Multiple Local Models for System Identification Using Vector Quantization Algorithms.
Luís Gustavo M. Souza, Guilherme A. Barreto · 2010
We introduce a novel method to build multiple local regres- sion models based on the prototype vectors of the SOM network and other well-known vector quantization (VQ) algorithms. The resulting models are evaluated in the task of identifying the inverse dynamics of a heat ex- changer data set. Additionally, we evaluate through statistical hypothesis testing the influence of the VQ algorithm on the performance of the local model. Simulation results demonstrate that the proposed method consis- tently outperforms previous MLP- and SOM-based approaches for system identification.