Full-automatic loading control method for loading platform

Abstract

The application discloses a full-automatic material discharging control method for a loading building, and relates to the technical field of automatic loading control.The method comprises the following steps: step S1, collecting real-time data and positioning a carriage; step S2, performing three-dimensional real-time modeling and state analysis; step S3, dividing a loading unit and generating a loading strategy; step S4, starting loading and performing self-adaptive material discharging control; step S5, deviation identification and dynamic adjustment; and step S6, recording and generating a report.Through the cooperation of a laser radar, visual AI and a PLC, the full-process automation from vehicle number identification, carriage alignment, material discharging control to quality detection is realized, and the operation risk caused by human judgment errors is reduced.

Description

Technical Field

[0001] This invention relates to the field of automated loading control technology, and in particular to a fully automated material feeding control method for loading towers. Background Technology

[0002] Loading bulk cargo onto trains is a crucial link in the logistics chain of commodities such as coal and ore. Traditional loading operations rely heavily on manual labor, requiring operators to monitor screens in a central control room and manually control chute raising and lowering, gate opening and closing, and train traction based on experience. This model suffers from problems such as high labor intensity, susceptibility to spillage and uneven loading, low loading accuracy, low operational efficiency, and safety hazards from chute collisions with wagons. Furthermore, data from each stage is isolated, making precise management difficult.

[0003] In recent years, although some automated loading technologies have emerged, such as using sensors for material level detection or simple program control, they often struggle to achieve high-precision, adaptive, and uniform material feeding when faced with complex and variable on-site conditions, such as different vehicle models, different coal densities, and fluctuations in vehicle speed. Furthermore, existing technologies lack real-time sensing capabilities and automatic correction capabilities for the loading process, resulting in unstable loading quality and requiring significant manual intervention to address overloading and off-center loading issues. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a fully automated material feeding control method for loading towers.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A fully automated material unloading control method for a loading tower, the method comprising the following steps: Step S1: Perform real-time data acquisition and carriage positioning.

[0006] By deploying multi-line lidar, high-definition network cameras and speed measuring radar at key locations in the loading building, multi-dimensional data of the moving train carriages are collected in real time, including the precise position coordinates of the carriages, real-time speed, direction and three-dimensional point cloud data of the carriage outline. Step S2: Perform 3D real-time modeling and state analysis.

[0007] The multi-source point cloud data collected in step S1 is fused with video images, and real-time 3D modeling is performed through an intelligent detection algorithm engine to generate a dynamic 3D scene including an empty car model and a real-time material accumulation model; key status information such as car interval, car edge, relative position of chute, material drop height and material surface flatness is analyzed and obtained. Step S3: Divide the loading units and generate the loading strategy.

[0008] Based on the car model and loading plan, each car is virtually divided into n consecutive loading units along its direction of travel; based on the target load, material density and the three-dimensional model generated in step S2, the theoretical loading volume and weight required for each loading unit are calculated to form an initial loading prediction model and an optimal control strategy library. Step S4: Start loading and implement adaptive material feeding control.

[0009] The automatic loading process is initiated, and the electric car-pulling machine is controlled to pull the car body through the loading chute at a constant speed. During the material unloading process, the opening and closing timing of the quantitative bin gate and the lifting height of the chute are controlled according to the strategy in step S3. At the same time, steps S1 and S2 are continuously executed to obtain the latest three-dimensional model of the material level, and the actual loading model is compared with the theoretical prediction model in real time. Step S5: Deviation Identification and Dynamic Adjustment

[0010] Based on the real-time comparison results of step S4, the system intelligently identifies whether there are deviation trends such as overloading, unbalanced loading, short tonnage, or material level exceeding the limit during the loading process. If a deviation is detected, the system immediately and dynamically adjusts the control parameters, including but not limited to fine-tuning the chute height, correcting the gate opening and closing sequence, adjusting the feed flow rate, or adjusting the traction vehicle speed, in order to eliminate or reduce the deviation until the car is fully loaded. Step S6: Record and generate a report

[0011] After a single car is loaded, the system automatically records the final three-dimensional material level model, actual loading weight, off-center load data, and all control parameter adjustment records for that car, generating a loading quality report. This data is fed back to the AI ​​learning model to continuously optimize and train the prediction model and control strategy for subsequent loading operations.

