Research on Faulted Prediction Method of Marine Gyroscope Based on GT-RVM

Marine gyroscope is an important guarantee for the safety and reliability of ships. Aiming at the issue of fault prediction technique of marine gyroscope, a method based on grey theory (GT) and relevance vector machine (RVM) for the prediction of marine gyroscope was proposed. After analyzing the failure mode of gyroscope in marine inertial navigation system, the random drift data was extracted as the characteristic parameter of the fault. Considering the randomness, nonlinearity, non-stationary and weak time variability of random drift data, the relevance vector machine was used to predict the fault. The accumulative generating operation (AGO) and inverse accumulated generating operation (IAGO) were used to process the original data. To improve the accuracy of the long term prediction of the RVM, the grey relational analysis is used to update the RVM model dynamically. The experimental results show that the proposed method can trace the characteristic parameters' trend and can be effectively applied in fault prediction of marine gyroscope.