Controle adaptativo utilizando o critério da máxima correntropia aplicado a um sistema de bombeamento de água

Abstract

Given the scarcity of water resources concomitantly with the prospects of urbanization, it has become essential to establish practices that make it possible to reduce losses and improve the water distribution system. Therefore, over the years, several control techniques have been applied in these systems, mainly focusing on the reduction of hydraulic pressure in the ducts, among them the adaptive control. However, the method adopted for its tuning is empirical and exhaustive due to the large number of parameters to be adjusted. Moreover, the criterion adopted by this class of controller is the Mean Squared Error (MSE) for minimizing the tracking error and parametric estimation, however, this criterion is limited to extract information only from the first and second order statistical moments, consequently losing information contained in the higher order statistical moments. Therefore, the present work uses as a criterion for minimizing the tracking error and parameter estimation of a Reference Model Adaptive Controller the Maximum Correntropy with Gaussian kernel. This controller was implemented using Matlab in a LABVIEW environment for pressure control in an experimental water pumping system. The results obtained showed that the controller presents similar characteristics to the control that uses the MSE criterion, for errors close to zero, with a maximum error of 15% during the demand variation, percentage overshoot less than 5% and steady state error less than 2%. However, it is robust to outliers, which does not happen when the MSE criterion is used, this allows for greater reliability in water supply systems that use the controller with the MCC.