An AUV integrated navigation method based on improved genetic neural network

As extended Kalman filter (EKF) is liable to get divergence in the process of data fusion in autonomous underwater vehicle (AUV) integrated navigation system, a neural network (NN) based on the genetic algorithm (GA) is applied in the system. But there are many drawbacks such as prematurity, bad stability, fixed cross and mutation probability in the conventional GA, so an improved GA is proposed. The improvements include float coding, competition selection strategy, reservation of the best individual, “migration” mechanism, and redefined operators including crossover operator and adaptive crossover-mutation operator. The simulation results indicate that the algorithm is more effective, and achieves the precision of EKF.