Avaliação de desempenho baseada em variância mínima de controladores para malha de pH

Abstract

This work aims to assess the performace of a nonlinear control system. The operacional anddynamicdetails of a pH control plant aredescribed. The system gain changes nonlinearly on operational conditions, besides that its exhibits sensibility about others parameters like reagent concentration. The control system performance assessment is provided by a ratio of a minimum variance performance lower bound with the actual variance. This method is named as Minimum Variance Index. The minimum variance performance lower bound are hardly described. Basically, depends upon time delay, that is not constant over the operation conditions of the pH system. Nevertheless, having the notion of limitatins, a unique value of time delay is chosen for all operation setpoints of pH control. Before that, the minimum variance is estimated by three dif-ferent methods. The first one is described by (Desborough e Harris, 1992); the second is based on the prediction error defined by difference between system outuput and its predictor. The last one is based on noise determination and on parameters estimation. The methods are compared by experiments changes at configurations like model structure, estimator parameters and disturbance on the system. Results are analized to investigate the influence of nonlinearty at each estimation. Then, this estimation are implement to calculate the Minimum Variance Index. The performance of an adaptive controller and PI controller are compared by the analysis of the index, and finally, theconclusions are discussed.