Implementação do algoritmo dos mínimos quadrados recursivo com fator de esquecimento na estimação de ganhos de um controlador adaptativo por modelo de referência

Abstract

This undergraduate thesis designed and simulated an Model Reference Adaptive Control (MRAC) system using the Recursive Least Squares Method with Forgetting Factor (RLS-FF) for estimation and adjustment of the controller gains, applied to a second-order plant with unknown and varying parameters. Currently, in the field of Control and Automation, the concept of Industry 4.0 has become essential for new factories, and for existing ones, the modernization of the industrial process. In this context, industrial processes require control for safe and adequate operation, but there are varying and even unknown parameters, resulting from partially known dynamics and unmodeled disturbances in the process. Thus, an algorithm was proposed to estimate the gains of the control law of an adaptive controller using the RLS-FF method, as well as the implementation of a modified version with adaptive or variable forgetting factor. Next, the MRAC simulation was performed in an attempt to control the speed of a DC motor modeled using state equations. With this simulation, numerical results were obtained regarding the target parameters of the control system. Through mathematical description and computational simulation, the results were promising, based on the convergence of the controller gains. Therefore, this article analyzed the advantages and limitations of coupling the RLS-FF algorithm in the MRAC system, especially with regards to conventional MRAC techniques.