Adaptive calibration for ultrasound reflectometry methods used in the detection of lubrication film thickness is of great research interest. This is mainly due to the versatile non-destructive implementation of the technology in industrial applications, allowing for measurements of the lubrication film thickness, which directly relates to the friction, wear, and overall efficiency of the system. This study reviews and compares a curve fitting, extended Kalman filter, and resonance frequency detection adaptive calibration approach. Furthermore, the study compares two different regression models, used for the curve fitting and the extended Kalman filter approach. The study compares the methods and regression models based on both a theoretical and experimental analysis. The experimental analysis is based on data with varying lubrication film thickness. The lubrication film thickness is varied such that the resonance frequency is both detectable and non-detectable within the ultrasound transducer bandwidth. It is found that all three methods give results very similar to those achieved through manual calibration. However, since the extended Kalman filter approach is the only method that allows for continuous calibration, this method is found to have the greatest potential of the methods compared.

Lubrication film thickness monitoring between moving contact surfaces within industrial machinery is of great research interest because it is linked to the efficiency and rate of maintenance of the machine. Ultrasound reflectometry shows potential within this field of research, primarily because the ultrasound transducers can be conveniently implemented in operation-ready machinery. In this article, the lubrication film thickness inside an operational radial piston motor is estimated using two adaptive ultrasound reflectometry methods. A new Gaussian random walk based adaptive ultrasound reflectometry method is compared against the existing layer phase-lag method with adaptive Extended Kalman filter based calibration. It is found that the new method proposed gives less noisy and more accurate lubrication film thickness estimates, but it is more difficult to tune. It is found that both methods can detect the significant lubrication film thickness dynamics present in a radial piston motor, but they need further investigation regarding their robustness and reliability.