Research On Train Positioning Algorithm Based On Wireless Sensor Network

In recent years,China's railway construction has experienced substantial growth,and the advantages of railway transportation have become increasingly apparent.However,at the same time,this increase has also put tremendous pressure on the safe operation of railways.In the process of train operation,the real-time position information of train is the key to ensure the safe operation of train.The traditional wired train positioning system can provide reliable positioning accuracy,but there are some practical problems such as high cost,complicated deployment,and difficult maintenance.Global Positioning System(GPS)can achieve high positioning accuracy and it has been widely used in vehicle systems including railways,but it has many restrictions,such as train may often pass GPS blind spots of tunnels,hilly areas,or urban canyons.In view of the above problems,the wireless positioning method was introduced in the safety monitoring,tracking and positioning of train operation in recent years.Because of its low cost,simple deployment and maintenance,non-centralized,self-organizing,multi-hop routing and dynamic topology,wireless positioning method which is based on wireless sensor network(WSN)shows great potential in the field of train tracking and positioning.And it is innovated by people continuously.This dissertation focuses on the study of train positioning in a complex environment.Based on the theories of WSN,Monte Carlo principle and approximate Bayesian filtering algorithm,a WSN train positioning algorithm based on extended Kalman-particle filter is proposed.The main contents and research results of the paper are as follows:(1)Firstly,the model of WSN train positioning system is established according to the complexity of train operation.By deploying the anchor node on the side of the track and the gateway node on the track roof,the WSN construction for train positioning is completed.In the deployment of the WSN positioning system model,different deployment methods are adopted to realize the deployment of anchor nodes without blind spots,and to ensure the gateway nodes are always within the communication coverage of the anchor node at any time.In order to establish a non-blind zone and real-time two-way communication link between the anchor node and the gateway node,the satisfaction condition of the communication radius of the anchor node is calculated.(2)Due to the problems of particle degradation in Monte Carlo WSN mobile node positioning and the fact that particle filtering is not suitable for target positioning in the case of linear motion observation equations,this paper uses extended Kalman filter theory.In terms of train motion equation representation and data processing,the data received by the gateway node is classified,the signal strength and position coordinates of the anchor node are recorded,the appropriate positioning anchor node is selected,and the distance from thegateway node to the positioning anchor node is calculated,indicating the train motion equation.Using the advantage of the extended Kalman filter algorithm in the linear environment(motion observation equation),calculating and selecting appropriate importance probability density function to solve the particle degradation,establishing the resampling maneuvering model in the train movement process,and achieving the weight updating factor and updating weight value in different movement conditions to improve the positioning accuracy of train.Simulation experiments show that the proposed algorithm improves the positioning accuracy and the robustness of the algorithm compared with the traditional algorithm,and can further improve the positioning accuracy.(3)For the problem that each experiment may be influenced by different external factors and then it may produce large errors,this paper selects appropriate and effective positioning anchor nodes,and further uses resampling to select the optimal positioning anchor nodes,According to the optimal positioning anchor node combined with this algorithm,the positioning accuracy can be further improved.Through simulation experiments,compared with the particle filter positioning algorithm,under the linear motion model and curve motion model,the positioning accuracy of the algorithm is increased by 20% to 28%,which proves the effectiveness of the algorithm.