DLLE-EWMA based Incipient Fault Detection for Satellite Attitude Control System

In this paper, an incipient fault detection (FD) method that is applied to satellite attitude control system (ACS), is proposed based on the locally linear embedding algorithm with dynamic neighborhood parameters (DLLE) and the exponentially weighted moving average (EWMA) strategy. Dynamic neighborhood parameter selection is introduced to LLE, so that the weight reconstruction matrix can be determined according to the sample density of the manifold. Then, the DLLE algorithm is integrated with EWMA in both the latent and residual subspaces to establish the EWMA-T2 and EWMA-SPE statistics for incipient fault detection. Case study is conducted using the real telemetry data from two on-orbit satellites, and the results can demonstrate the effectiveness and feasibility of this proposed incipient FD method.