Automation based electrohydraulic systems are most common in nowadays industry because they overcome the problems of pneumatic systems such as the high compressibility of air which gives poor dynamics and low power used. Electrohydraulic systems are those hydraulic systems which use valves actuated by electrical power. Three most common types of electrically operated valves
are on/off solenoid, proportional, and servo valves. Those types of valves suffer from some nonlinear effects like deadbands and solenoid hysteresis. The performance of electrohydraulic valves greatly affects the complete hydraulic circuit performance; therefore if some methods are introduced to solve or enhance the dynamic performance of the valves, this will increase the overall performance of those automated systems.
The dynamic performance can be enhanced by either using a different actuating hardware such as micro positioning devices (e.g. piezoelectric actuators) which don’t suffer from the hysteresis
problem, or by applying a software technology through using an intelligent control algorithm.
This research focuses on the proportional valve types. The proposed enhancement system is to apply a hybrid control algorithm based on an artificial intelligent controller to solve the
nonlinearities of the system through using a fast learning algorithm. Consequently, a hybrid control system based Cerebellar Model Articulation Controller (CMAC) neural network is presented. The proposed controller is composed of two parallel and instantaneous working controllers. The first is a conventional Proportional-Velocity (PV) servo type controller which is used to decrease the initial large error of the closed-loop system and the second is a CMAC neural network which is used as an intelligent controller to overcome the
nonlinear characteristics of the system used.
A fourth order model for the electrohydraulic system is introduced and all parameters are estimated by using Matlab/Simulink software. PV controller parameters are tuned to get optimal
values by using parameter estimation toolbox on Matlab/Simulink. CMAC is implemented using Simulink environment and its parameters are tuned to get the best CMAC control action. Three
different learning algorithms are tested, using a constant learning rate, using a variable learning rate, and learning by the PV control action. The tracking performance is measured using the root-mean-square (RMS) error and/or the average of the absolute errors (AAE) between the desired and the actual trajectories as performance indices for PV controller only and for comparing it with the PV-plus-CMAC proposed control scheme. Simulation and experimental results show a good tracking performance obtained using the proposed controller.
The robustness of the proposed controller is measured in two working environments. The first is to add two different inertia loads, and the second is working into two different noise level input
signals. The proposed controller gives better values of RMS error and/or AAE for all tests in simulation and experimental work, thus the proposed controller is robust. |
