Finite Impulse Response Type Multilayer Perceptron Artificial Neural Network Model For Bacteria Growth Modeling Inhibited by Lemon Basil Waste Extract

Abstract

Abstract The tools to predict the growth of bacteria over the time is essential to maintain the process stability in bio processes. Currently, not all tools have been fully used to fulfil these interests which can be applied in industry and laboratory. In this paper, a mathematical modelling approach based on the type of multi layer perceptron artificial neural network created by Finite Impulse Response (FIR) is proposed. The neural network model was developed using data collected from laboratory work. A total of 75% the growth of bacteria (S. Aureus, B. Cereus and S. Typhimurium) which is inhibited by lemon basil waste extract, over the time data are used to train Artificial Neural Network (ANN), and the rest of the data are used to validate the model. ANN has been model the growth of S. Aureus, B. Cereus and S. Typhimurium which is inhibited by lemon basil waste extract over the time. Mean Square Error (MSE) results during training and validation obtained from this modeling were 0.087 and 0.147, respectively. It means the mathematical modeling approach used in this study is suitable for capturing nonlinear characteristics of bacterial growth that is inhibited by lemon basil waste extract.