Optimization of a gas turbine engine fuzzy control

Modern techniques are required for many complex systems where increasingly strict performance and regulatory requirements must be achieved. In this paper, a genetic-fuzzy control is employed in order to meet the engine performance requirement and constraints. Knowledge and understanding of the system behavior is an essential factor in control system development. Particularly, this factor is more serious for defining fuzzy membership functions and developing fuzzy rules which both are based on knowledge and experience of the controller designer. First, a successful approach based on Artificial Neural Networks (ANNs) is used for simulating the engine performance in a reversed manner, and achieving a set of suitable fuel flow functions. The fuzzified forms of these functions are used to produce first generation of chromosomes (fuzzy fuel functions). Indeed, the method of neural networks is used as an effective idea to define first generation of fuzzy fuel functions. Subsequently, tuning of the FLC parameters is obtained based on an optimization problem solved by a multi-objective genetic algorithm (GA). Finally, a fuzzy logic controller is developed using the engine simulation model, and the GA results. Simulation results of the optimized fuzzy controller, simple fuzzy controller and a conventional controller are presented to demonstrate the capability of proposed optimized fuzzy controller in achieving a fast engine response with consideration of all safety constraints.