First steps on fan matrix condition monitoring and fault diagnosis using an array of digital MEMS microphones

Abstract

Condition monitoring and fault diagnosis of complex mechanical systems were based on vibration analysis in the last decades. The sensors which were employed in these methods needed to be in contact with the vibrant surfaces of the machines. This fact was an important limitation of the corresponding methodologies. In the last years, this problem is trying to be avoided by means of the analysis of the noise, i.e. the acoustic signals, which are directly related with the vibrant surfaces, instead of the vibrations themselves. Both, acoustic and vibrational signals can reveal information related with machinery operation conditions. Using arrays of digital MEMS (Micro-Electro-Mechanical Systems) microphones allows creating systems with a high number of sensors, without paying a high cost. This work has studied the use of acoustic images, obtained by an array with 64 MEMS microphones (8×8) inside a hemianechoic chamber, in order to detect, characterize and, eventually, identify failure conditions of a fan matrix. The acoustic images have been processed to extract different geometric parameters. Afterwards, these parameters have been used in classification algorithms, based on Support Vector Machines (SVM), in order to identify failures on the fans of the matrix