Operating condition detection and fault diagnosis are very important for reliable operation of reciprocating compressors. Machine learning is one of the most powerful tools in this field. However, there are very few comprehensive reviews which summarize the current research of machine learning in monitoring reciprocating compressor operating condition and fault diagnosis. In this paper, the recent application of machine learning techniques in reciprocating compressor fault diagnosis is reviewed. The advantages and challenges in the detection process, based on three main monitoring parameters in practical applications, are discussed. Future research direction and development are proposed.

The health management of the reciprocating compressor is crucial for its long term steady operation and safety. Online condition monitoring technology for the reciprocating compressor is almost mature, whereas the fault diagnosis technologies are still insufficient to meet the need. Therefore, in this paper, a novel fault detection method for the reciprocating compressor based on digital image processing and artificial neural network (ANN) was proposed. This method is implemented to the sectionalized pressure–volume (p–V) curves, which are obtained by dividing a working cycle in the cylinder into four thermal processes, including expansion, suction, compression and discharge. Hit-or-miss transform is adopted to extract the comprehensive gradients of expansion and compression curves, and vertical projection transform is applied to extract the vertical projection features. Finally, all of the features are fed to an ANN to do classification. To validate the proposed method, a seeded fault testing was conducted to collect real running data. The results showed that the new approach shows a good performance, with a high classification accuracy of 97.9%.