Comparative Evaluation of Intelligent Agent Based Improved Control Designs of Electro-pneumatic Clutch Actuation System for Heavy Duty Vehicles
Ndubuisi Paul-Darlington Ibemezie
Rectory Division, Federal Polytechnic, Ngodo-Isuochi, Abia State, Nigeria.
Julius Egwu Arua *
Department of Mechatronics Engineering Technology, Akanu Ibiam Federal Polytechnic, Unwana, Ebonyi State, Nigeria.
Igwe Lazarus Uduma
Department of Mechanical Engineering Technology, Akanu Ibiam Federal Polytechnic, Unwana, Ebonyi State, Nigeria.
John Ukanu
Department of Mechanical, Faculty of Engineering, Ahmadu Bello University, Zaria. Kaduna State, Nigeria.
Ali, Uche Egwu
Department of Electrical Electronics Engineering. Akanu Ibiam Federal Polytechnic, Unwana, Afikpo, Ebonyi State, Nigeria.
Ukoima Katoubokmelek Thompson
Department of Mechanical Engineering, Faculty of Engineering, University of Uyo, Akwaibom State, Nigeria.
*Author to whom correspondence should be addressed.
Abstract
The comparative performances of electro-pneumatic clutch actuation system in heavy-duty vehicles using intelligent agent-based control adaptation technique is presented. Conventional control techniques in clutch actuation uses on/off, servo mechanism and other non-intelligent methods of actuation control. These techniques demand for calibration of clutch actuators. To eliminate calibration and its defects, intelligent control methods of clutch actuation are implemented. The specific methodology was predicated on three intelligent agent systems of Fuzzy logic, Neural Network and Hybrid Neuro-Fuzzy. The design started with the development of an intelligent agent-based actuation control rule modelled in a forty-nine fuzzy logic rules pattern for improving the clutch actuation process. A backpropagation standard training algorithm of weight adjustment in a neural network architecture was also designed. The developed fuzzy and neural network models were combined in a hybrid neuro-fuzzy model. Simulink models for Conventional, Fuzzy Logic, ANN and hybrid Neuro-Fuzzy controllers were also developed. Finally, the models were simulated in a Simulink platform and the levels and percentages of improvements determined and compared in order to justify the study. The mean improvements on Conventional controllers compared to that of the Neuro-Fuzzy improved controllers stood at an error reduction in clutch travel from 0.720mm to 0.0244mm given a percentage decrease of 96.6% thereby reducing error to a tolerable level of only 3.39 % while that for torque was increased from 0.1786 NM to 0.4166 NM given a percentage increase of 133.26%. Similarly, increases were recorded for angular speed which also increased from 1005 RPM to 2344 RPM given an increase of 133.23% and power which increased from 16.88 kilowatts to 26.03 kilowatts resulting in 54.21% increase. The fuzzy and ANN controllers also recorded some degrees of improvements over the conventional controllers but to a lesser extent comparatively. It is thus recommended that the use of intelligent agent-based controllers be adopted in the design of clutch actuator system control for its effective improvements.
Keywords: Actuation, calibration, fuzzy-logic, neural network, neuro-fuzzy