Feature extraction has been widely used in sensor fault detection. Commonly used feature extraction methods such as PCA and MDS involve signal process of liner time-invariant systems, which are less effective in dealing with the nonlinear systems. In this paper, we will present that Local Linear Embedding (LLE) concept is adopted to solve the fault detection problems and that certain enhancement have been made to make LLE approach more efficient and robust in the extraction of signal features. Test results are given to demonstrate the effectiveness of the enhanced LLE method.
Zhang W,Li B,Zhou WJ. A LLE-based Approach to Sensor Fault Detection[C]. International Joint Conference on Neural Networks. Hong Kong, China. June 1-8, 2008.A LLE-based Approach to Sensor Fault Detection.