NEURAL PREDICTIVE FORCE CONTROL FOR A HYDRAULIC ACTUATOR: SIMULATION AND EXPERIMENT

Abstract

& This paper proposes a neural-based predictive control algorithm for online control of a forceacting industrial hydraulic actuator.In the algorithm, a multilayer feedforward neural network is employed to modeling the highly nonlinear hydraulic actuator.The nonlinear neural model is instantaneously linearized at each sampling point.Estimated parameters from the linearized model are used in the generalized predictive control (GPC) algorithm to control the contact force.Simulation and experimental results show that the neural-based predictive controller can adapt to different environments and keep the contact force in a desired value despite high nonlinearity and uncertainty in the hydraulic actuator system.Many tasks completed by robots need the robotic machines to interact with uncertain environments in a controlled manner (Zeng and Hemami 1997).For example, robots perform the tasks of pushing, scraping, grinding, and twisting.All of these capabilities intrinsically require that the robots be force-controlled.Furthermore, in bioengineering area, when bio-robots are engaged in surgery and other human-like operations, such as human skin harvest and muscle exercise (Duchemin et al. 2005;Jaax and Hannaford 2004), force contact control is a critical issue to guarantee satisfactory performance.There are various ways to implement force controls.Among them, hydraulic servo actuators are considered ideal candidates for many practical cases due to their standard, safe, high-load capacity, and reliable performance.However, because of high nonlinearity and uncertainty in the hydraulic system, designing a controller to effectively control the force generated byThe experiments were conducted at the Experimental Robotics and Tele-Operation Laboratory at the University of Manitoba.