"ANFIS" Neuro-Fuzzy Modeling of a Pneumatic Leak Testing System

The Neuro-Fuzzy modeling research of a pneumatic leak testing system based on the construction of a series of models built from data generated using a mathematical model of the system is presented. Sub-problems related with fuzzy modeling such as structure specification, parameter estimation and model evaluation were analyzed. Many alternatives were tested and compared. The model selected can predict the internal pressure in the object to be tested with a little RMS error and achieve an optimal balance between prediction accuracy and model complexity, based on statistical information criteria.