State Classification of Reciprocating Compressor Valves using Event per Cycle Measurements
Sprache des Titels:
Englisch
Original Buchtitel:
Proceedings of the MFPT 2010
Original Kurzfassung:
This paper presents a novel approach for diagnosing faults in reciprocating compressors using analysis of event information per cycle. Based on easily available system variables, selected event information of the valves is used to detect damaged valves.
Then the discussed method tries to find an optimal combination of valve events to distinguish between normal and faulty operating conditions. Finally, the fault state of the valve is classified by applying pattern recognition techniques to the space of the valve events.