The application of evolutionary algorithms could be a rapid and acceptable means of predicting the elemental composition of biomass applicable in power generation. This article introduces a hybrid model comprising of adaptive neuro-fuzzy inference system (ANFIS) and particle swarm optimization (PSO) called PSO-ANFIS model. The model is based on the proximate values of biomass materials to predict their corresponding elemental compositions. This intelligent hybrid model was trained with 581 biomass data and further tested with a new 249 biomass data. The carbon, hydrogen, and oxygen contents of solid biomass were predicted based on the ash content (Ash), fixed carbon (FC), and volatile matter (VM) as the inputs. The model was evaluated based on some known performance criteria. A root mean squared error (RMSE), mean absolute deviation (MAD), mean absolute percentage error (MAPE), log of accuracy ratio (LAR), coefficient of correlation (CC), mean absolute error (MAE) value of 3.5816, 2.3279, 4.8521, 0.0012, 0.9329, 0.0227 at computation time (CT) of 47.04 secs for Carbon (C); 0.6383, 0.4047, 10.2187, 0.002, 0.7986, 0.0261 at CT of 35.2 secs for Hydrogen (H); 4.2042, 2.7417, 10.7612,0.0016, 0.9137 at CT of 28.16 secs for Oxygen (O) respectively was obtained. A regression analysis was also carried out to determine the level of correlation between actual and predicted values. The reported indices show that PSO-ANFIS can be used as a novel computation approach for the prediction of elemental compositions of biomass, which is vital to the combustion, thermochemical, gasification, and pyrolysis process toward energy production.
Read full abstract