Friction Behavior of a Wet Clutch Subjected to Accelerated Degradation

Abstract

This study aims at experimentally investigating the sliding friction characteristics of a wet clutch during its lifetime. More precisely, the objective is to understand how the Stribeck and the frictional lag (i.e, sliding hysteresis) parameters evolve as the clutch degradation progresses. For this purpose, a novel test procedure has been proposed and a set of experiments has been carried out on a fully assembled (commercial) clutch using a modified SAE#2 test setup. Furthermore, a systematic methodology for the Stribeck and the frictional lag parameters identification is developed. Regardless of the applied pressure, it appears that the first three identified Stribeck parameters tend to decrease with the progression of the degradation, while the last parameter tends to increase. In regard to the frictional lag parameter, the trend shows pressure dependency.

Preventing Catastrophic Forgetting using Prior Transfer in Physics Informed Bayesian Neural Networks

“…For the training of F 1 , the hyperparameter S that makes a trade-off between the likelihood cost and the KL-divergence cost (see (11)), is tuned for each layer of the network. Taking S = 1 1500l , with l the amount of links connected to the layer, provided good results.…”

“…The trade-off between the likelihood cost and the KLdivergence cost can be shifted by adding an additional hyperparameter S to the cost function (11). Additionally the likelihood cost is calculated batchwise, thus the likelihood cost is multiplied by the amount of batches b.…”

“…Van Heck, A. Coene and G. Crevecoeur are member of EEDT-DC, Flanders Make, Belgium. accommodate for changes in the system such as wear or lack of lubrication leading up to changing friction levels [11]. In other words, physics-based models have the ability to capture system dynamics but when uncertainties arise, certainly in the face of changing operating conditions, their predictive capabilities deteriorate [12].…”