Adaptive soft sensor model using online support vector regression with time variable and discussion of appropriate hyperparameter settings and window size

• Published in: Computers & chemical engineering Vol. 58; pp. 288 - 297
• Main Authors: Kaneko, Hiromasa, Funatsu, Kimito
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 11.11.2013; Elsevier
• Subjects: Applied sciences; Computer science; control theory; systems; Control theory. Systems; Data processing. List processing. Character string processing; Degradation; Exact sciences and technology; Hyperparameter; Memory organisation. Data processing; Modelling and identification; Online support vector machine; Process control; Process control. Computer integrated manufacturing; Soft sensor; Software; Time variable; Degradation; Online support vector machine; Time variable; Hyperparameter; Soft sensor; Process control; Data analysis; Chemical plant; Measurement sensor; High precision; Change of state; Rapid process; Regression analysis; Modeling; Adaptive method; Time response; Variable transformation; Drop; Time varying system; Chemical reactor; Vector support machine; Data fusion; Data models; System identification
• ISSN: 0098-1354; 1873-4375
• DOI: 10.1016/j.compchemeng.2013.07.016

•Soft sensors are widely used to predict process variables in chemical plants.•Our goal is to achieve high prediction accuracy of soft sensors for new data.•We employ the online support vector regression (OSVR) and time variable.•The hyperparameters and the window size of the OSVR model were discussed.•The performance was confirmed with simulation data and real industrial data. Soft sensors have been widely used in chemical plants to estimate process variables that are difficult to measure online. One crucial difficulty of soft sensors is that predictive accuracy drops due to changes in state of chemical plants. The predictive accuracy of traditional soft sensor models decreases when sudden process changes occur. However, an online support vector regression (OSVR) model with the time variable can adapt to rapid changes among process variables. One crucial problem is finding appropriate hyperparameters and window size, which means the numbers of data for the model construction, and thus, we discussed three methods to select hyperparameters based on predictive accuracy and computation time. The window size of the proposed method was discussed through simulation data and real industrial data analyses and the proposed method achieved high predictive accuracy when time-varying changes in process characteristics occurred.