Research On Intelligent Electric Locomotive Environmental Perception System Based On Multi-sensor

Due to the frequent occurrence of safety accidents in the transportation of electric locomotives during the construction of subway tunnels,people's attention to driving safety of electric locomotives has been further improved and the environmental perception system of intelligent electric locomotives has become a hot research topic.Collision and derailment are the most common safety accidents of electric locomotives,but there is still no effective detection method.This paper focuses on the obstacle and derailment detection technology,designs a multi-sensor-based intelligent electric locomotive environment perception system,which provides more guarantee for driving safety.The main research work of this paper are presented as follows:Firstly,according to the working characteristics of electric locomotives,the overall design of the environmental perception system is completed.The working principle and performance characteristics of each type of sensor is analyzed,which helps to determine the sensor type.Model selection and parameter configuration of sensors are carried out according to the environmental sensing requirements of electric locomotive.According to the two functional modules of obstacle detection and derailment detection,the overall architecture and communication network distribution of the environment perception system are designed.Secondly,based on the image processing technology and the working characteristics of Lidar,a method for detecting obstacles in front of the tracks is proposed.Extract the track outline in the camera picture through image processing technology to determine the track type.According to the size of vehicle body and the installation position of Lidar,determine the detection limit conditions to remove the interference points and the obstacle detection on the straight track is realized.Aiming at the problem that Lidar is prone to misdetection on curved track,the internal parameters of the camera are calibrated to establish a spatial transformation model of coordinate systems.The point cloud at the same time is projected onto the image plane.According to the projection point and the track boundary position,the interference of the target point outside the track is eliminated,the false detection rate on the curved track is reduced.Thirdly,a derailment detection method based on D-S evidence theory is proposed.By analyzing the characteristics of the motion information when the electric locomotive is derailed,the evidence theoretical model is established based on the four types of information obtained by the motion attitude sensor and sound sensor: lateral acceleration,longitudinal acceleration,roll angle,and noise intensity in the vehicle.By constructing the fuzzy set between the derailment state and information,the membership function is taken as the basic probability assignment function to obtain the initial evidence.For the high conflict appearing in the fusion process of the original evidence,the fuzzy similarity is used to assign the weight of each evidence,the initial evidence is weighted and then fused by the Dempster combination rule.After improving the combination rules,the confidence of decision-making is improved,and more accurate derailment detection results are obtained.Finally,Lidar and camera are installed on the vehicle to build the experimental platform,the obstacle detection function verification experiments were carried out on straight track and curved track respectively.The experimental results prove that the obstacle detection function basically meets the design requirements.The performance of the derailment detection algorithm is tested based on the historical data.The comparison experiments with other detection methods verify the robustness of the method under different vehicle speeds.