Correlation-based Just-In-Time modeling for soft-sensor design

Abstract

Abstract Soft-sensors are widely used for estimating product quality or other key variables when on-line analyzers are not available. However their estimation performance deteriorates when the process characteristics change. To cope with such changes and update the model, recursive methods such as recursive PLS and Just-In-Time (JIT) modeling have been developed. When process characteristics change abruptly, however, they do not always function well. In the present work, a new method for constructing soft-sensors based on a JIT modeling technique is proposed. In the proposed method, referred to as correlation-based JIT modeling, the samples used for local modeling are selected on the basis of the correlation among variables instead of or together with distance. The proposed method can adapt a model to changes in process characteristics and also cope with process nonlinearity. The superiority of the proposed method over the conventional methods is demonstrated through a case study of a CSTR process in which catalyst deactivation and recovery are considered as changes in process characteristics.