Application of response surface methodology in controller fine-tuning

Despite the developments in advanced control techniques, the traditional PID controller is still being used in the majority of industrial processes. However, due to process non-linearities and modelling difficulties, common tuning techniques often yield tuning parameters that are not optimum. The subsequent fine-tuning stage is time-consuming because it is performed by trial and error. Several researchers have suggested that a statistically designed, experimental approach to controller tuning may be fruitful, e.g. Box and Kramer [1].

Using Response Surface Methodology (RSM), a model of the system performance as a function of the tuning parameters can be obtained. RSM can systematically lead the operator to improved tuning and provide a picture of the sensitivity of the process to the tuning parameters. The application of this technique in the fine-tuning of a simulated and real-time process is shown.