Estimation of biomass higher heating value (HHV) based on the proximate analysis by using iterative neural network-adapted partial least squares (INNPLS)

The higher heating value (HHV) of biomass fuels is a crucial factor in the techno-economic analysis and subsequent development of bioenergy projects. In this study, iterative neural network-adapted partial least squares (INNPLS) was applied to estimate the HHV of biomass fuels as a function of fixed carbon (FC), volatile matters (VM), and ash content. The ANN paradigm was used to correlate the inputs and the outputs of PLS score vectors thorough iterative training procedure. The prediction capability of the proposed model was compared with those of the classical PLS, the coupled principle component analysis and ANN paradigm (PCA-ANN), and the neural network-adapted partial least squares (NNPLS). The presented models were developed, trained, and tested using 350 data points obtained from the published literature. According to the results obtained, the INNPLS showed an excellent capability to model the HHV of biomass fuels over the other methods. This approach was then embedded into a simple and user-friendly software for estimating the HHV of biomass fuels on the basis of their proximate data. The developed software can be utilized for reliable and accurate estimation of biomass HHV based on only three input parameters as an alternative to the lengthy and costly laboratorial measurements. (C) 2017 Elsevier Ltd. All rights reserved.