ANFIS based prediction model for biomass heating value using proximate analysis components

This paper proposes a new biomass higher heating value (HHV) prediction model using adaptive neuro-fuzzy inference system (ANFIS) approach. The 444 data related to wide range biomass based materials are composed from the open literature. The input set for the prediction model is involved of the proximate analysis components such as fixed carbon, ash and volatile matter. Three methods called grid partition, sub-clustering and fuzzy c-means are considered in the ANFIS model building process, in order to generate fuzzy inference system (FIS) structure. For determining the best ANFIS based prediction model, a number of simulation studies are performed for each FIS method. The optimal result obtained from each method is compared with each other and the results of the models given in the related literature by prediction performance criteria. The results show that sub-clustering based ANFIS model is the best biomass HHV prediction model. Its obtained coefficient of regression (R2) and root mean square error (RMSE) are 0.8836 and 1.3006, respectively, in the testing phase. As a conclusion, it can be said that the proposed ANFIS based model is an efficient technique to obtain high accuracy biomass HHV prediction.