Analysis and implementation of robust algorithms for classification of samples in electronic noses

VOROBIOFF, JUAN; C. A. RINALDI; BOGGIO, NORBERTO G

Ciudad Autonoma de Buenos Aires

Congreso; 15th International Meeting on Chemical Sensors; 2014

Comision Nacional de Energia Atomica

There are many pattern recognition algorithms for classify different samples [1-4]. These algorithms can be combined for data analysis of sensor array. In turn, they can be implemented in different ways and with different parameters. In this work different methods for signal processing for electronic nose are implemented, each of which consist of several algorithms. This methods are analyzed, using our own software in an automated manner, signals of measurements of fish are processed and their results are analyzed for each method. Using cross validation, foreach method correct recognition index (IR) was calculated. After compare IR and extensive testing and analysis of different methods, very robust algorithms were implemented and excellent results were obtained.