TY  - JOUR
AU  - Anyene, Chinedu C. C.
AU  - Nwajinka, Charles O.
AU  - Okpala, Chukwunonso D.
AU  - Amaefule, Desmond O.
AU  - Anonye, Onyemauche F.
AU  - Nwakuba, Nnaemeka R.
PY  - 2026
DA  - 2026/10/05
TI  - Neuro-Predictive Modelling and Thermo-Energetic Optimization of Convective Drying Kinetics and Mass Transfer Dynamics in Black Bean (Akidi) Seeds
JO  - Agricultural Science and Food Processing
T2  - Agricultural Science and Food Processing
JF  - Agricultural Science and Food Processing
VL  - 3
IS  - 4
SP  - 124
EP  - 141
DO  - 10.62762/ASFP.2026.204903
UR  - https://www.icck.org/article/abs/ASFP.2026.204903
KW  - Artificial neural network
KW  - Vigna unguiculata
KW  - black bean
KW  - moisture diffusivity
KW  - drying kinetics
AB  - Convective drying extends the storage life of legume seeds, but its nonlinear kinetics and energy demand are difficult to predict with conventional models. This study developed and validated an artificial neural network (ANN) to predict the drying kinetics, drying time and energy consumption of black bean (\textit{Vigna unguiculata}) seeds during convective drying at 45-65~\degC, and to identify optimal drying conditions. Fresh seeds had an initial moisture content of 18.84% (wet basis), above the 12-14% safe storage limit. A 2-10-10-3 feedforward network (Levenberg-Marquardt training) used drying time and temperature as inputs and moisture ratio (MR), drying rate (DR) and energy consumption (E) as outputs. Increasing the temperature from 45 to 65~\degC{} reduced the drying time from 1100 to 800~min (27.3%), while the effective moisture diffusivity increased from $2.84\times10^{-10}$ to $6.15\times10^{-10}$~m$^{2}$\,s$^{-1}$, with an activation energy of 34.70~kJ\,mol$^{-1}$. On a dry matter basis, protein (25.92-22.92%) and fat (2.26-1.85%) decreased, whereas ash, fibre and carbohydrate increased with temperature. The ANN achieved overall $R^{2}$ of 0.9988-0.9992 and RMSE of 0.0141-0.0190, with maximum prediction errors of 0.61% for MR and 0.40% for DR. ANN-based optimization identified 65~\degC{} as the optimum within the investigated range, reducing the predicted drying time by 25.8% and energy consumption by 27.3% relative to 45~\degC. The model provides a decision-support tool for energy-efficient drying of black bean seeds.
SN  - 3066-1579
PB  - Institute of Central Computation and Knowledge
LA  - English
ER  - 

@article{Anyene2026NeuroPredi,
  author = {Chinedu C. C. Anyene and Charles O. Nwajinka and Chukwunonso D. Okpala and Desmond O. Amaefule and Onyemauche F. Anonye and Nnaemeka R. Nwakuba},
  title = {Neuro-Predictive Modelling and Thermo-Energetic Optimization of Convective Drying Kinetics and Mass Transfer Dynamics in Black Bean (Akidi) Seeds},
  journal = {Agricultural Science and Food Processing},
  year = {2026},
  volume = {3},
  number = {4},
  pages = {124-141},
  doi = {10.62762/ASFP.2026.204903},
  url = {https://www.icck.org/article/abs/ASFP.2026.204903},
  abstract = {Convective drying extends the storage life of legume seeds, but its nonlinear kinetics and energy demand are difficult to predict with conventional models. This study developed and validated an artificial neural network (ANN) to predict the drying kinetics, drying time and energy consumption of black bean (\textit{Vigna unguiculata}) seeds during convective drying at 45-65~\degC, and to identify optimal drying conditions. Fresh seeds had an initial moisture content of 18.84\% (wet basis), above the 12-14\% safe storage limit. A 2-10-10-3 feedforward network (Levenberg-Marquardt training) used drying time and temperature as inputs and moisture ratio (MR), drying rate (DR) and energy consumption (E) as outputs. Increasing the temperature from 45 to 65~\degC{} reduced the drying time from 1100 to 800~min (27.3\%), while the effective moisture diffusivity increased from \$2.84\times10^{-10}\$ to \$6.15\times10^{-10}\$~m\$^{2}\$\,s\$^{-1}\$, with an activation energy of 34.70~kJ\,mol\$^{-1}\$. On a dry matter basis, protein (25.92-22.92\%) and fat (2.26-1.85\%) decreased, whereas ash, fibre and carbohydrate increased with temperature. The ANN achieved overall \$R^{2}\$ of 0.9988-0.9992 and RMSE of 0.0141-0.0190, with maximum prediction errors of 0.61\% for MR and 0.40\% for DR. ANN-based optimization identified 65~\degC{} as the optimum within the investigated range, reducing the predicted drying time by 25.8\% and energy consumption by 27.3\% relative to 45~\degC. The model provides a decision-support tool for energy-efficient drying of black bean seeds.},
  keywords = {Artificial neural network, Vigna unguiculata, black bean, moisture diffusivity, drying kinetics},
  issn = {3066-1579},
  publisher = {Institute of Central Computation and Knowledge}
}