Recursive Least Squares Identification with Extreme Learning Machine (RLS-ELM)

This work investigates the artificial neural networks models in systems identification. The search for process optimization applied to robotics in industry has constantly increased with the course of the computerization of the industry over the years. Identify an automation and control of a process in a manipulator are requirements to have a better quality for the industry final product with a continuous improvement using system identification. To obtain an optimized control in a system identification it is necessary to exist the premise (the output of the model of the system is closer to the real output). This work aims to demonstrate the identification by methods: Least Squares (LS), Recursive Least Square (RLS) and a hybrid model that takes a RLS with Extreme Machine Learning (ELM), applied to the robotic manipulator joint model. The Coefficient of Determination ( \(R^{2}\) ) results are used in the follow identification models: LS, RLS and RLS-ELM.