A joint research group comprising Kasetsart University and MI-6 Ltd. has conducted research on a technique to predict the elemental composition and higher heating value (HHV) of biofuels rapidly and non-destructively using Fourier-transform infrared (FTIR) spectral data combined with deep learning. The results were published in the international journal 'Results in Engineering'.
Accurate determination of carbon, hydrogen, oxygen, and nitrogen (CHON) elemental composition is essential for evaluating the quality and environmental impact of biofuels. However, traditional quantification methods using elemental analyzers or mass spectrometry, while highly accurate, require destructive sample preparation and time-consuming procedures. These limitations present practical bottlenecks that make conventional testing unsuitable for high-throughput screening in large-scale biofuel development programs.
To overcome these difficulties, this study implemented the "AggMap" algorithm to restructure raw 1D FTIR spectra into 2D feature maps that spatially cluster chemically correlated wavenumber regions. This conversion allowed a convolutional neural network (CNN) to effectively capture structural relationships, utilizing a hybrid classification-regression architecture that couples element presence detection with percentage quantification to achieve high predictive accuracy on gas-phase datasets. On the other hand, the verification using liquid fuel samples is currently limited to a small-scale proof-of-concept stage. Expanding the data to cover diverse fuel characteristics and verifying the model's applicability are positioned as the roadmap for future research.
This achievement forms the basis of a digital technology that allows for highly accurate and rapid estimation of the energy characteristics of biofuels directly from spectral data. It is expected to open up the possibility of low-cost and ultra-rapid fuel screening using your FTIR platform and contribute to the efficiency of the initial evaluation process in bioenergy research and development.
Paper Information
- Title: Non-destructive combustion analysis of biofuels through FTIR and deep learning for elemental composition and HHV determination (Link)
- Journal: Results in Engineering
- DOI:https://doi.org/10.1016/j.rineng.2026.110248
- Authors ※Affiliations are based on the information at the time of publication.
- Sivakorn Kanharattanachai, Pongsapak Lueangratana, Thanakrit Yoongsomporn, Elisa Margarita Mendoza Zamarripa, Yuan Weilin, Chia Hsiu Chen, Mitsuru Irie, Chaiyanut Jirayupat (MI-6 Ltd.)
- Napat Kaewtrakulchai (Kasetsart University)
