System and method for intelligent inspection of railway freight train

Abstract

A railway freight station is not allowed to be connected with an external network, auxiliary equipment cannot be laid, a 4G / 5G communication network cannot be used, inspection routes are complex, and various carriage types and inspection content types exist. Although many types of inspection equipment and systems exist in the market at present, the inspection equipment and the systems are not suitable for railway freight inspection in view of the limitation of environmental conditions of railway freight stations. The invention designs a novel inspection scheme method, and the scheme uses high and new technologies such as a software technology, an intelligent navigation technology, a robot technology, a video monitoring technology, an artificial intelligence technology and long-distance wireless transmission. An integrated solution method is designed from different perspectives of inspection equipment design, multi-navigation mixing, an artificial intelligence model and the like, the limitation that an existing inspection system cannot meet railway freight station inspection conditions is solved, and intelligent inspection landing implementation of the railway freight station freight cars is achieved.

Description

Technical Field

[0001] This invention belongs to the fields of inspection robots and artificial intelligence. Background Technology

[0002] After a railway freight train stops at a freight station, goods need to be loaded and unloaded. According to railway freight station requirements, the train carriages must be inspected before and after loading and unloading to ensure the safety of the goods and the train's operation. Currently, the inspection is mainly conducted manually, with inspectors using cameras to check each carriage. If any abnormalities are found, photos are taken and corrective action is arranged. The inspection process presents challenges due to the large number of inspection items, the heavy daily workload, and the high risks associated with inspections in inclement weather. While there are many types of inspection robots on the market, none are suitable for railway freight inspection due to the environmental limitations of railway freight stations. Railway freight stations have specific restrictions on their operating environment due to safety regulations and other requirements, including: prohibition of connecting inspection systems to external networks, prohibition of installing navigation aids along inspection routes, prohibition of using 4G / 5G communication networks, varying distances and elevations between inspection routes and freight cars, and different elevations for different types of freight cars. These restrictions hinder the promotion and use of current inspection vehicles and systems in railway freight stations. Currently, no inspection equipment or system on the market can meet the limitations of railway freight inspection. Summary of the Invention