[0012] Preferably, during real-time 3D modeling in step S2, the following operations are performed based on the intelligent detection algorithm engine: Register and fuse local point cloud data from two or more lidar scanners to synthesize a complete 3D point cloud data map of the carriage and cargo. A three-dimensional mesh model is established based on the three-dimensional point cloud data and input into a convolutional neural network model to calculate the real-time volume of the goods inside the carriage. The average density of the cargo is calculated in real time based on the cumulative weight of the loading machine and the volume of the cargo. By fitting the point cloud coordinates of the cargo surface, a cargo distribution curve is generated to analyze the uniformity of material accumulation.

[0013] Preferably, the method for intelligently identifying whether a deviation trend occurs during the loading process in step S5 includes: The carriage is divided into horizontal and vertical sections, and the three-dimensional model of the cargo is divided into multiple blocks; Calculate the material volume of each block and convert it into weight distribution by combining it with the average density calculated in real time; Set the left and right deviation height thresholds for the front and rear axles, and the front and rear deviation height thresholds; The actual height of each block is compared with the expected height calculated based on the theoretical uniform model. If the deviation of any block exceeds the set threshold, it is determined to be an off-center load, and an alarm and automatic adjustment program are triggered.

[0014] Preferably, the method further includes a coordinated control step with the upstream material supply system, as follows: Communicate with the stacker-reclaimer and belt conveyor control systems to obtain real-time material reclaiming location, belt scale flow rate, and material conveying information; Based on the loading progress, the traction speed of the wagon, and the material level in the buffer silo, the system dynamically calculates and sends suggested material handling flow and tonnage upstream to prevent the buffer silo from overflowing or running out of material.

[0015] Preferably, the method combines lidar positioning, anti-collision sensors, and photoelectric switches during material feeding to monitor the distance between the chute and the carriage baffle in parallel. If any sensor detects a collision risk, the chute is raised to a safe height.

[0016] Preferably, the method monitors the remaining material through a quantitative bin weighing instrument during material discharge, calculates the remaining capacity based on the position of the carriage, and makes a real-time judgment to close the gate in advance to prevent material spillage.

[0017] Preferably, in the method, the vehicle number recognition adopts a dual-system verification of visual AI and RFID, and the vehicle compartment positioning adopts multi-radar cross-detection.

[0018] A fully automated material unloading control system for a loading tower, used to implement the method described above, the system comprising: Centralized control unit: With PLC controller as the core, it connects to and controls all key equipment on site, including hydraulic station, chute, gate, quantitative bin instrument, electric trolley machine and various sensors, and executes control instructions from AI analysis unit; AI perception and analysis unit: includes multi-line high-precision lidar, high-definition network camera, speed radar and AI analysis server deployed in loading channel; the server runs intelligent detection algorithm and is responsible for real-time 3D modeling, status analysis, deviation identification and strategy calculation. Vehicle traction unit: including electric toggle machine and its control system, receiving instructions from the centralized control unit to realize automatic traction, alignment, speed adjustment and return in sync with the loading rhythm; Data interaction interface unit: Through industrial communication protocols and middleware, it realizes data interaction with the port production management system, belt conveyor process control system and vehicle number recognition system to obtain loading plans, material information and vehicle information; Human-machine interaction monitoring unit: including host computer, control panel and monitoring screen, providing full-process visual monitoring, parameter setting, one-click switching between manual / automatic mode, alarm prompts and report generation functions.

[0019] Preferably, in the AI ​​perception and analysis unit, the installation and deployment of the lidar meets the following conditions: at least two lidars are deployed above each loading lane, with an installation spacing of 5-7 meters and a distance of 2-3 meters from the upper surface of the goods.

[0020] Preferably, the system also includes an intelligent inspection subsystem for empty vehicle status. This subsystem deploys a line array camera and a vehicle number recognition camera in front of the loading area to automatically identify safety anomalies such as unclosed doors and foreign objects in the vehicle compartment, and generates an inspection report.

[0021] The beneficial effects of this invention are as follows: 1. This invention achieves full automation of the entire process from vehicle number recognition, carriage alignment, material feeding control to quality inspection through the collaboration of LiDAR, visual AI and PLC, reducing the operational risks caused by human error.

