Non-Contact Measurement for Mechanical Fault detection in Production Line

Abstract

Appliance manufacturing companies more often ask for an automatic on-line inspection system to accurately monitor the characteristics of all their products. It is well known that vibration tests enable discrimination between good and faulty products and hence the analysis of the vibration signals can be used for quality control of household appliances on the production lines. Laser Doppler Vibrometry (LDV) is now an established technique for vibration measurements in industrial applications where non-contact operations are essential. Despite the advantages of the LDV, speckle noise occurs when rough surfaces are measured and the object is moving. Therefore, spike removal is a crucial point for a reliable system of mechanical defects detection. This paper deals with the integration of pattern recognition techniques into an automatic test system for data acquisition and classification, in order to detect mechanical faults of washing machines (WM) in the production line. In particular, as the electrical motor is one of the most critical part of the assembled system, the goal is to detect the faults related to the motor by the use of a Laser Doppler Vibrometer pointing the tub of the washing machine. First, data acquisition and its problems are introduced. Then, the adopted pre-processing techniques for speckle noise reduction is illustrated. Finally, feature extraction and real examples are shown to test the system.