For a tribological experiment involving a steel shaft sliding in a self-lubricating bronze bearing, a semi-supervised machine learning method for the classification of the state of operation is proposed. During the translatory oscillating motion, the system may undergo different states of operation from normal to critical, showing self-recovering behaviour. A Random Forest classifier was trained on individual cycles from the lateral force data from four distinct experimental runs in order to distinguish between four states of operation. The labelling of the individual cycles proved to be crucial for a high prediction accuracy of the trained RF classifier. The proposed semi-supervised approach allows choosing within a range between automatically generated labels and full manual labelling by an expert user. The algorithm was at the current state used for ex post classification of the state of operation. Considering the results from the ex post analysis and providing a sufficiently sized training dataset, online classification of the state of operation of a system will be possible. This will allow taking active countermeasures to stabilise the system or to terminate the experiment before major damage occurs.

Tribologists often rely on triboexperiments to investigate factors that affect a tribosystem. The inherent dynamic behavior of the respective tribometer setups and its effect on data interpretation remain often unknown. In this study, a comprehensive analysis of sensor data obtained from lubricated and dry triboexperiments is performed. Data are generated on a pin-on-disc test rig with a silicon nitride ball on a steel disc contact with a rotation frequency of ~3 Hz. High-speed acquisition of sensor data up to 5 kHz is performed to resolve changes in the data within individual cycles. The characteristic frequencies of the system and their temporal evolution are determined via time-frequency analysis, which reveals periodic patterns in the sensor data. Cycle-based data evaluation allows the detection of localized events and changes during an operation and considerably reduces the apparent measurement uncertainty, as compared with an unreduced dataset. The data analysis and visualization routines presented herein may serve as a prototype for further application to tribometer setups.

This article presents results from an experimental setup for a dual energy-band vacuum spectrometer for total-reflection X-ray fluorescence analysis allowing simultaneous efficient excitation of low, medium, and high Z elements. The spectrometer is equipped with an air-cooled 35 W low power Rh X-ray tube and a 17 mm² silicon drift detector with a thin 8 µm beryllium window. A Pd/B₄C multilayer monochromator is used at the same time as a Bragg reflector for Rh-Kα radiation and as a high-energy cut-off reflector above 5 keV, where the characteristic Rh-L radiation is totally reflected and present in the spectrum of the exciting radiation. This leaves one broad low energy band below 5 keV and one high energy band around the energy of Rh-Kα. As Rh-L radiation would be absorbed on its path through air, a new beam entrance system was designed in order to guide the Rh-L photons into the vacuum chamber for efficient excitation of low Z elements. With this setup, elements down to sodium (Z = 11, E = 1.04 keV) could be detected. First results are presented, and spectra obtained in air as well as in vacuum are compared and discussed. Detection limits in the range of 1000 µg/kg for Na and around 140 µg/kg for Mg were achieved using the NIST SRM 1640 (trace elements in water).