The traditional applications of subspace-based fault detection method are in health monitoring of mechanical structures, and the method traditionally only uses output measurements and assume that the input to the system is Gaussian noise. But, sometimes results are found unsatisfactory in real applications. In this research, we started to use real values of input data, thereby considerably improving the quality of fault detection. As a second approach to improve the method, the idea of active fault detection was applied to subspace-based fault detection method. For more information see (A. Esna Ashari and L. Mevel 2012) and (A. Esna Ashari and L. Mevel 2013).