Procedure design of modeling a sterilization reactor with artificial neural network (ANN) using MATLAB

Artificial neural networks (ANNs) are parallel computing systems like biological neural networks. Computation via ANNs is one of the growing scopes of artificial intelligence. ANNs include of large quantity of processing components with their interconnections. ANNs do not require previous science about the nature of the relationships between the data. This is one of the chief advantages of ANNs when compared with most experimental and statistical methods. ANNs can be specified by three components; neurons, weights and transfer functions. Neurons are the fundamental processors of neural networks and are connected to each other by links known as weights. Neurons are commonly arranged in layers; an input layer, an output layer, and one or more hidden layers. In ANNs, every neuron in hidden layer(s) and output layer applies the transfer function to its input. Transfer function calculates output from neurons. In this study, we design and develop procedure of modeling a sterilization reactor with artificial neural network (ANN) using MATLAB. We propose a 3-layer artificial neural network with back propagation training algorithm. Input parameters of network are initial microbial content of microorganism (C), time of process in sterilization reactor (t), temperature of sterilization reactor process (T) and gas flow rate (Q). Network output is sterilization efficiency. Proposed modeled network has 4-X-1 architecture; that X is the neuron numbers of hidden layer. Optimum network is calculated by trial and error method with using MATLAB toolbox. The criteria of selecting optimum network is being highest value of R (coefficient of correlation) and the lowest value of MSE (mean square error) for both training and testing data.

Keywords:

Procedure , Artificial neural network (ANN) , Modeling , Sterilization reactor , Process , MATLAB