ABSTRACT

One of the major trends driving and driven by technological advancement is increasing use of the sensors and instruments in the world around us. This chapter focuses on multivariate data-driven soft sensors that rely on data modeling techniques. It describes the three most common methods for the development of soft sensors. The principal component analysis (PCA) can be either used as a data preprocessing method or in combination with a regression method as a full soft sensor, in such case the method is referred to as principal component regression. The partial least squares method is another popular method for soft sensing. This method is in particular very useful for adaptive soft sensors. The most common application of soft sensors is the prediction of values that cannot be measured online using automated measurements. This may be for technological reasons (e.g., there is no equipment available for the required measurement), economical reasons (e.g., the necessary equipment is too expensive), etc.