Statistical Classification Approach to Valve Condition Monitoring Using Pressure Features

This paper provides a novel health monitoring classification algorithm to estimate the level of valve leakage in reciprocating compressors. It is expected that will change shape pressure-volume diagram. method constructs feature space using polytropic exponent during expansion and compression phases as well discharge suction loss power. These features are extracted by measuring in-cylinder pressure, crank angle. Linear quadratic discriminant classifiers used machine learning approaches classify compressor. implemented on single-stage double-acting industrial gas compressor operating air. Faults seeded precisely machining poppets simulate common leakage. The approach shows high accuracy determining degree promise for future work prognostics.