Evaluación Comparativa de Métodos de Sintonización Para Modelamiento de Controladores

Abstract

Abstract: Contex: Controller tuning methods for the identification of models, both first order plus dead time, second order overdamped plus dead time, and second underdamped order plus dead time can express quite high or absolute prediction errors, which do not describe appropriately the controller to identify, due to the implementation of an incorrect method for its identification. Method: This work evaluated three techniques or methods of controller tuning, the Miller modified tangent method, the Smith colon method and the Strejc method, on the other hand using Matlab software and its tool "System Identification "is made the respective model for comparison with the methods exposed for a better evaluation, the absolute prediction error and the quadratic prediction error will be taken as comparison index for their respective comparison. Results: The results support that the most suitable method for the tuning of controllers, compared to the three methods exposed, is the Strejc method of multiple poles due to its low error of absolute and quadratic prediction in comparison with the others. Conclusions: The tuning of controllers with multiple poles is a very good alternative of modeling, which guarantees an adequate follow up to the native function of the controller, due to its multiple poles the controller can present slow responses action.