Adaptive composite control of electronic throttle using local learning method

This paper deals with an application of receptive field weighted regression (RFWR) method for the composite control of automotive electromechanical throttle. Simulation experiments use the plant model optimized for hardware-in-the-loop simulation and tuned according to experimental results. Composite controller consists of PID and feedforward inverse dynamics implemented using RFWR. Because of significant dry friction in the system, two independent RFWR approximators are used to solve the problem of discontinuity at zero velocity. Satisfactory simulation results are presented.