Sensor fault diagnosis method based on Hilbert marginal spectrum and supervised locally linear embedding and support vector machine

Sensor plays an important role in complex industrial environment. Therefore, researches on sensor fault diagnosis technology are important for improving the reliability of industry system. A sensor signal which is non-linear and non-stationary, has many kinds of structural characteristics and sensing properties. On the basis of supervised locally linear embedding (SLLE), support vector machine (SVM), and Hilbert marginal spectrum (HMS), as a means of sensor fault diagnosis is proposed in this thesis. HMS is put forward to feature analysis from obtained sensor signals. Then SLLE is put forward to reduce dimensionality of high dimensionality fault signal, that is more effective than other dimensionality reduction means, for example multi-dimensional scaling (MDS), locally linear embedding (LLE) and principal component analysis (PCA). In the end, support vector machine (SVM) is used to complete the sensor fault diagnosis on base of the obtained feature vector. The result shows that the above methods improve the feature extraction and recognition result observably. In the perspective of the results of the simulation, this mean may not merely be significantly used to the fault diagnosis of gas sensor but also offer a direction for other sensors.