Design and optimisation of a filter based on neural networks. Application to reduce noise in experimental measurement by TGA of thermal degradation of 1-ethyl-3-methylimidazolium ethylsulfate ionic liquid

Abstract

In this paper a new computerised approach to the accurate determination of thermal degradation of 1-ethyl-3-methylimidazolium ethylsulfate ionic liquid is considered. In this approach, an integrated neural network (NN)/thermogravimetric analyser (TGA) is designed to filter a noisy signal from TGA equipment. Data collected (sample weight loss as a function of sample temperature) from the TGA equipment were transferred into a NN trained computer for modeling and filtering of output. Such an integrated NN/TGA filter is capable to filter the noisy TGA output signal based on the created models and patterns, without any previous phenomenological knowledge. The NN/TGA equipment is able to filter the output of the experimental process with mean error less than 0.3%.