Development of proactive non contact condition monitoring system for rotating machine elements

Abstract

Condition based maintenance and condition monitoring are associated with maintenance of equipments based on the real-time condition of subsystem(s) of the machine. Vibration, temperature and acoustic signals have been used for machine health monitoring, vibration signature being the most widely used parameter. In the present work, a laser beam based non-contact vibration pickup has been designed in which an array of Light Dependant Resistors (LDR) were used to receive the vibration signal by studying the change in intensity and deviation of the laser beam due to the vibration of the machine. Non- contact Optimal Sensor Placement (NC-OSP) methodology has also been proposed in order to optimize sensor position. Experiments were conducted with different bearing conditions for vibration signals based bearing condition monitoring with speed, load and bearing faults as input parameters. The acquired signals were preprocessed using Hilbert transform for demodulation in order to obtain fault characteristic frequencies. Statistical methods have been utilized for extracting the features and the dimensionality reduction was carried out with Principal Component Analysis (PCA). Sequential Floating Forward Selection (SFFS) based feature selection method has been used for identifying the features that significantly contribute to classification and determining the number of such features. Finally, classification and performance evaluation was done using different classifiers viz. SVM, kNN and ANN have been discussed. Among the three machine learning classifiers used in this study, ANN outperformed SVM and kNN. The performance of developed non-contact sensor was validated using a standard accelerometer for both raw and envelope signals. The results reveal that the vibration signatures obtained from the developed non-contact sensor compare well with the accelerometer data obtained under the same conditions in predicting the health of the machines.