APLICA O DE REDES NEURAIS ARTIFICIAIS EM PROCESSO DE DESTILA O DE óLEO ESSENCIAL

In the Amazon region, among different plant species stands Piper aduncum L., rich in essentialoil which it is used as insecticide. Thus, this study aimed to provide a distillation of the products generatedin an agro-industrial system of the species, using Neural Networks (NN). Were collected the followinginput variables: mount of biomass and time distillation and the output variable, “water smell”, to train thenetwork was used the method backpropagation. To reach the efficiency of the network, was determined aminor error and a closer relationship between the estimated data and actual data, such approximation wasevaluated by the r value of correlation. The method showed a closer approximation to the actual values whencompared with another study that also used the same data, but applying the method of multiple regression.