A Transferable Thruster Fault Diagnosis Approach for Autonomous Underwater Vehicle under Different Working Conditions with Insufficient Labeled Training Data

Existing thruster fault diagnosis methods for AUV (autonomous underwater vehicle) usually need sufficient labeled training data. However, it is unrealistic to get data each working condition in practice. Based on this challenge, a transferable approach proposed. In the approach, an IPSE (instantaneous power spectrum entropy) and STNED (signal-to-noise energy difference) are added SPWVD (smoothed pseudo Wigner-Ville distribution) identify time frequency boundaries of local region time-frequency caused by fault, forming TFE (time-frequency energy) method feature extraction. addition, RCQFFV (relative change quantity value), MSN (multiple scale normalization) LSP (least square prediction) SVDD (support vector description) align distributions samples, contributing TSVDD (transferable SVDD) classification samples. The experimental results prototype indicate that monotonic percentage thrust loss proposed but not SPWVD. has higher overall accuracy comparison conventional under conditions with no