A Novel SINS/USBL Tightly Integrated Navigation Strategy Based on Improved ANFIS

Abstract

Strapdown inertial navigation system (SINS)/ ultra- short baseline (USBL) integrated navigation system is widely used in navigation and positioning of underwater vehicle. In an unknown and complex underwater environment, unstable USBL measurement information can lead to decreased navigation accuracy. In the traditional method, the interference to USBL is uniformly modeled as outliers, but in practice, it is also represented as slowly varying errors, update irregular errors. The latter two errors are not easy to be detected at the beginning of occurrence. Therefore, a SINS/USBL integrated navigation technology based on novel adaptive neural fuzzy inference system (ANFIS) is proposed. This model can detect and compensate abnormal USBL information and maximize USBL information. In order to improve the prediction accuracy of ANFIS, an ANFIS algorithm based on variational Bayesian Kalman filter and principal component analysis (PCA) is proposed to avoid the decrease of positioning accuracy caused by false compensation. The simulation shows that the designed ANFIS algorithm can accurately predict the above anomalies. And it can overcome the problem that USBL measurement anomalies are difficult to be detected at the initial stage. Combined with the proposed integrated navigation technology, it obtain the same positioning accuracy in complex environment as in good environment. The effectiveness and robustness of the proposed method are verified by river experiment.