University : İstanbul Technical University
|
IDENTIFICATION AND POSITION CONTROL OF A SYSTEM CONSISTING OF A PROPORTIONAL VALVE AND A HYDRAULIC CYLINDER İlyas İSTİF This study considers the application of neural network model based control of a hydraulic cylinder, which is driven by a proportional valve, subjected to variable reference trajectory and variable load conditions. For the simulation, a simulink model is developed for a proportional valve controlled hydraulic cylinder system. The fourth order nonlinear system model is fully implemented in Matlab’s Simulink simulation package and the model of each hydraulic components is developed and combined in a library for easy reuse. Bond graph models of the hydraulic components are also developed and bond graph model of complete system is obtained by combining bond graph models of the hydraulic system components. Identification of the electrohydraulic system is performed. For the identification process, PRBS signal is generated and applied to the experimental setup and the position of the piston is measured. Well known model types, such as ARX, ARMAX, State Space, Box Jenkins and Output Error, are used to model the electrohydraulic system by input-output data. Neural network predictive control and feedback linearization control are applied to the fourth order nonlinear system model and position control is performed. The performance of the neural network based control algorithms are also investigated by changing system parameters and by using sinusoidal and step reference trajectory. Both sinusoidal and step reference trajectory are followed by the piston with very small errors. Command signal, which is obtained by simulation, is applied to the experimental setup and simulation results and hydraulic system response are discussed. Keywords: Proportional valve, hydraulic system, neural networks, control, identification. |