Prediction of lung nicotine concentration based on novel GA-ANFIS system approach

This paper presents novel GA-ANFIS expert system approach for the prediction of lung nicotine concentration by using "on inspect machine" recorded data. Smoking causes the majority of lung cancers - both in smokers and in people exposed to secondhand smoke. Total yields of cigarette smoke constituents are greatly influenced by smoking behavior. The size of particles in the smoke inhaled directly from a cigarette is controlled by ventilation and it has the largest influence on reducing tar, nicotine and carbon monoxide yields. Our goal is to get better results in prediction of ventilation as well as lung nicotine concentration for different type of cigarette data under the same smoking condition. The data, recorded for different type of cigarettes, are collected by special equipment in real conditions within cigarette factory. GA-ANFIS expert system performs optimization in two steps. In the first step, it generates eight different ANFIS structures, after which it makes second level of optimization using results from ANFIS structure and finally resulting in optimal fuzzy model structures. Validation of the novel GA-ANFIS expert system approach is performed in MATLAB environment by using data set that was not used in the fuzzy model training and testing.