On-line friction modeling, estimation and compensation for position control

Problems associated with friction have plagued high precision servo-control applications (i.e., manufacturing as well as other industries) for many years. To avoid these problems researchers have developed various compensation techniques to deal with time-invariant friction. However, in complex applications where variable operating environments or wear exist, friction may not be time invariant. Thus, time invariant friction models can't produce an accurate representation of the friction. These inaccuracies will lead to poor compensation. To address this problem it is necessary to determine the best suited model, which can be used for time-varying friction, in order to continuously track any variations in the frictional resistance.; In this research a four-parameter friction model is extended for on-line application and compared to other standard models for validation. Next, an on-line approach for friction estimation using an extended Kalman filter is developed to allow for time-varying frictional parameters. The real-time estimated friction is then used in a feedforward compensator to provide high-precision servo control. To illustrate the method without introducing undue complexity, the study utilizes a single-degree-of-freedom mass-spring-damper system. In addition, sensitivity, controllability, observability, and convergence issues are also studied. Finally, simulations and experimental results are used to validate the proposed technique.