TY - GEN
T1 - Comparative analysis of the heating values of biomass based on Ga-ANFIS and PsO-ANFIS models
AU - Olatunji, Obafemi
AU - Akinlabi, Stephen
AU - Madushele, Nkosinathi
AU - Adedeji, Paul
AU - Fatoba, Samuel
N1 - Publisher Copyright:
Copyright © 2019 ASME.
PY - 2019
Y1 - 2019
N2 - This article applied a hybridized, adaptive neuro-fuzzy inference system ANFIS-genetic algorithm (GA-ANFIS) and ANFIS -Particle swarm optimization (PSO-ANFIS) to predict the HHV of biomass. The minimum input parameter for the prediction model is based on the proximate values of biomass which are fixed carbon (FC), ash content (A) and volatile matter (VM). The 214 data which cover a wide range of biomass classes were extracted from reliable literature for the training and testing of the models. The optimal results obtained based on each modelling algorithm were compared. The proposed algorithms were evaluated by statistical indices which are the Coefficient of Correlation (CC), Root Mean Squared Error (RMSE), Mean Absolute Percentage Error (MAPE), Mean Absolute Deviation (MAD) estimated at 0.9189, 1.2369,7.4575 and 1.3560 respectively for PSO-ANFIS and 0.9088, 1.1200, 6.3960, 0.8895 respectively for GA-ANFIS. The GA showed exceptional ability to generalize in term of MAPE though at the expense of lesser CC which is obtained in the case of PSO. The reported indices showed that PSO-ANFIS and GA-ANFIS could be applied as an approach to the prediction of HHV based on proximate analysis instead of lengthy experiment procedures.
AB - This article applied a hybridized, adaptive neuro-fuzzy inference system ANFIS-genetic algorithm (GA-ANFIS) and ANFIS -Particle swarm optimization (PSO-ANFIS) to predict the HHV of biomass. The minimum input parameter for the prediction model is based on the proximate values of biomass which are fixed carbon (FC), ash content (A) and volatile matter (VM). The 214 data which cover a wide range of biomass classes were extracted from reliable literature for the training and testing of the models. The optimal results obtained based on each modelling algorithm were compared. The proposed algorithms were evaluated by statistical indices which are the Coefficient of Correlation (CC), Root Mean Squared Error (RMSE), Mean Absolute Percentage Error (MAPE), Mean Absolute Deviation (MAD) estimated at 0.9189, 1.2369,7.4575 and 1.3560 respectively for PSO-ANFIS and 0.9088, 1.1200, 6.3960, 0.8895 respectively for GA-ANFIS. The GA showed exceptional ability to generalize in term of MAPE though at the expense of lesser CC which is obtained in the case of PSO. The reported indices showed that PSO-ANFIS and GA-ANFIS could be applied as an approach to the prediction of HHV based on proximate analysis instead of lengthy experiment procedures.
KW - Biomass
KW - GA-ANFIS
KW - HHV
KW - PSO-ANFIS
KW - Proximate analysis
UR - http://www.scopus.com/inward/record.url?scp=85076423997&partnerID=8YFLogxK
U2 - 10.1115/POWER2019-1825
DO - 10.1115/POWER2019-1825
M3 - Conference contribution
AN - SCOPUS:85076423997
T3 - American Society of Mechanical Engineers, Power Division (Publication) POWER
BT - ASME 2019 Power Conference, POWER 2019
PB - American Society of Mechanical Engineers (ASME)
T2 - ASME 2019 Power Conference, POWER 2019
Y2 - 15 July 2019 through 18 July 2019
ER -