On the use of cyclostationary indicators in IC engine quality control by cold tests

• Published in: Mechanical systems and signal processing 2015-08, Vol.60-61, p.208-228
• Main Authors: Delvecchio, S., D’Elia, G., Dalpiaz, G.
• Format: Article
• Language: English
• Subjects: Cold tests; Condition monitoring; Engines; Faults; IC engine; Indicators; Indicators of Cyclostationary; Joining; Mechanical systems; Quality control; Roller bearings; Screws; Signal processing; Vibration
• ISSN: 0888-3270; 1096-1216
• DOI: 10.1016/j.ymssp.2014.09.015

This paper addresses the use of first- and second-order cyclostationary (CS1 and CS2) tools to process the vibration signals picked up from internal combustion (IC) engines during cold tests. This type of analysis is needed in order to detect and diagnose irregular operations for quality control purposes. The effectiveness of indicators such as Mean Instantaneous Power (MIP), Degree of Cyclostationarity (DCSα) and Indicator of Cyclostationarity (ICSnx) in detecting assembly faults has been tested on real signals concerning three faulty conditions: inverted piston, connecting rod with incorrectly tightened screws, connecting rod without one bearing cap. In the past several authors have mainly used cyclostationary metrics for diagnostics purposes in rolling bearings and gear systems. Moreover, a signal model, qualitatively reproducing the features of actual cold test signals, has been formulated and used in order to preliminarily study the influence of signal parameters on the Indicators of Cyclostationarity. The results indicate that the cyclostationary tools – mainly CS2 tools – are effective in detecting and diagnosing all tested faulty conditions. In particular, indicator IC⌢S2x is highly sensitive to faults and it is suitable as pass/fail tool in quality control at the end of the engine assembly line. As a further second-order cyclostationary metric, the MIP is effective for detection, as well for fault identification, since it is able to localize regular and fault events within the engine cycle. In addition DCSα effectively characterizes the CS2 periodicities, giving the cyclic order distribution. Since these CS2 tools require a moderate computation cost, they can be considered ready for on-line industrial applications.