A learning unidirectional linear response fuzzy controller with golden section search

This study proposes a learning Undirectional Linear Response (ULR) fuzzy controller using a golden section search. This fuzzy controller can be realized with ULR elements which can be implemented simply by a diode or a drain gate connected MOS transistor. This understanding of ULR system is further extended by incorporating the Golden Section Search (GSS) technique of finding the minimum point using the bracket method. We define GSS by evaluating the function at a chosen point in the larger of the two intervals. The midpoint is then replaced and the previous is now chosen to be the new endpoint. This is repeated until the width achieves the desired tolerance and the new minimum point is found. The simulation is carried out on the inverted pendulum problem based on the basic functions of a fuzzy controller that includes membership function, minimum and defuzzification functions of the ULR elements.