A Design of Soft Sensor Based on Data Fusion

Soft sensor has been commonly used as a valuable alternative to the traditional means for the acquisition of critical quality variables in chemical processes. Many researches and applications of soft sensor have been reported, especially the modeling technique. However, due to the nature of non-linear characteristics of chemical process, it is difficult for a single soft sensor model to identify and provide accurate predictions for the entire dynamic process. The measurements from field instrumentations, such as lab analysis, should be used to compensate the model to improve soft sensor estimations. The objective of this work is to report a novel soft sensor design method, namely data fusion soft sensor, which integrates the soft sensor model estimations with filed measurements by data fusion technique to improve the accuracy and reliability of soft sensor. The performance of the algorithm is evaluated through a lab experiment. The results illustrate the effectiveness of the proposed technique and the potential. The limitations and future directions of research are also outlined.