Fault Diagnosis of Remainders Inside PIGA Based on Pattern Recognition Algorithm

The three floated pendulous integrating gyro accelerometer (PIGA) is a key component in inertial navigation which is widely used in the aerospace field. PIGA is composed of numerous components, resulting in its complex model, high failure rate and difficulties in the failure detection. Aiming at the characteristic of PIGA, a new data-driven fault diagnosis algorithm, utilizing wavelet packet and energy entropy to extract the time-frequency distribution features of data then probabilistic neural network (PNN) and support vector machine (SVM) to classification is proposed. The method is validated in the fault data of surplus inside the PIGA. The classification accuracy of SVM based on radial basis kernel function (RBF) is 96.67% on the circumstance that the time window length is 1 second. The results demonstrate that the proposed algorithm can effectively and rapidly recognize the fault caused by remainder inside the PIGA, which does not depend on the model and can be conveniently transplanted to other kinds of inertial devices.