A GA-based real time controller for the flexible pole-cart balancing problem

• Main Author: Fernandez, Patrick S.
• Format: text
• Language: English
• Published: Animo Repository 1999
• Subjects: Real-time control; On-line data processing; Programming (Electronic computers); Computer Sciences
• Online Access: https://animorepository.dlsu.edu.ph/etd_masteral/2035
• Institution: De La Salle University

A controller for a highly non-linear system is presented in this research. The Flexible Pole-Cart Balancing Problem (FPCBP) is used as the benchmark to investigate the capability of Genetic Algorithms (GAs) in real time control. Over the past five (5) years research on GA-based controller captured the interest of many scientists. However, published results show only off-line application. This research investigates an on-line application of GA-based control system. The controller software is initially trained using a set of data taken from an actual run of a fuzzy logic based Flexible Pole Cart Balancing System (Dadios). The GA-based controller then 'evolves' as it acquires data during its actual run.Prerequisite to the objective of developing a GA-based controller is to speed up the standard GA implementation necessary for real time on-line operation. Instead of strings, real number chromosome representation was used to achieve a high speed GA-based equation fitting program. Results show that the controller developed is able to stabilize the FPCBS at any given initial condition which are within the limits of the system. The experiment also shows that the GA-based controller is able to adapt to changes in the system caused by noise and mechanical inaccuracies of the model.