Adaptive friction compensation with partially known dynamic friction model

This paper illustrates the application of a model-based adaptive friction compensation on a DC motor servomechanism. The dynamic friction model and the control structure studied previously by the authors were used as a basis for this study. The paper first proposes a two-step off-line method to estimate the nominal static and dynamic parameters associated with the model. Then two adaptive globally stable mechanisms are introduced to deal with structured normal forces and temperature variations. Assuming that a nominal friction model is known and that the friction variations can be suitably structured, adaptation is performed on the basis of only one parameter. The paper presents experimental results validating the identification of the dynamic friction model and the adaptive control scheme. These results show that the adaptive loop improves over a fixed compensation scheme and over a PID controller without friction compensation.