Faults identification of underwater vehicle based on the states-switching unscented Kalman filter

Fault detection and identification play a key role in the active fault-tolerant control of underwater vehicles. A fault identification model based on the states-switching unscented Kalman Filters is constructed in this paper to the common sensor and actuator faults of the underwater vehicles. Firstly, states of underwater vehicle in each work mode are estimated by two groups of states-switching unscented Kalman Filters respectively. Then, each probability of the underwater vehicles work mode is calculated by Bayesian formula recursively. According to the probability value, faults can be identified. The simulation results indicate that the states-switching unscented Kalman Filters can estimate the underwater vehicle states rapidly and accurately with less computation. The faults identification system can identify various faults mode timelier and more accurately based on the states-switching unscented Kalman Filters states estimation.