BP_Adaboost Algorithm-based Approach For Rail Surface State Recognition

Restricted by objective conditions such as wide geographical span,complex terrain and lines,the operating environment of trains is complex and changeable,and the rail surface state of different lines and stages may change randomly.In order to ensure the safe operation of the train,it is necessary to identify the state of the rail surface.Based on the characteristic data of rail surface adhesion state,this paper adopts the neural network model and takes the data of rail surface adhesion state as input to classify and identify the current state of rail surface.The research content of this paper is as follows:In view of the existing rail surface recognition accuracy is not ideal,the problem of recognition speed is slow,study a rail surface based on Adaboost algorithm of neural network identification model,in the case of small sample data,the algorithm can still keep good recognition effect.Aiming at the compensation problem in data collection,a strong predictor based on BP_Adaboost algorithm is designed,and the validity and feasibility of the method is verified by simulating the data which is difficult to be collected.In the traditional BP neural network,the particle is easy to converge to the local minimum point in the solution space,an improved PSO algorithm is adopted to optimize the network parameters,and the simulation experiment results show that this method can better solve the problem of local extremum.For the slower speed of PSO,has the problem of computing resources waste,the MEA optimize the network weights and threshold value,compared with improved particle swarm optimization algorithm are simulated,the results show that mind evolutionary algorithm to solve local extreme value at the same time,than the improved particle swarm algorithm faster calculation speed,saves computing resources.