Hydraulic actuator control with open-centre electrohydraulic valve using a cerebellar model articulation controller neural network algorithm

Abstract

Real-time learning control of a hydraulic actuator is studied using a cerebellar model articulation controller (CMAC) neural network architecture. Experiments are conducted on a one-degree-of-freedom hydraulic cylinder. The main applications considered are the earth-moving equipment applications. The electrohydraulic valve used to control the flow in the hydraulic circuit is an open-centre non-pressure-compensated type of valve, the type used in most earth-moving vehicles. Large hysteresis and dead band are observed in the hydraulic system which result in a large delay and poor dynamic tracking performance. A comparison of the experimental results on the control of the hydraulic actuator for the CMAC controller and a conventional proportional-integral controller with a model-based feedforward valve transform is made.