[0003] To address the limitations of existing inspection systems and equipment for railway freight car inspection, this invention proposes a novel inspection scheme and method to meet the constraints of actual inspection operations and achieve intelligent inspection. The scheme employs intelligent technologies such as software, intelligent navigation, robotics, image recognition, video surveillance, OCR text recognition, and long-distance wireless transmission to achieve intelligent freight car inspection while meeting the requirements of railway freight station environments. This invention is a smart inspection system and method specifically designed for railway freight car inspection tasks, supporting safety inspection of various types of freight vehicles, including open wagons, covered wagons, flatbed wagons, container wagons, and tank cars. It can detect various vehicle anomalies in real time and identify various carriage information. To achieve the aforementioned intelligent inspection objectives, the solution utilizes various sensor devices and information acquisition equipment, including high-resolution cameras, 3D LiDAR, millimeter-wave radar, and depth cameras. It also employs multiple navigation technologies such as RTK and LiDAR for precise positioning and navigation. High-performance network bridge equipment and data encryption technology ensure the transmission and security of inspection data and management information. Artificial intelligence (AI) technology is used for image recognition and text detection of streaming media data, with GPU graphics servers completing the detection and recognition tasks in real time. Finally, a software-based inspection dispatch system is developed to coordinate with inspectors in dispatching inspection vehicles to complete inspection operations. Currently, outdoor inspection vehicles on the market generally have chassis weighing no more than 150 kg; the lifting equipment typically has a lifting height of no more than 2 meters, and the lateral extension of the top of the lifting equipment is no more than 1 meter. This type of equipment structure is completely unsuitable for inspecting railway freight cars. The maximum height of a railway freight train car from the ground is 4.9 meters, and the maximum distance between the inspection route and the freight car can reach 3.5 meters. Therefore, to ensure that the roof can be inspected, the onboard lifting equipment must be raised to at least 5 meters, and the lateral extension distance must be at least 3.6 meters. Railway freight stations typically cover an area of ​​2-3 kilometers in radius, with inspection routes generally exceeding 1 kilometer in length. Due to safety requirements, equipment at freight stations is not permitted to connect to the external network, use 4G / 5G mobile networks for data transmission, or lay data transmission auxiliary equipment (such as wireless access points) or navigation auxiliary equipment (such as magnetic nails) along the inspection routes. This necessitates high standards for long-distance data transmission security and navigation methods. Currently, inspection vehicles on the market generally use 4G / 5G wireless transmission for data transmission, and navigation methods such as visual navigation, laser navigation, magnetic navigation, and GPS navigation cannot fully meet the navigation needs of inspections. Given the aforementioned limitations, a new solution and method have been invented to meet these limitations and achieve the goal of intelligent inspection. The new solution and method mainly comprises four parts: an inspection vehicle chassis and superstructure, an intelligent navigation system and equipment, an AI server and inspection management system, and relay poles and network transmission equipment. (I) The design specifications for the inspection vehicle chassis and superstructure are as follows ( Figure 1 ): 1) Overall weight: 330 kg - 380 kg; 2) Chassis structure: Ackermann steering, front-to-rear-wheel drive, solid tires. This primarily addresses issues such as excessive load, uneven road surfaces, and the risk of tire blowouts on gravel roads. 3) The chassis design dimensions of the inspection vehicle are: length * width * height: 1.5 meters * 1 meter * 0.5 meters, with a ground clearance of 10-12 centimeters. If the size is too large, it will be difficult to turn around at railway inspection intersections; if the size is too small, it will be easy to overturn (because the lifting boom is too high and the extension distance is too long). The low ground clearance is to lower the center of gravity and prevent the inspection vehicle from overturning. 4) Drive motor type: stepper motor or servo motor. Stepper motors require the addition of an odometer encoder and angle converter to support navigation; 5) Lifting mast: also known as the main mast. Hydraulic lifting control, 3 sections, 2.2 meters when fully lowered, and 5.5 meters when fully raised. The lifting height is precisely controlled using a pull rope sensor. 6) Lifting side pole: DC motor lead screw lifting, lifting height is 1 meter-2.5 meters, to support wide-angle / fisheye cameras and fill lights; 7) Telescopic crossbar: DC motor lead screw telescopic, 2 sections, 2 meters when fully retracted, 3.6 meters when fully extended; supports wide-angle / fisheye cameras and fill lights; 8) Crossbar Rotation Device: The rotation device supports 360-degree rotation and automatically rotates to a preset angle for shooting according to the actual shooting needs on the inspection route; 9) Image acquisition equipment: wide-angle / fisheye camera and fill light. Floodlights are used to meet the lighting requirements for shooting; 10) Mobile Station Bridge: A 180-degree bidirectional bridge installed on top of the main lifting mast. When the lifting mast is raised, the bridge also rises, its height exceeding that of the truck bed, thus ensuring uninterrupted data signal transmission. The bridge supports proprietary protocols and encrypted data transmission to guarantee data security. (II) Intelligent navigation system and equipment for patrol vehicles: 1) LiDAR: Mounted on the chassis, 0.45 meters from the center of the front of the inspection vehicle, and 0.43 meters above the vehicle's surface. It uses a 16-line 3D LiDAR, connected to the navigation controller, responsible for collecting data to support positioning, navigation, and obstacle avoidance. 2) Depth Camera: An obstacle avoidance information acquisition device, installed in the center of the front plane of the inspection vehicle, to collect information on forward obstacles at medium to long distances. The depth camera uses an optical camera and works in conjunction with LiDAR in different environments; 3) Millimeter-wave radar: Installed in the middle of the front and rear panels of the inspection vehicle, it collects data on obstacles in front of and behind the inspection vehicle at medium to long distances. Navigation controller: Installed inside the inspection vehicle, it is the core device for intelligent navigation. The navigation controller receives external signals and directs the inspection vehicle's movement or obstacle avoidance according to a preset route; 4) Mobile GNSS equipment: The GNSS antenna is installed on the left and right sides of the front of the inspection vehicle, parallel to each other, to receive GNSS signals and then send them to the navigation controller through the receiver to support inspection navigation; the receiver is installed inside the inspection vehicle; the GNSS receiver also receives the positioning signal emitted by the transmitter on the high pole, and uses RTK technology to provide high-precision positioning for the navigation system; (III) AI Server Equipment and Inspection Management System: 1) AI Server: A server equipped with a GPU chip is used. The GPU is responsible for executing the AI ​​model and completing image recognition and text analysis tasks. 2) Inspection Management System: This system primarily implements functions such as inspection vehicle dispatching, video monitoring of inspection tasks, real-time alerts for inspection anomalies, retention of historical inspection information, inspection reports, and anomaly management. It is an application system integrating machine vision, image processing, and deep learning. Image feature-based recognition and detection involves feature extraction, feature calculation, parameter analysis and selection, and determination of abnormal event characteristics. Finally, an AI model is trained using big data. The AI ​​server uses this model to analyze the image data collected during the inspection process, quickly obtaining anomaly information and ensuring the efficient and safe operation of the inspection work. The intelligent inspection process is as follows: 2.1) The inspector selects the route to be inspected and an available inspection vehicle to carry out the inspection task; 2.2) The inspection vehicle raises the barrier, extends the barrier, turns on the camera (lights), and navigates along the pre-planned inspection route; 2.3) The inspection vehicle transmits camera video streams to the AI ​​server in real time; 2.4) The AI ​​server receives the video stream and extracts frames of images according to the rules; 2.5) The AI ​​server uses an AI model to detect and identify extracted frames of the image, checking for any anomalies; 2.6) If any abnormality occurs, the system will display a real-time notification to the on-duty personnel on the system interface; 2.7) The AI ​​server monitors the operation and inspection status of the inspection vehicle in real time; 2.8) Save the abnormal results data from the inspection and generate an anomaly report; (iv) Relay poles and network transmission equipment: Relay poles are auxiliary devices for completing long-distance data transmission. They can be placed in open, high places (such as rooftops) to ensure that the data collected by the cameras on the inspection vehicle is transmitted to the AI ​​server. 1) Elevated pole: approximately 2 meters high, with an "X"-shaped support frame at the bottom and a box for placing equipment in the middle; 2) GNSS antenna: mounted on a panel at the top of an elevated pole, it receives positioning satellite signals and transmits the signals to the transmitter; 3) GNSS transmitter: Installed in a box in the middle of the elevated pole. The transmitter receives satellite signals and transmits them to the receiver on the inspection vehicle; 4) Base station bridge: It communicates with the mobile station bridge on the inspection vehicle using a private protocol and data encryption, and connects to the AI ​​server via a network cable to transmit data to the AI ​​server; Compared with existing market inspection systems, the system implemented using the method described in this invention has the following advantages and characteristics: 1) It technically solves the limitation that current market inspection robots are not suitable for railway freight inspection scenarios; 2) Reduced the workload and risks for inspection staff; 3) It solved the requirement of simultaneously performing high-altitude and low-altitude inspection operations; 4) The use of AI technology has improved the efficiency of inspection, detection, and identification; 5) Develop an intelligent inspection and dispatch system using software technology to manage and schedule inspection tasks; 6) Use a hybrid navigation technology combining lidar and GNSS to achieve precise positioning and navigation; 7) Use data encryption technology in network transmission to ensure data transmission security; Attached Figure Description