[0022] 2. By integrating quantitative bin weighing data with real-time 3D point cloud modeling, the system can dynamically calculate and adjust the material feeding amount, ensuring that the loading weight of each car is stably controlled within the target deviation range, effectively solving the problems of overloading and underloading.

[0023] 3. The unit-based control scheme of this invention, combined with real-time three-dimensional scanning, enables the system to detect the material accumulation pattern in real time during the loading process. It identifies the off-center loading trend through AI algorithms and dynamically adjusts the chute position or feeding strategy to correct the off-center loading. This ensures the uniformity of loading from the source and avoids transportation safety risks and subsequent coal leveling costs caused by off-center loading.

[0024] 4. The system of the present invention is linked with the electric trolley to realize automatic matching of traction speed and loading rhythm; it communicates with the upstream stacker-reclaimer and belt conveyor system to realize accurate prediction and adjustment of material supply flow, avoiding buffer bin overflow or material shortage waiting, and shortening the overall operation time of a single train. Attached Figure Description

[0025] Figure 1 This is a system architecture diagram of the present invention; Figure 2 This is a train carriage image captured for the present invention; Figure 3 This is an automatic annotation diagram of the problematic carriages in this invention; Figure 4 This is a laser radar scanning model diagram of the present invention; Figure 5 This is a three-dimensional point cloud data diagram of the present invention. Detailed Implementation

[0026] The technical solution of the present invention will be further described in detail below with reference to specific embodiments.

[0027] Example 1: A fully automated material unloading control method for a loading tower. This method is applied to the intelligent transformation of the dual-track train loading tower at Guotou Qinzhou Port. The specific steps are as follows: Step S101: Multi-dimensional data synchronous acquisition; When the train is pulled into the loading area by the electric towing machine, the lidar installed on the pillars on both sides and the top of the loading tower begins to scan synchronously, acquiring dense point clouds of the carriage outline and the area below the chute. Simultaneously, high-definition network cameras capture video streams, and millimeter-wave radar accurately measures the train's real-time speed. All data is uploaded to the AI ​​analysis server in real time via industrial Ethernet.

[0028] Step S102: Dynamic 3D Scene Reconstruction; The AI ​​server calls the built-in Matrix-LoadDetector algorithm engine to perform timestamp alignment, coordinate registration, and data fusion on point clouds from different radars, generating a complete point cloud model of a C70 type carriage. Simultaneously, the algorithm segments features such as carriage panels and couplers from the video stream to assist in semantic enhancement of the point cloud model. Before loading, an empty 3D bottom model of the carriage is first established.

[0029] Step S103: Intelligent Division and Strategy Calculation; Based on the standard length of the C70 car, the system automatically divides it into 10 equally long virtual loading units along the direction of travel. The operator inputs the target load of this car section as 70 tons and the material as bituminous coal with a calorific value of 5500 kcal into the host computer. The system matches the reference density of this coal type to the historical database and, combined with the car's floor area, calculates the theoretical loading weight of each loading unit as 7 tons and the theoretical stockpile height as H0. Simultaneously, based on the car's height, the system sets the chute collision avoidance warning height and emergency lifting height.

[0030] Step S104: Execute closed-loop adaptive feeding; the electric traction machine pulls the car forward at a constant speed of 0.8 km / h. When the front end of the car reaches directly below the chute, the PLC controls the quantitative bin gate to open, and at the same time controls the chute to descend to the preset initial height (below the upper beam of the car); the material begins to fall.

[0031] Step S105: Real-time comparison and dynamic correction; During the loading process, the LiDAR continuously scans, and the AI ​​server generates the latest 3D model of the material level every 0.1 seconds. The system performs the following analyses in real time: 1) Compare the current total unloading weight (from the quantitative silo scale) with the theoretical total weight to prevent overloading or underloading; 2) Analyze the latest 3D model to calculate whether the material accumulation center deviates from the physical center of the carriage. For example, when loading reaches the 5th unit, the AI ​​model analysis finds that the material surface curve is slightly lower at the front and slightly higher on the right, indicating a slight off-center loading trend. The decision engine immediately generates adjustment instructions: during the subsequent 6th and 7th units of unloading, slightly reduce the descent of the left side of the chute and fine-tune the gate opening to increase the flow rate on the left side. These instructions are sent to the PLC for execution in real time.

