지하역사의 미세먼지 농도 예측을 위한 소프트센서(soft sensor) 모델 개발

An objective of this research is to develop a soft sensor for estimating particulate matter (PM) concentration at underground subway stations. The soft sensor infers difficult-to-measure variable based on information of easy-to-measure variables. To develop the soft sensor, an independent component regression (ICR) is used, where the ICR is an optimal technique to construct a linear regression model by extracting essential information from multivariate data. Experimental results in S-subway station show that the proposed method can improve the performance for estimating PM10 concentration at subway station as compared to the conventional soft sensor which is developed using principal component regression (PCR).