Title :
An unscented Kalman filter approach to designing GMDH neural networks: Application to the tunnel furnace
Author :
Luzar, Marcel ; Mrugalski, Marcin ; Witczak, Marcin ; Korbicz, Jozef
Author_Institution :
Inst. of Control & Comput. Eng., Univ. of Zielona Gora, Zielona Gora, Poland
Abstract :
This paper presents an identification method of dynamic systems based on the Group Method of Data Handing. In particular, a new structure of the dynamic neuron in pole representation is proposed. Moreover, a new training algorithm based on the Unscented Kalman Filter is presented. The final part of this work contains an illustrative example regarding the application of the proposed approach to an identification of the tunnel furnace.
Keywords :
Kalman filters; data handling; neural nets; GMDH neural network; dynamic system; group method of data handling; identification method; pole representation; training algorithm; tunnel furnace; unscented Kalman filter; Biological neural networks; Furnaces; Kalman filters; Neurons; Parameter estimation; Temperature sensors; Training;
Conference_Titel :
Methods and Models in Automation and Robotics (MMAR), 2011 16th International Conference on
Conference_Location :
Miedzyzdroje
Print_ISBN :
978-1-4577-0912-8
DOI :
10.1109/MMAR.2011.6031345
