the kth time step, L m Indicates the length of the mth link, LN m represents the number of lanes in the mth link, M represents a constant, and w o (k) represents the queue length of container trucks waiting to perform transportation tasks at the kth time step, N p Represents the model prediction time.

9. The method according to claim 1, It is characterized in that The method further comprises: Set the maximum number of iterations, initial control variables, and step size parameters, and obtain state variables through forward recursive calculations and co-state variables through backward recursive calculations; Calculating the gradient of the target path optimization function with respect to the control variable according to the state variable and the co-state variable; Based on the current gradient and the gradient of the previous iteration, dynamically adjust the step size parameter to obtain the updated step size parameter, and calculate the updated control variable after iteration according to the control variable iteration function, the updated step size parameter and the current gradient; If the number of iterations reaches the preset maximum number of iterations, the iteration is stopped and the optimal control variable is output.

10. A container truck path planning system based on Beidou positioning, It is characterized in that The container factory area container truck path planning system (1) comprises a data collection management module (11), a model building optimization module (12) and a path prediction optimization module (13), wherein: The data collection management module (11) is used to collect geographic space data, dynamic operation data and container truck transportation data, wherein the container truck transportation data includes real-time location information and real-time status information of container trucks; The model building optimization module (12) is used to build a real-scene three-dimensional model corresponding to the geospatial data based on the geospatial data, extract road network information from the real-scene three-dimensional model to identify each transport node in the transport area and the link formed between any two adjacent transport nodes, and build a dynamic traffic allocation model based on the real-time location information and the real-time status information; The path prediction optimization module (13) is used to input the average density of container trucks, the average speed of container trucks and the average flow of container trucks in each link into the dynamic traffic allocation model to predict the state variables of all container trucks in the transportation area of ​​the container factory, and generate the optimal transportation path for each container truck based on the state variables and the dynamic operation data in combination with the target path optimization function and the time cost function.

Citation Information

Patent Citations

  • Dynamic optimal distribution method for logistics tasks based on distribution vehicle end real-time state information drive

    CN103473659A

  • Three-dimensional digital sandbox system

    CN107945270A

  • Intelligent port dispatching command system and method for coping with container trucks

    CN116092285A

  • Real scene three-dimensional modeling method and device based on remote sensing satellite image and medium

    CN116468869A

  • Logistics freight vehicle transportation supervision system

    CN118115071A

Cited By

  • Multi-terrain comprehensive path planning method fusing road factors

    CN120702466A

  • Method for optimizing semiconductor wafer conveying efficiency by using equipment end cache system

    CN120893609A

  • A method for optimizing semiconductor wafer delivery efficiency using a device-side caching system

    CN120893609B

  • Wavelength division imaging device and image registration following method

    CN121309941A

  • Driving path planning method for unmanned vehicle in airport

    CN122531242A