Soft Computing Based Partial-Retuning of Decentralised PI Controller of Nonlinear Multivariable Process

Recent developments in nature-inspired algorithms motivate the control engineers to work towards its application in industrial processes. Almost all the industrial processes are difficult to control since it involves many variables, strong interactions and inherent nonlinearities. In the present work the authors propose cuckoo search, a recent metahuristic algorithm to fine tune the parameters of decentralised PI controller of coal gasifier which is a highly nonlinear multivariable process having strong interactions among the control loops. With the existing controller parameters the response does not able to meet the performance requirements at 0% load for sinusoidal pressure disturbance test. The PI controller for pressure loop is retuned using Cuckoo search algorithm and the best optimal values for its parameters are obtained. Performance of the system with tuned optimal controller settings is evaluated for pressure disturbance test, load change test and coal quality test.