AI-Based New Control Algorithm

This paper introduces a new model following an approach to the design of an artificial intelligence control technique that employs Mamdani type fuzzy logic control. The model following this approach has the inherent advantage of being suitable for practical applications, as it eliminates the requirement to specify a complicated cost function as a figure of merit for the system performance. It enables the designer to successfully pass the second order time domain specifications to the design of fuzzy logic controller. The generation of error signal and its derivatives that are required in the formation of a rule base in fuzzy logic control is facilitated in a very systematic way, thus doing away with the involvement of a human expert. In the proposed scheme, the error between plant and model states is used to compute the error and error derivatives, and a fuzzy law is developed. The proposed technique also brings in a systematic approach to the fuzzy logic control, thus overcoming a lot of heuristics that were in vogue with earlier fuzzy logic applications. The suggested technique is potentially applicable in the field of robotics and aerospace wherein a large number of linear models of process have to be gain scheduled to overcome the uncertainty of parameters. In this paper, a model following the fuzzy logic control has been applied to a second order model of a roll autopilot of a missile. It has been found that the proposed scheme is robust and works satisfactorily even when the parameters are perturbed as much as 50% from their geometric mean value