[0032] Step S106: Data recording and model optimization; after loading of this car section is completed, the gate closes and the chute is lifted. The system records the final 3D model of the material level; the actual loading weight is 70.1 tons (deviation +0.1 tons), and the off-center loading rate is less than 0.5%. All process data, including adjustment commands, sensor data, and final results, are stored in the historical database. Every week, the system uses the accumulated new data to perform offline reinforcement training on the Matrix-LoadDetector model, making it more adaptable to the actual coal characteristics and conditions of Qinzhou Port.

[0033] Example 2: This example provides a system for implementing the above method; the system is deployed at the loading terminal of Qinzhou Port as follows: Centralized Control Unit: A PLC-based control cabinet is installed in the control room of the loading tower. This PLC directly controls four sets of hydraulic gates, two telescopic chutes, and four quantitative bin weighing instruments on the two loading lines via remote I / O modules and a Profinet network. Simultaneously, it communicates with the PLCs of the two electric car-turning machines via a DP bus.

[0034] AI Sensing and Analysis Unit: XT32 LiDAR sensors are installed on both sides of each loading line, 3.5 meters from the center of the track and 6 meters high. High-definition PTZ cameras are installed behind the chute to specifically monitor the material drop point. All sensing devices are connected to an AI server deployed with the Matrix-LoadDetector algorithm. This server not only performs real-time point cloud processing but also runs a vehicle number recognition model, comparing the results with the existing RFID system to ensure a vehicle number recognition rate of over 99.5%.

[0035] Coordinated feeding interface: The data interaction unit of the system connects with the existing port coal process control system (CPS) through the OPC UA protocol. During the loading process, this system real-timely receives the belt scale flow data from the CPS, dynamically calculates the recommended reclaimer quantity of the upstream reclaimer in combination with the radar level gauge data of the buffer bin, and sends it through the interface, so as to realize the linkage of loading and reclaiming, and meet the technical requirement of controlling the residual material in the buffer bin below 3 tons.

[0036] With the above method and system, the present invention transforms the original loading operation that relies on manual experience into a standardized process of data-driven, intelligent decision-making and automatic execution. Verified by practice, the system can stably achieve a loading deviation within ±0.2 tons for each vehicle (with a load of more than 60 tons) and an unbalanced loading rate of less than 1%, effectively eliminating material scattering and collision accidents, and truly realizing the intelligent upgrading of the loading station.

[0037] The above description only illustrates the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any equivalent replacement or modification made by those skilled in the art within the technical scope disclosed by the present invention based on the technical solution and the inventive concept of the present invention shall be covered within the protection scope of the present invention.

Claims

1. A fully automated material unloading control method for a loading tower, characterized in that, The method includes the following steps: Step S1: Perform real-time data acquisition and carriage positioning. By deploying multi-line lidar, high-definition network cameras and speed measuring radar in the loading building, multi-dimensional data of the moving train carriages are collected in real time, including the precise position coordinates of the carriages, real-time speed, direction and three-dimensional point cloud data of the carriage outline. Step S2: Perform 3D real-time modeling and state analysis. The multi-source point cloud data collected in step S1 is fused with video images to perform real-time 3D modeling and generate a dynamic 3D scene including an empty car body model and a real-time material accumulation model; the information on car body spacing, car body edge, relative position of chute, material drop height and material surface flatness is analyzed and obtained. Step S3: Divide the loading units and generate the loading strategy. Based on the car model and loading plan, each car is virtually divided into n consecutive loading units along its direction of travel; based on the target load, material density and the three-dimensional model generated in step S2, the theoretical loading volume and weight required for each loading unit are calculated to form an initial loading prediction model and an optimal control strategy library. Step S4: Start loading and implement adaptive material feeding control. The automatic loading process is initiated, and the electric car-pulling machine is controlled to pull the car body through the loading chute at a constant speed. During the material unloading process, the opening and closing timing of the quantitative bin gate and the lifting height of the chute are controlled according to the strategy in step S3. At the same time, steps S1 and S2 are continuously executed to obtain the latest three-dimensional model of the material level, and the actual loading model is compared with the theoretical prediction model in real time. Step S5: Deviation Identification and Dynamic Adjustment Based on the real-time comparison results of step S4, the system intelligently identifies whether a deviation trend occurs during the loading process; if a deviation is detected, the control parameters are immediately and dynamically adjusted until the loading of the car is completed.

