Analysis And Research On Multi-sensor Information Fusion Method

Urban rail transit has been widely popularized in major cities due to its large volume,fast speed,and effective relief of traffic congestion.However,the rapid development also brought many disadvantages,mainly reflected in the deformation of the track and tunnel.The conventional detection methods are mainly handheld detectors and large-scale rail inspection vehicles.The detection efficiency of the handheld detectors is low,and the accuracy is not high.Large-scale rail inspection vehicles are expensive,bulky,and inconvenient to go on and off the line.The information collected by the sensors used in the above detection method is processed separately and in isolation,resulting in an increase in the workload of information processing.The internal connection between the sensor information is cut off,and the internal characteristics of the combined information are lost.The various sensors cannot be effectively combined,which causes a waste of information resources and may even lead to wrong decisions.In order to solve the above problems,it is necessary to be able to effectively fuse multi-source sensor data.This paper realizes multi-dimensional detection of tracks and tunnels by mounting a variety of sensors on the small track detection vehicle,namely inertial navigation system,lidar sensor,laser displacement sensor,shaft encoder and camera.The system adopts the PXI chassis to achieve the consistency of multi-sensor data acquisition,that is,the consistency in time and the scalability in space are improved.A single sensor has its own advantages and disadvantages in the deformation detection of rails and tunnels.The camera's biggest challenges are embodied in: no depth information,limited field of view,and greater influence from external conditions.The biggest challenges of lidar sensors are embodied in: close sensing range and limited angle resolution.The information of each sensor complements each other's advantages,complements each other's strengths and effectively merges,overcoming the uncertainty and limitations of a single sensor,resulting in more accurate track and tunnel status information.The key technology of multi-sensor fusion lies in the choice of fusion strategy and corresponding algorithm.In this paper,the fusion of laser displacement sensors and inertial navigation system data to obtain more accurate track status information;fusion of lidar sensor and camera data to obtain more accurate tunnel status information.The data from the shaft encoder and the inertial navigation system are fused to obtain more accurate mileage information.