Control of electro-hydraulic servo system for a material test system using fuzzy nerual network

This paper deals with the problem of the controller design for the electro-hydraulic servo position control system in a material test system of rock mechanics using fuzzy neural network. Because rock mass is a nonlinear elastic and heterogeneous medium with the ubiquitous micro-holes and micro-cracks and easily broken, its stiffness and damping coefficients are variable during the experimental process. In this paper, the resistant force of the rock specimen is treated as to be a function of time-varying stiffness, which can be estimated during the experimental process. A fuzzy neural network controller with the inputs of the position error and its derivative, and the time-varying stiffness is designed to control the electro-hydraulic position servo system, which is trained by computer simulation of the system with the known parameters. The experimental results of a compressive test on a rock specimen with six loading-unloading cycles show the effectiveness of the proposed method.