Abstract
The application of Instrumental Odour Monitoring Systems (IOMS) for environmental odour monitoring has become increasingly relevant in recent years, thanks to their capability to operate continuously, allowing near real-time discrimination of odour events generated by industrial plants, which can cause annoyance to nearby populations. This study refers to the application of an IOMS for monitoring odour emissions at the fenceline of a Waste Treatment Plant (WTP). The main focus is the development of appropriate data processing approaches applied to the sensor signals recorded by the IOMS during operation, in order to counteract interference related to variations in ambient humidity and drift effects. The study was conducted over a period of more than one year, during which dedicated olfactometric campaigns were carried out to train and test the IOMS odour classification and regression models. The results show that the application of specific data pre-processing strategies effectively reduce the impact of interferents on IOMS outputs, leading to improved model performance: acceptable IOMS performance is maintained even though the system is affected by drift effects. Conversely, models developed without proper compensation for humidity and drift exhibit significantly reduced performance.
