Development of Self-Validating Soft Sensors Using Fast Moving Window Partial Least Squares

Abstract

In the development of soft sensors for an industrial process, the colinearity of the predictor variables and the time-varying nature of the process need to be addressed. In many industrial applications, the partial least-squares (PLS) has been proven to capture the linear relationship between input and output variables for a local operating region; therefore, the PLS model needs to be adapted to accommodate the time-varying nature of the process. In this paper, a fast moving window algorithm is derived to update the PLS model. The proposed approach adapted the parameters of the inferential model with the dissimilarities between the new and oldest data and incorporated them into the kernel algorithm for the PLS. The computational loading of the model adaptation was therefore independent of the window size. In addition, the prediction performance of the model is only dependent on the retained latent variables (LVs) and the window size that can be predetermined from the historical data. Since a moving window approach is sensitive to outliers, the confidence intervals for the primary variables were created based on the prediction uncertainty. The inferential model would not be misled by the outliers from the online analyzers, whereas the model could be updated during the transition stage. The prediction performance of a soft sensor is not only dependent on the capability of the inferential model, but also relies on the data quality of the input measurements. In this paper, the input sensors were validated before performing a prediction. The deterioration of the prediction performance due to the failed sensors was removed by the reconstruction approach. A simulated example of a continuous stirred tank reactor (CSTR) with feedback control systems illustrated that the process characteristics captured by the PLS could be adapted to accommodate a nonlinear process. An industrial example, predicting oxygen concentrations in the air separation process, demonstrated the effectiveness of the proposed approach for the process industry.

Adaptive soft sensor for quality prediction of chemical processes based on selective ensemble of local partial least squares models

“…For comparison purpose, the performance of several state-of-the-art and commonly used PLS based adaptive soft sensing methods was also provided and analyzed. These benchmark methods consist of the recursive PLS (RPLS) [Qin, 1998], the locally weighted PLS (LWPLS) [Kano and Fujiwara, 2013], the moving window PLS (MWPLS) [Liu et al, 2010] and the localized adaptive soft sensor (LASS) [Ni et al, 2014]. The estimation accuracy is evaluated on the test dataset by several commonly used error indexes including the root mean squares error (RMSE), the relative RMSE (RRMSE) and the maximum absolute error (MAE), which are defined as …”

“…These drifts may result from catalyst deactivation, mechanical ageing, change of operating conditions, variation of feed properties, or even climatic change, et al Therefore, developing adaptive soft sensors to adapt them to new process dynamics automatically is necessary for prolonging their life time in industrial applications. Moving window (MW) models [Kaneko et al, 2009;Zhang et al, 2013;Liu et al, 2010], recursive models [Dayal and MacGregor, 1997;Qin, 1998;Tang et al, 2012a;Shao et al, 2012] and just-in-time learning (JITL) models [Chen et al, 2009;Kim et al, 2013b;Liu et al, 2012;Liu and Chen, 2013;Fujiwara et al, 2009] are commonly applied to achieve such target and successful applications of these methods have been reported. However, there are some limitations associated with these methods that need to be analyzed.…”

Online soft sensor design using local partial least squares models with adaptive process state partition

“…These benchmark methods included the least squares SVM (LSSVM) [49] based design, the conventional distance based JITL PLS (JITLPLS) [47], the recursive PLS (RPLS) [19], the moving window PLS (MWPLS) [18] and the CoJIT [21].…”

Virtual Sensing Technology in Process Industries: Trends and Challenges Revealed by Recent Industrial Applications

“…e proposed method was applied to an industrial grinding circuit. Liu et al (2010) proposed fast moving window PLS (FMWPLS) based on the dissimilarities between the new and oldest data that were incorporated into the kernel algorithm for PLS, FMW-PLS was applied to oxygen concentration estimation in an air separation process. proposed the incremental local learning so sensing algorithm (ILLSA) by exploiting the local learning framework and applied it to a polymerization reactor.…”

“…Later, Liu et al (2010) extended this reconstruction-based approach to the case of multiple sensor failures. Since a moving window approach including their FMW-PLS is sensitive to outliers, the con dence intervals of predictions, which were derived from the uncertainties of the output and input variables, were created to avoid inferential models being misled by abrupt noise from the online analyzers.…”