Comparison between MLP and RBF network in improving CHEMFET sensor selectivity

This paper presents a comparison between two Artificial Neural Network (ANN) approaches, specifically, Multilayer Perceptron (MLP) and Radial Basis Function (RBF) networks, in order to improve selectivity of chemically field-effect transistor (CHEMFET) sensor towards the main ion concentration in mixed solution. MLP and RBF models were developed in Matlab Software. Those models will be able to estimate the main ion in mixed solution by learning the pattern of the input and output based on sensor reading extracted. To validate the architecture of ANN as the optimum model, there are three parameters that will be varied specifically number of hidden neuron, learning rate and momentum. The purposed of parameters optimization is to fit the network outputs to the given inputs. Mean Square Error (MSE) and Regression analysis were used for performance evaluation of the models. The MLP network model showed an absolutely better output than the RBF network model in estimating the main ion concentration in mixed solution.