In order to ensure the safety of railway transportation, it is necessary to regularly check for faults and defects in the railway system. Visual inspection technology is conducive to improving the low efficiency, poor economy and inaccurate detection results of traditional detection methods. This paper introduces the research and contribution of various scholars in the field of visual inspection, summarizes the application and development of visual inspection technology in the railway industry, and finally forecasts the future research direction of visual inspection technology.

To ensure the safe operation of high-speed trains, catenary anomaly detection and alerting security have become an urgent problem to solve. In this paper, we propose a novel method for abnormal behavior localization of a pantograph-catenary for high-speed trains. First, a modified faster RCNN is proposed to detect the pantograph faults. By adjusting the parameters of the faster RCNN, the positional accuracy of the candidate box and accuracy of the algorithm are guaranteed. We perform the arc detection after detecting the pantograph head area. The detection accuracy is over 99%. The height of the pantograph center point is also obtained during detection. Then, the actual running mileage of the fault point is calculated. Experiments show that the method proposed in this paper is also applicable to various complex scenes and that this method can determine the fault localization in the shortest time, narrow the maintenance scope, and improve the overhaul efficiency.

This paper proposes a current sensor fault detection method based on a sliding mode observer for the torque closed-loop control system of interior permanent magnet synchronous motors. First, a sliding mode observer based on the extended flux linkage is built to simplify the motor model, which effectively eliminates the phenomenon of salient poles and the dependence on the direct axis inductance parameter, and can also be used for real-time calculation of feedback torque. Then a sliding mode current observer is constructed in αβ coordinates to generate the fault residuals of the phase current sensors. The method can accurately identify abrupt gain faults and slow-variation offset faults in real time in faulty sensors, and the generated residuals of the designed fault detection system are not affected by the unknown input, the structure of the observer, and the theoretical derivation and the stability proof process are concise and simple. The RT-LAB real-time simulation is used to build a simulation model of the hardware in the loop. The simulation and experimental results demonstrate the feasibility and effectiveness of the proposed method.

Detecting obstacles accurately is a vital part of realizing autonomous driving. But utilizing single sensors, for example, RGB camera or LiDAR itself, is difficult to meet the needs of autonomous driving under the condition of sparse environmental sensing data. In contrast, equipments that can collect data in large quantities are expensive and difficult to be utilized in practical applications. Hence, under the premise of ensuring low application costs, the difficult task is to obtain high-precision three-dimensional information of obstacles in real-time from sparse data. To overcome this problem, a real-time detection technique that combines LiDAR data and image information for obstacles detection is proposed in this article. It can detect vehicles close to the ART in sparse data in real-time. The sparse point cloud is de-noised initially and the ground is separated from the initial cloud data. The remained cloud, which includes surrounding vehicles, are processed by the clustering algorithm according to the structural relationship and scanning mechanism of the LiDAR. Then the processed LiDAR data is fused with the classification information generated from images based on a neural network to locate suspicious obstacles. The ratio values between the central values of vehicles and image width are obtained in the image coordinate system. Then the category information and ratio values are mapped on the processed LiDAR clusters during ART operation. The method is tested and verified in the real field. The evaluation results show that the proposed method can achieve an average accuracy of 85.5% for obstacle classification with a cost-effective sensing suit.

This paper deals with a current sensor fault reconstruction algorithm for the torque closed-loop drive system of an interior PMSM. First, sensor faults are equated to actuator ones by a new introduced state variable. Then, in αβ coordinates, based on the motor model with active flux linkage, a current observer is constructed with a specific sliding mode equivalent control methodology to eliminate the effects of unknown disturbances, and the phase current sensor faults are reconstructed by means of an adaptive method. Finally, an αβ axis current fault processing module is designed based on the reconstructed value. The feasibility and effectiveness of the proposed method are verified by simulation and experimental tests on the RT-LAB platform.