PENGENALAN POLA CITRA TERMOGRAFI PADA PEMANTAUAN KONDISI MESIN DENGAN METODE PENGOLAHAN CITRA DIGITAL

Abstract

Rotating machines and reciprocating machines like turbines, combustion engines and electric motor are the most widely used in industry. These machine consist of components which generate vibration dan heat during operation. Excessive vibration and heat can be indicators of fault in the machine. Therefore, machine condition monitoring is needed to monitor the real time machine condition with proper devices and methods. One of those being developed now is using infrared thermography. This method is categorized in the nondestructive testing (NDT) or inspection without damage. The aim of this study is to identify the patterns of digital image thermography by means image clustering and image features for machine condition monitoring. Image features used in this study are area, perimeter and central moments of hotspots region. This study was conducted by capturing thermal image of bearing from machine fault simulator by varying the conditions which are common in rotating machinery such as, mass unbalance, misalignment, looseness, and bearing faults. Self Organizing Map (SOM) method was used to pattern recognition of machine condition based on extracted image features. Using this method, the fault in the machine can be diagnosed. There are several factors that affect the accuracy of diagnosis such as the size of data image, clustering process and image features. The small data image leads the low accuracy of diagnosis process.