Bio-Inspired Optimization to Improve Neural Identifiers for Discrete-Time Nonlinear Systems
Juan F. Guerra, Ramón García-Hernández, Miguel A. Llama · 2021 18th International Conference on Electrical Engineering, Computing Science and Automatic Control (CCE) · 2021
This work aims to apply metaheuristic optimization, offered by bio-inspired algorithms, to enhance the behavior of neural identifiers for unknown nonlinear systems in discrete-time whose model, for this purpose, is assumed unknown. Due to this fact, a new and efficient training algorithm based on the UKF approach is used in combination with the bio-inspired algorithms that are used in order to tune the covariance matrices for the training algorithm. Simulations on a two degree of freedom robot arm are performed obtaining encourage results.