Abstract
Microbial volatile organic compounds (mVOC) are indicators of the mycological load of buildings. It is an unfavourable phenomenon and has a negative impact on the health of residents, as it is associated with the problem of Sick Building Syndrome and buildings affected by it should be subject to repair processes. In order to diagnose the presence of mould fungi in the air of closed rooms, diagnostic tests should be carried out, which are usually time-consuming and require large financial outlays. The basic diagnostic methods include microbiological methods, associated with the cultivation and counting of fungal colonies. The second type of tests includes chemical tests, including gas chromatography, which allows the detection of chemical substances that are markers of the presence of fungi (including mVOC). This article proposes the use of an array of MOS gas sensors (the so-called electronic nose) as a tool for detecting mycological infestation of rooms. The signals obtained from the gas sensor array were then subjected to advanced analysis consisting in the development of predictive models for estimating concentrations. For this purpose, deterministic models based on linear regression without multidimensionality reduction were developed, and then artificial intelligence tools were used - Multilayer Perceptron (MLP) and Support Vector Machine (SVM). The analyses carried out confirmed the effectiveness of artificial intelligence models for predicting fungal marker concentrations and smaller concentration estimation errors compared to deterministic models based on linear regression.