Bio-Tribo-Acoustic Emissions: Condition Monitoring of a Simulated Joint Articulation

Acoustic emissions have been used to interpret the frictional processes observed in a simulated metal-on-polymer joint replacement articulation during in vitro testing. The coefficient of friction profile is predicted from AE features using a nonlinear autoregressive neural network with an external input model, and the evolution of surface damage is identified using k-means clustering of the distribution of emission types from running-in to prolonged sliding states. The predicted coefficient of friction profiles were found to exhibit a similar response to the actual coefficient of friction profiles. Clustering showed that a higher percentage of continuous emissions are generated during the prolonged sliding stage, indicating sliding friction being the most dominant process during that state. The findings of this study provide a significant pathway toward achieving the potential of AE testing as a more intuitive and dynamic process of monitoring the tribological conditions of artificial joints and diagnosing the pathologies of the natural joints. [Display omitted] •Acoustic emission signals from a simulated joint replacement articulation are acquired.•A NARX neural network is used to predict coefficient of friction profile from AE features.•K-means clustering is used to classify signals into emission types.•Distribution of continuous emission is instrumental in surface damage identification.•AE frequency spectra is used to identify wear mechanisms present during test.