Document Type : Original Article
Education and Pedagogical Sciences Department, Taras Shevchenko National University of Kyiv, Ukraine
In the present work, the multi-layer perceptron neural network is applied to model biomass gasification in the fixed-bed downdraft gasifier. Therefore, the multi-layer perceptron neural network is implemented to analyze and predict the gas composition in the outlet flow of the gasifier concerning the CH_4, CO_2, H_2 and CO concentrations. On the other hand, the input data for the prediction includes biomass element content (C,H, and O), the value of the ash and moisture contents, and the temperature of the reduction zone. Extensive values which are derived from the experimental data are used to train the Multi-Layer Perceptron Neural Network. The obtained results from the model prediction show a satisfying agreement with the empirical data. The result of statistical analysis in the case of R^2 values for CH_4 and CO is higher than 0.99, and for the CO_2 and H_2 is higher than 0.98, which shows a good agreement between the experimental and predicted data. Also, a comparative study between MLP and other well-known methods demonstrates the superiority of MLP for gasification yield prediction. Hence, this model can be a useful tool for the analysis and performance evaluation of the gasifier modules.