[0004] Figure 1 It is an inspection vehicle and its superstructure equipment; Figure 2 It is an elevated pole device; Detailed Implementation

[0005] This invention includes an inspection vehicle, a lifting device, a data collection device, a navigation device, a communication device, an AI server, and an inspection system. The specific implementation steps are as follows: 1) Inspection vehicle chassis: Figure 1 The 101 section is a four-wheeled vehicle with an Ackerman chassis structure, rear-wheel drive and front steering; solid tires are used to prevent tire blowouts; the vehicle's weight must be over 300 kg, otherwise it cannot support the lifting equipment on top; the drive motor can be a servo motor or a stepper motor. 2) Main pole lifting equipment: Figure 1 The 105th section employs a hydraulic mast structure, driven by a stepper motor. It is divided into three sections, measuring 2.2 meters when fully lowered and 5.5 meters when fully raised. The base of the main mast is fixed to the inspection vehicle chassis; the top features a rotating device. 3) Side pole lifting equipment: Figure 1 In section 103, the side rod is divided into a fixed part and a sliding part. The fixed part is 2.5 meters high and its bottom is fixed next to the main rod. The sliding part is supported by the fixed part and is driven by a stepper motor to slide up and down to raise and lower the data acquisition device. 4) Horizontal bar telescopic device: Figure 1 In section 107, one end of the crossbar is fixed to a rotating device at the top of the main pole. The crossbar is divided into two sections, 2 meters long when retracted and 3.6 meters long when fully extended; 5) Rotating equipment: Figure 1 In section 106, the supporting crossbar rotates; 6) Data acquisition equipment: Figure 1 The middle 104 section, including the camera and fill light, is fixed to the top of the crossbar and the sliding part of the side bar, and can extend and slide up and down with the crossbar and side bar; 7) 3D laser equipment: Figure 1 In section 102, the laser equipment is responsible for collecting positioning and navigation data; the lidar also provides the function of scanning and detecting obstacles, so its installation height and position should be considered in conjunction with other obstacle avoidance equipment to achieve the best effect. 8) GNSS equipment: Figure 1 Part 109 of the Chinese curriculum. Figure 2 Section 201. GNSS equipment is responsible for collecting positioning and navigation data; a hybrid laser and GNSS navigation system is used to achieve precise positioning and navigation for the inspection vehicle; 9) Obstacle avoidance devices: Figure 1 Sections 102, 110, and 111 include lidar, depth cameras, and millimeter-wave radar. These three types of equipment (10) work together to enable the inspection vehicle to safely avoid obstacles during its journey. 11) Communication equipment: Figure 1 Part 108 Figure 2 Section 202. Wireless bridge, installed at the top of the main pole and the top of the elevated pole of the inspection vehicle; the bridge equipment is raised simultaneously when the main pole of the inspection vehicle is raised; 12) Inspection server: This includes an AI server and a monitor. The AI ​​server uses a GPU to run AI models. 13) Inspection and Dispatch System: A management system developed using software technology. This system mainly performs three types of tasks: 13.1) Dispatch inspection vehicles to perform inspection operations according to task requirements; 13.2) Receive streaming media data from the inspection vehicle and perform image detection and OCR recognition on the data using an AI model; the image detection model is responsible for detecting abnormalities on the sides and top of the truck bed, while the OCR recognition is responsible for identifying data information on the sides of the truck bed. 13.3) Value-added services developed based on inspection data.

Claims

1. The main lifting boom and crossbar assembly, through a rotating support device, achieve 360-degree rotation. This is only applicable to the railway freight inspection industry.

2. Railway freight intelligent inspection system: This system monitors and schedules inspection tasks. Using a trained artificial intelligence model, it performs real-time frame extraction, detection, and identification of video streams from railway freight trains. This system is designed specifically for the railway freight inspection industry.

Images