High temperature sensor array for simultaneous determination of O2, CO, and CO2 with kernel ridge regression data analysis

A sensor array comprising of three chemical gas sensors was evaluated to predict the concentrations of O2, CO, and CO2 in a gas stream with the sensors at 600 °C. The data analysis involved a non-linear multivariate regression method (kernel ridge regression, KRR) along with a searching algorithm to predict gas concentrations. The sensors in the array included a resistance-based 2% CuO/10% La2O3/TiO2 sensor, and two potentiometric sensors, including a yttria stabilized zirconia (YSZ) sensor with a metal/metal oxide internal reference electrode, and a lithium phosphate-based sensor. In addition, the possibility of using the KRR algorithm to predict gas concentrations beyond the training data is explored.