Analysis of Large Sets of Seemingly-Random Experimental Data: The Case of Electron Emission from Frictional Contacts
Analysis of seemingly random experimental data can be difficult when no phenomenological model is known. If available data is output for even simple but unknown-dynamics systems, hypotheses could be made and tested for possible underlying processes, being them from pure random to deterministic ones, but testing of such hypotheses can be difficult in the practice. For instance, the study of theoretical systems leading to chaos is an established mathematical field, but the testing of given experimental data for the hypotheses of deterministic origin versus a stochastic one is not simple and it may not be possible. Available analysis techniques are reviewed and discussed. The author and colleagues have carried out extensive experimental research work on emission of electrons as a probe for in-situ on-time surface monitoring, seeking a better understanding of this triboemission from dry-sliding contacts and during wear, and of the fractoemission that occurs during plastic deformation and failure. Particle outputs may be very complex and carry limited information; Electron triboemission data are typically composed of seemingly deterministic bursts of emission which are superimposed to lower but seemingly constant levels of random emission. They are large sets that are acquired in very short-time windows, were the discrete occurrence of counts is matched to detected particles, but no information about their energy or paths can be simultaneously obtained. Analysis of triboemission outputs required new approaches and techniques: the author studied different stochastic-process distributions for fitting to such data, and he also tested the hypothesis of deterministic-chaos origin. The proposed data analysis can be tools for the study of similar complex systems. For instance, the author believes that understanding of triboemission and of related mechanisms is a key to modeling of frictional and charging processes for insulators, mainly for ceramics, and for semiconductors.
Molina, Gustavo J..
"Analysis of Large Sets of Seemingly-Random Experimental Data: The Case of Electron Emission from Frictional Contacts."
Mecanica Computacional, XXVI (2007): 2728-2741.