This paper describes an electronic nose system (B_NOSE) of low cost, developed through an array of 16 sensors of metal oxide gases (TGS) and signal processing techniques such as PCA, MLP and PNN for the classification of different chemicals compounds. We used a total of three volatile organic compounds (VOCs), of type aromatic hydrocarbons (benzene, toluene and xylene) in order to verify the sensitivity and selectivity of the system to 500, 1000 and 1500 ppm (parts per million). We compared the results with different classification techniques, obtaining a 100% success rate and determining that an electronic nose system is useful for detecting and identifying a large number of pollutants chemicals.