Kalman filtering approach to multirate information fusion for soft sensor development

Accurate and frequent measurements of quality variables are important for real-time process monitoring and control. However, because online measuring instruments commonly have the limitations of high investment and low accuracy, while offline laboratory analyses are obtained manually and infrequently, neither online instruments nor offline analyses can satisfy the requirements of real-time applications in industries alone. In order to obtain more reliable information of quality variables, this paper develops two kinds of adaptive soft sensors in the framework of Kalman filter. The idea is to take the advantages of fast-rate sampling of online data and high-accuracy of lab data by synthesizing these two sources of measurements at different sampling rates. The CSTR case study and applications in a chemical production process demonstrate the effectiveness of the proposed methods.