Prediction of the effects of process informatics parameters on platinum, palladium, and gold-loaded tin oxide sensors with an artificial neural network

Abstract

In this study, we investigated the effect of the preparation process i.e., the process informatics (PI) parameters of sensor elements on the responses of Pt-, Pd-, and Au-loaded SnO2 sensors. These responses were predicted by an artificial neural network (ANN) using a dataset comprising 441 data points that had been fabricated and evaluated under many parameters in our previous studies. We reported an optimal data preprocessing method based on the relational expression between the sensor response of a semiconductor sensor and the concentration of a target gas and the effect of each PI parameter based on predicted sensor responses under untested conditions.