Izvorni znanstveni članak
Adaptive Neural Network Controller for Thermogenerator Angular Velocity Stabilization System
Krunoslav Horvat
; Brodarski Institute, Zagreb
Ines Šoić
; Brodarski Institute, Zagreb
Ognjen Kuljača
; Brodarski Institute, Zagreb
Sažetak
The paper presents an analytical and simulation approach for the selection of activation functions for the class of neural network controllers for ship’s thermogenerator angular velocity stabilization system. Such systems can be found in many ships. A Lyapunov-like stability analysis is performed in order to obtain a weight update law. A number of simulations were performed to find the best activation function using integral error criteria and statistical T-tests.
Ključne riječi
activation function; adaptive neural network; tracking problem
Hrčak ID:
104494
URI
Datum izdavanja:
30.6.2013.
Posjeta: 928 *