Levy guided Firefly Algorithm based tuning of decentralised PI Controller of Nonlinear Multivariable system - Coal Gasifier

Coal gasifier, an important and primary unit in Integrated Gasification Combined Cycle, converts coal into coal gas. Control of gasifier is highly difficult since it involves many complicated reactions, nonlinear behavior, interactions among control loops and operating constraints. The present work deals with tuning of decentralized PI controller of coal gasifier system so as to meet the performance requirements at all operating points and also for all disturbances. Firefly algorithm, a recently developed metaheuristic approach is used to tune the pressure loop PI controller. Here, a random walk via Levy flight is considered for exploring the search space. Controller parameters at base load are taken as the decision variables while the Maximum Absolute Error for pressure is taken as the objective function. Firefly algorithm determines the optimum settings for the PI controller. The tuned PI controller satisfies all the performance requirements for pressure disturbance and coal quality variations at 0%, 50% and 100% operating points and smooth tracking of outputs for load change is observed. Hence, Levy guided Firefly algorithm can be effectively used to get optimum response from a highly nonlinear multivariable process.