Low cost green pectin extraction from lemon peels using ultrasound technique: Artificial intelligence – genetic algorithm modeling and characterization

Abstract

The hybrid optimization approach integrating response surface methodology box-behnken design (RSM-BBD) and artificial neural network-genetic algorithm (ANN-GA) was employed to enhance extraction of pectin from lemon peel via ultrasound-assisted extraction techniques. By using these techniques, extraction conditions were improved, and a model was constructed. The extraction time (30–60min), amplitude (25–50%), and pH (1.2–4.2) were considered as input variables, while the output variable was pectin yield (PY%). The results showed that the prediction efficiency of RSM was higher than ANN-GA. With an ANN as the fitness function, the genetic algorithm determined a maximum pectin yield of 38.67% at a pH of 1.2, 30 min extraction time, and 50% amplitude. The experimental pectin extraction yield was found to be 38.92%. At an extraction time of 60 min, with an amplitude of 25% and a pH 4.2, the RSM-predicted value was 39.77%, closely aligning with the actual experimental value of 40.23%. FTIR and SEM were employed for further characterization of the extracted pectin. These analyses demonstrated that RSM has proven to be an effective technique for enhancing the extraction of pectin from lemon peel.