Pectin extraction from Helianthus annuus (sunflower) heads using RSM and ANN modelling by a genetic algorithm approach

In this work, Response Surface Methodology (RSM) and Artificial Neural Network coupled with genetic algorithm (ANN-GA) have been used to develop a model and optimise the conditions for the extraction of pectin from sunflower heads. Input parameters were extraction time (10-20 min), temperature (40-60 degrees C), frequency (30-60 Hz), solid/liquid ratio (S/L) (1:20-1:40 g/mL) while pectin yield (PY%) was the output. Results showed that ANN-GA had a higher prediction efficiency than RSM. Using ANN as the fitness function, a maximum pectin yield of 29.1 +/- 0.07% was searched by genetic algorithm at the time of 10 min, temperature of 59.9 degrees C, frequency of 30 Hz, and solid liquid ratio of 1:29.9 g/mL while the experimental value was found to be 29.5 +/- 0.7%. Extracted pectin was characterised by FTIR and C-13 NMR. Thus, ANN coupled GA has proved to be the effective method for the optimization of process parameters for pectin extraction from sunflower heads. (C) 2018 Elsevier B.V. All rights reserved.