keywords: Pineapple waste, oxalic acid, ANOVA, ANN, regression analysis
Oxalic acid is one of the important organic acids produced by fermentation and its production is affected by several factors. This study investigated the effect of three independent variables namely; potassium dihydrogen phosphate (KH2PO4), magnesium sulphate (MgSO4) and sodium nitrate (NaNO3) and their mutual interactions on oxalic acid production from pineapple waste using Box Behnken Design (BBD). Modelling was carried out using Response Surface methodology (RSM), Artificial Neural Network (ANN) and Adaptive Neuro Fuzzy Inference System (ANFIS). A quadratic model was obtained to predict the concentration of oxalic acid as a function of the three independent variables. For ANN, Incremental Back Propagation (IBP) with hyperbolic tangent function (Tanh) was the best model for predicting oxalic acid production. For ANFIS, the Sugenoinference system combined with hybrid learning algorithm, Gaussian membership function was found suitable for the prediction of oxalic acid production. The developed RSM, ANN and ANFIS models described the fermentation with high accuracy as indicated by their high R2 values (0.957, 0.9894 and 0.9893), low RMSE (1.0923, 0.5417 and 0.5422) and low AAD (7.8692, 1.1887 and 1.3130), respectively. RSM, ANN and ANFIS coupled with genetic algorithm were applied to optimize the process for best operating condition and ANN gave the maximum value of oxalic acid (20.73 g/L) with the best combination of the input variables (0.77 g/L of KH2PO4, 0.09 g/L of MgSO4 and 1.78 g/L of NaNO3). Based on the statistical indices used for evaluation, ANN performed slightly better than ANFIS-GA and both were better than RSM. RSM performed the least.