Prediction of influence of stepwise increment of initial acetic acid concentration in charging medium on acetification rate of semi-continuous process by artificial neural network

Abstract

Based on industrial vinegar production, ethanol concentration in charging medium is normally considered as a strong variable influencing the acetification for a given initial acetic acid concentration. Moreover, high initial acetic acid concentration is considered when higher than 100 g L−1 of acetic acid as finished product is obtained. This study assessed the effect of a stepwise increment of initial acetic acid concentration in fermentation medium of 45, 55, and 65 g L−1 after charging at constant ethanol concentration of 35 g L−1 on acetification rate (ETA) by high acid-tolerant strain of Acetobacter aceti WK. Average ETA was 8.144 + 0.09 g L−1 d−1 at 45 g L−1 and 8.655 + 0.09 g L−1 d−1 at 55 g L−1, and significant decreased to 6.819 + 0.23 g L−1 d−1 at 65 g L−1. An artificial neural network (ANN) model was applied to predict the ETA in semi-continuous acetification under the conditions of the study. The optimized ANN structure was revealed to contain two hidden layers and seven neurons per layer. The experimental acetification correlated to the predicted data with R2 of training and testing data set of 0.858 and validation data set of 0.831, respectively. Results indicated that the inputs as acetic acid and ethanol concentrations successfully predicted the ETA of semi-continuous acetification process.