Abstract
Abstract Inferential (or soft) sensors are used in industry to infer the values of imprecisely and rarely measured (or completely unmeasured) variables from variables measured online (e.g., pressures, temperatures). The main challenge, akin to classical model overfitting, in designing an effective inferential sensor is to select a correct structure represented by the number of sensor inputs. This work is focused on the design of an inferential sensor for bottom product composition of an industrial distillation column. We study effectiveness of various subset selection methods that regard different model-overfitting criteria. Our results show that the subset selection is a viable methodology to sensor design and that we are able to improve accuracy of the current refinery sensor by around 15 %.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call