Vehicle positioning algorithm based on particle swarm optimization BP neural network

For multi-sensor information fusion in vehicle navigation system, adaptive particle swarm optimizer (APSO) is used to substitute gradient descent in training parameters of BP neural network to improve BP performance. Cascade fusion architecture and Kalman model of GPS/DR/MM vehicle navigation system are proposed. The algorithm of APSO optimized BP neural network is described in detail. Three improvement strategies of APSO parameter setting are presented to advance the accuracy and speed of training. Test results showed that the proposed algorithm to be easily carried out, had strong global optimization ability, reduced the location error and improved vehicle location-matching accuracy.