Implementation of CMAC as a neural network controller on hydro-mechanical systems

Real time learning control of a non-linear hydro-mechanical system is studied using a cerebellar model articulation controller (CMAC) neural network architecture. Through simulation and experimentation, the control scheme was shown to perform well in the presence of system non-linearities. This control scheme showed that the CMAC controller can cope with complexities associated with the highly non-linear hydraulic system, which is more desirable than conventional control schemes that are typically designed using linearised models. A comparison of simulation and experimental results is made on the control of the hydro-mechanical system employing the CMAC controller and a conventional PID controller.