System and method for on-board data analysis for multi-gas sensing

Abstract

Gas sensors are disclosed having an on-board, low-power data processor that uses multivariable gas classification and/or gas quantitation models to perform on-board data processing to resolve two or more gases in a fluid sample. To reduce computational complexity, the gas sensor utilizes low-power-consumption multivariable data analysis algorithms, inputs from available on-board sensors of ambient conditions, inputs representing contextual data, and/or excitation responses of a gas sensing material to select suitable gas classification and/or gas quantitation models. The data processor can then utilize these gas classification and quantitation models, in combination with measured dielectric responses of a gas sensing material of the gas sensor, to determine classifications and/or concentrations of two or more gases in a fluid sample, while consuming substantially less power than would be consumed if a global comprehensive model were used instead. Thus, the data processor is utilized for linear, nonlinear, and non-monotonic multivariate regressions.