This paper presents a condition monitoring technique for the identification and correction of faults in reciprocating machines by recording engine vibration signals arising from different processes. The proposed data acquisition set-up allows the analyst to record vibration signals
from different locations, analyse them and make the appropriate decisions regarding predictive maintenance. Data are recorded before and after carrying out major overhauling to assess the effectiveness of the proposed condition monitoring technique. For a complete health assessment, vibration
signals and ultrasonic signature with respect to the crank angle are recorded and interpreted. All of the data are obtained at constant engine speed to depict the mechanical condition of the components and indicate any poor performance. A scuffed main bearing and a broken piston ring are diagnosed
successfully. Suspected components are replaced based upon the findings of the initial analysis and data are recorded again to verify the results. Post-evaluation data confirm a normal mechanical condition and satisfactory performance except for some initial friction. The main objective of
this experimentation is the identification of major tasks related to maintenance prior to major overhauling. This work shows that the use of non-intrusive condition monitoring techniques can reduce the frequency of failures by identifying developing faults.
Amplitude and phase interaction in Hilbert demodulation of vibration signals: Natural gear wear modeling and time tracking for condition monitoring
