Position Fusion Navigation and Location Method of Train Inspectors Based on Particle Swarm Optimization Neural Network
The train inspection site is large,and it is difficult to judge whether a personn is in the safe area.In order to obtain the position information of the train inspector in real time,reasonably and accurately,a train inspector position fusion naviga-tion positioning method based on particle swarm optimization neural network is proposed.This paper determines the hierarchi-cal structure of the comprehensive management platform for train inspection safety operation,and uses the platform as the basis for navigation algorithm implementation.It collects the initial position and movement speed of train inspectors in the way of combining inertial orientation navigation and global navigation satellite positioning system to ensure the pertinence of the basic input and output process.It analyzes the process of particle swarm optimization,including particle initialization and population evaluation,determines the network architecture,select the activation function,and establishes the classical neural network model to ensure the rationality of the fusion navigation process.The method of particle swarm optimization(PSO)is used to improve the network searching ability and avoid falling into local optimum.The collected fusion information is taken as the net-work input,the connection weight values and network related parameters are set,and the particle speed and position are upda-ted according to the fitness value.When the convergence accuracy meets the requirements,the navigation and positioning re-sults are output.The experimental results show that the positioning results of this method agree well with the actual position.Only 300 iterations are needed to achieve the algorithm convergence,and the training error is basically kept below 0.002.It can effectively locate the specific operation position of the train inspector,improving the real-time navigation and positioning of the train inspector.
particle swarm optimizationneural networktrain inspectorposition fusionnavigation and positioning