Artificial intelligence-based modeling for prediction of biogas production rate from full-scale anaerobic sludge digestion reactors: Neural networks and fuzzy logic applications

In this study, artificial intellingence-based prognostic methods (i.e. artificial neural networks (ANN) and fuzzy-logic) and a multiple regression-based analysis were conducted to predict biogas production rate from a full-scale sludge digestion process. In the scope of the present work, five process-related parameters such as influent sludge flow rate, total solids (TS) content, total volatile solid (VS) content, alkalinity and volatile fatty acids (VFA) concentration were considered as the independent input variables of the proposed models.In the first part of this study, an ANN-based modeling study was performed by using MATLAB® V7.9.0.529 (R2009b, License No: 161051) software. Following to pricipal component analysis (PCA), a benchmark comparison of 11 back-propagation (BP) algorithms was employed by means of their respective mean squared errors (MSE), and scaled conjugate gradient algorithm was found as the best of 11 BP algorithms. According to the selected training algorithm (trainscg), the number of neurons at the hidden layer was optimized as 14, and the coefficient of determination (R2) was obtained as 0.6496 for the optimal three-layer ANN structure (5:14:1).In the second part of the present study, a fuzzy-logic-based methodology was conducted to predict biogas production rate from the full-scale sludge digestion process. In this part, a "Fuzzy Logic Module" (Fuzzy Logic Toolbox) was created by using a "Fuzzy Inference System" (FIS) within the framework of MATLAB® V7.9.0.529 software. For this purpose, a MISO (multiple inputs single output) type fuzzy-logic model was developed and five input variables were fuzzified in a knowledge-based manner. Trapezoidal membership functions with ten and twenty levels were implemented for the fuzzy subsets, and a Mamdani-type fuzzy inference system was used to implement a total of 394 rules in IF-THEN format. The most widely used methods in the literature, prod, max, prod, sum, centroid, were employed as the inference operators for IMPLICATION, AGGREGATION and DEFUZZIFICATION methods conducted in this study.Fuzzy logic predicted results were compared with the outputs of both ANN-based model and a polynomial multiple regression-based model (the third part of the present work) derived in this study. Findings of this study clearly indicated that, compared to ANN model and conventional multiple regression approach, the proposed MISO fuzzy-logic-based model produced smaller deviations and exhibited a superior predictive performance on forecasting both COD removal efficiency and biogas production rates with satisfactory determination coefficient (R2) about 0.8765.