Biomass chemical exergy estimation using novel intelligent tool: application to biorefinery sustainability

In response to the growing importance of sustainable energy and environmental considerations in biorefinery technologies, this study introduces a novel hybrid artificial intelligence approach for predicting biomass chemical exergy. The proposed least-squares support vector machine method based on an imperialist competitive algorithm (ICA-LSSVM) combines composition and energy analysis methods to model the chemical energy of biomass accurately. Extensive data were collected to train the model, resulting in a high-precision prediction with a coefficient of determination (R2) of 0.999 and a root mean square error (RMSE) of 30.3. The main point of this study is the development of this accurate model to determine the chemical energy, and also different forms of comparison are used to prove this issue. The sensitivity analysis highlights the direct impact of carbon (C%), hydrogen (H%), and oxygen (O%) percentages on biomass chemical exergy, while nitrogen (N%) and sulfur (S%) exhibit an inverse relationship. The model outperforms existing correlations in the literature, emphasizing its accuracy and practicality for biomass chemical exergy prediction, and contributing to both laboratory and industrial applications. The study emphasizes the model’s real-data reliance and suggests further exploration with an expanded dataset for enhanced industrial applicability.

This article extensively delves into the modeling and prediction of biomass chemical exergy using the ICA-LSSVM approach.
A new sensitivity analysis has been done on the input variables to show their impact on this target.
The study emphasizes the model’s real-data reliance and suggests further exploration with an expanded dataset for enhanced industrial applicability.

This article extensively delves into the modeling and prediction of biomass chemical exergy using the ICA-LSSVM approach.
A new sensitivity analysis has been done on the input variables to show their impact on this target.
The study emphasizes the model’s real-data reliance and suggests further exploration with an expanded dataset for enhanced industrial applicability.