Application of a multilayer perceptron based on the Levenberg-Marquardt algorithm to odour pattern classification and concentration estimation using odour sensing system

This paper presents the application of multilayer perceptron baed on the Levenberg-Marquardt algorithm to odour pattern classification and concentration estimation using odour sensing system. We adapted the Levenberg-Marquardt (L-M) algorithm having advantages both the steepest descent method and Gauss-Newton method instead of the conventional steepest descent method for the simultaneous classification and concentration estimation of odours. It is confirmed that the multilayer perceptron based on the L-M algorithm is capable of not only the identification of odour classes, but also the prediction of odour concentrations for volatile chemicals with different concentration levels simultaneously throughout the experimental trails.