Real-time monitoring of odour concentration at emission sources by IOMS: comparison of different regression models

Thanks to their ability to classify and quantify odours, Instrumental Odour Monitoring Systems (IOMS) are increasingly being used for environmental monitoring. In this context, this work discusses and compares two different regression models implemented in a system installed at the inlet of an abatement system at a WasteWater Treatment Plant (WWTP) with the purpose of continuously monitoring and estimating the odour concentration of the inlet gas stream. The challenging environmental conditions together with the intrinsic uncertainty of Dynamic Olfactometry (DO) required the development of more complex and nonlinear models compared to more common linear chemometrics techniques. The model developed using Support Vector Regression (SVR), which involves some key features and parameters that enable to account for the uncertainty of the reference method, proved capable to provide accurate results and minimize the error in the odour quantification.