Subspace aided data-driven fault detection for LTI systems

This paper mainly solves the issue of subspace aided data-driven fault detection for LTI systems, identifying the residual generator without the specific model. By the input and output data, Akaike information criterion and singular value decomposition are used to determined the system order firstly and a comparison between them is made. Then based on a certain algorithm, we can get some useful subspace to construct a residual generator for fault detection effectively. Finally, we apply the theoretical method into simulation studies to show its feasibility and effectiveness.