Prediction of Higher Heating Value of Solid Biomass Fuels Using Artificial Intelligence Formalisms
- PDF / 569,512 Bytes
- 12 Pages / 595.276 x 790.866 pts Page_size
- 10 Downloads / 159 Views
Prediction of Higher Heating Value of Solid Biomass Fuels Using Artificial Intelligence Formalisms S. B. Ghugare & S. Tiwary & V. Elangovan & S. S. Tambe
# Springer Science+Business Media New York 2013
Abstract The higher heating value (HHV) is an important property defining the energy content of biomass fuels. A number of proximate and/or ultimate analysis based predominantly linear correlations have been proposed for predicting the HHV of biomass fuels. A scrutiny of the relationships between the constituents of the proximate and ultimate analyses and the corresponding HHVs suggests that all relationships are not linear and thus nonlinear models may be more appropriate. Accordingly, a novel artificial intelligence (AI) formalism, namely genetic programming (GP) has been employed for the first time for developing two biomass HHV prediction models, respectively using the constituents of the proximate and ultimate analyses as the model inputs. The prediction and generalization performance of these models was compared rigorously with the corresponding multilayer perceptron (MLP) neural network based as also currently available high-performing linear and nonlinear HHV models. This comparison reveals that the HHV prediction performance of the GP and MLP models is consistently better than that of their existing linear and/or nonlinear counterparts. Specifically, the GP- and MLP-based models exhibit an excellent overall prediction accuracy and generalization performance with high (>0.95) magnitudes of the coefficient of correlation and low (0.97) and test (>0.96) set HHVs computed using GP model I and MLP model I are higher than (0.955, 0.954) that for the model by Parikh et al. [5]. Also, the RMSE magnitude in respect of the training (test) set HHV predictions by the GP model I is significantly lower, that is, by nearly 25 % (13 %) than the corresponding RMSE magnitude of 1.250 (1.127) for the model by Parikh et al. [5]. In this regard, the MLP model I also fares equally better as seen by the corresponding RMSE magnitudes of 0.931 (training set) and 0.996 (test set), which are nearly 25 and 11 % lower than that for Eq. 1. Although by lower margins, the MAPE magnitudes pertaining to both the AI models indicate a trend similar to that exhibited by their RMSE values. It can thus be inferred that the two AI-based models possess excellent biomass HHV prediction accuracies and generalization performance. Also, their performance is better than the currently available high-performing model. A comparative analysis of the HHV prediction accuracies and generalization performance of the GP model I, MLP model I, and Eq. 1 in different ranges of HHVs was also performed (see Table 3). Here, three regions of HHVs are considered: 0–16 MJ/kg (low range), 16–25 MJ/kg (midrange), and 25–35 MJ/kg (high range). A comparison of the
GP model II MLP model II Channiwala and Parikh [11] Milne et al. [12] Friedl et al. [13]
Training set
Test set
CC
RMSE
MAPE
CC
RMSE
MAPE
0.958 0.974 0.891
1.086 0.867 1.881
3.62 3.33 4.99
0.959 0.959
Data Loading...
