A COMPARATIVE INVISTIGATION OF ANFIS PARAMETERS FOR THE DEVELOPMENT OF AN ADAPTIVE FUZZY LOGIC SPEED CONTROLLERS FOR AC DRIVES

Abstract

Despite the extensive work carried out on developing fuzzy logic controllers (FLC) that is either more intelligent/adaptive or employs fewer rules, there seems to be a lacking of a comprehensive explanation on how the proposed FLC scheme came about as most claimed that it was by trial and error and /or experience. This paper attempts to fill in this gap, by providing an in-depth explanation on how the Adaptive Neuro Fuzzy Inference System (ANFIS) can be used to develop FLC that are not only adaptive, but also use as few as 15 rules. This is achieved by investigating and understanding the ANFIS parameters involved and taking into consideration how each parameter affects the resulting speed control performance of the FLC. The ANFIS parameters studied are the type and range of training data, initial parameters of the FLC architecture and training method that should be used to train a FLC to be adaptive. The simulations were carried out in the MATLAB® environment