Improved model of halophilic cellulase fermentation from lignocellulosic biomass using artificial neural network (ANN)
An artificial neural network (ANN) was employed to model the fermentation process for halophilic cellulase production from lignocellulosic biomass. Three critical parameters within the ANN architecture; weight decay, number of iterations, and the number of hidden neurons were optimized to achieve th...
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | en |
| Published: |
Taylor’s University
2025
|
| Subjects: | |
| Online Access: | https://eprints.ums.edu.my/id/eprint/45727/1/FULLTEXT.pdf https://eprints.ums.edu.my/id/eprint/45727/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | An artificial neural network (ANN) was employed to model the fermentation process for halophilic cellulase production from lignocellulosic biomass. Three critical parameters within the ANN architecture; weight decay, number of iterations, and the number of hidden neurons were optimized to achieve the most accurate prediction. The model was designed to predict halophilic cellulase activity as the output, based on three manipulated input variables: NaCl (sodium chloride), CMC (carboxymethylcellulose), and FeSO₄·7H₂O (iron (II) sulphate heptahydrate) concentrations. A total of 19 experimental datasets were used to train the ANN model. The ANN model's performance was compared with that of a multivariable regression model. The highest cellulase activity of 0.0602 U/mL was achieved under optimal medium conditions determined by the ANN: 9.05% NaCl, 1.50% CMC, and 2.66×10⁻⁴% (w/v) FeSO₄·7H₂O. The ANN model demonstrated improved accuracy in estimating halophilic cellulase activity, reducing the RMSE from 4.4×10⁻³ to 3.8×10⁻³ when compared with the multivariable regression model. Moreover, the ANN outperformed the response surface methodology (RSM), reducing prediction error from 3.19-5.56% (RSM) to 0.84-2.36% (ANN). The dynamic adaptability of the ANN architecture enabled it to better capture the complexities of the fermentation process, which involves living microorganisms. ANN models enhance enzyme production efficiency and promote sustainable biotechnology by optimizing fermentation processes using renewable lignocellulosic biomass. |
|---|