2. The fully automatic material feeding control method for a loading tower according to claim 1, characterized in that, During real-time 3D modeling in step S2, the following operations are performed based on the intelligent detection algorithm engine: Register and fuse local point cloud data from two or more lidar scanners to synthesize a complete 3D point cloud data map of the carriage and cargo. A three-dimensional mesh model is established based on the three-dimensional point cloud data and input into a convolutional neural network model to calculate the real-time volume of the goods inside the carriage. The average density of the cargo is calculated in real time based on the cumulative weight of the loading machine and the volume of the cargo. By fitting the point cloud coordinates of the cargo surface, a cargo distribution curve is generated to analyze the uniformity of material accumulation.

3. The fully automated material unloading control method for a loading tower according to claim 1, characterized in that, The specific method for intelligently identifying whether a deviation trend occurs during the loading process in step S5 includes: The carriage is divided into horizontal and vertical sections, and the three-dimensional model of the cargo is divided into multiple blocks; Calculate the material volume of each block and convert it into weight distribution by combining it with the average density calculated in real time; Set the left and right deviation height thresholds for the front and rear axles, and the front and rear deviation height thresholds; The actual height of each block is compared with the expected height calculated based on the theoretical uniform model. If the deviation of any block exceeds the set threshold, it is determined to be an off-center load, and an alarm and automatic adjustment program are triggered.

4. The fully automatic material feeding control method for a loading tower according to claim 1, characterized in that, The method also includes a coordinated control step with the upstream material supply system, as detailed below: Communicate with the stacker-reclaimer and belt conveyor control systems to obtain real-time material reclaiming location, belt scale flow rate, and material conveying information; Based on the loading progress, the traction speed of the wagon, and the material level in the buffer bin, the system dynamically calculates and sends suggested material handling flow rate and tonnage upstream.

5. The fully automatic material feeding control method for a loading tower according to claim 1, characterized in that, During material discharge, the method combines lidar positioning, anti-collision sensors, and photoelectric switches to monitor the distance between the chute and the carriage baffle in parallel. If any sensor detects a collision risk, the chute is raised to a safe height.

6. The fully automatic material feeding control method for a loading tower according to claim 5, characterized in that, The method monitors the remaining material through a quantitative bin weighing instrument during material discharge, calculates the remaining capacity based on the position of the carriage, and makes a real-time judgment to close the gate in advance to prevent material spillage.

7. The fully automatic material feeding control method for a loading tower according to claim 1, characterized in that, In the method described, vehicle number recognition uses a dual-system verification of visual AI and RFID, and vehicle compartment positioning uses multi-radar cross-detection.

8. A fully automatic material unloading control system for a loading tower, used to implement the method described in any one of claims 1-7, characterized in that, The system includes: Centralized control unit: With PLC controller as the core, it connects to and controls field devices and executes control instructions from AI analysis unit; AI perception and analysis unit: includes multi-line high-precision lidar, high-definition network camera, speed radar and AI analysis server deployed in loading channel; the server runs intelligent detection algorithm and is responsible for real-time 3D modeling, status analysis, deviation identification and strategy calculation. Vehicle traction unit: including electric toggle machine and its control system, receiving instructions from the centralized control unit to realize automatic traction, alignment, speed adjustment and return in sync with the loading rhythm; Data interaction interface unit: Through industrial communication protocols and middleware, it realizes data interaction with the port production management system, belt conveyor process control system and vehicle number recognition system to obtain loading plans, material information and vehicle information.

9. The system according to claim 8, characterized in that, The system also includes a human-machine interaction monitoring unit, which includes a host computer, an operating console, and a monitoring screen. It provides full-process visual monitoring, parameter setting, one-click switching between manual and automatic modes, alarm prompts, and report generation functions.

10. The system according to claim 8, characterized in that, The system also includes an intelligent inspection subsystem for empty vehicles. This subsystem deploys a line array camera and a vehicle number recognition camera in front of the loading area to automatically identify safety anomalies such as unclosed doors and foreign objects in the cargo compartment, and generates an inspection report.

Images