Diagnosing Faults of Reciprocating Air Compressor (RAC) Setup Using Signal Processing Technique and Machine Learning Approach

This paper presents a method for identifying RAC faults using acoustic signals obtained from both healthy and unhealthy conditions. The entire procedure is carried out with microphones. Accumulated one healthy and seven unhealthy signals of RAC setup processed using unconventional method of signal processing called “local mean decomposition” (LMD). Additionally, the ‘6’ statistical properties (SP) have been evaluated in order to extract features: mean (US), variance ( \(\sigma_{s}^{2}\) ), root square of mean (Mrms), root amplitude of mean (Mrma), absolute amplitude of mean (Mama), and kurtosis index (Ki). Extracted fault features are classified using lazy learning-based (LLB) classifiers. It has been observed that from different types of LLB classifiers, K-star classifier is quite accurate and has higher accuracy (93.05%) as compared to the other two classifiers accuracy.