A Semisupervised Soft-Sensor of Just-in-Time Learning With Structure Entropy Clustering and Applications for Industrial Processes Monitoring

Abstract

To monitor industrial processes properly, softsensors are widely used to predict significant but difficultto-measure quality variables. However, the prediction performances of traditional data-driven soft-sensors are usually unacceptable once suffering from high-nonlinear, high-dimension and imblance data issues. Therefore, a semi-supervised soft-sensor, which is learned by a just-intime method with structure entropy clustering (SS-JITL-SEC), is proposed aiming to improve prediction performance with a simpler way. Inspired by a divide and conquer strategy, a novel SEC method is proposed to achieve several clusters and then to translate the highly complex and nonlinear modeling problems into simple and linear ones. Moreover, the training dataset is extended through a mixed semi-supervised (SS) labeling approach. Finally, dissimilarity-based just-in-time learning (JITL) works together with the resulting clustering sub-datasets to formulate a local adaptive prediction model. Two datasets from different types of wastewater treatment plants are used to verify the effectiveness of the proposed soft-sensor. The results show that the SS-JITL-SEC soft-sensor can achieve better prediction performance than other standard counterparts, and even for effective process monitoring with the resulted residuals.Impact Statement -Proper processes monitoring of difficult-tomeasure quality-related variables is imperative for safe and stable operation of industrial processes, particularly under the case of suffering from significantly dynamic, highly dimensional behaviors during supervised learning. Data-driven soft-sensors together with adaptive learning and semi-supervised learning are currently the alternatives to achieve this goal. The novelty of

Modeling an Adaptive Hybrid Soft Sensor with Co-training Learning toward Applications in Wastewater Treatment

“…Thus, we can observe from eq 7 that the unlabeled data is chosen so that the regression is most consistent with the labeled data set. 21 Then, we put (x u , h 2 (x u )) into L 1 by cross-positioning and where λ(0 < λ ≤ 1) is the forgetting factor and X t and Y t are the newly labeled input and output data, respectively. As the training data change, the prediction model can be updated.…”

“…Thus, we can observe from eq that the unlabeled data is chosen so that the regression is most consistent with the labeled data set …”

“…Thus, we can observe from eq 7 that the unlabeled data is chosen so that the regression is most consistent with the labeled data set. 21 Then, we put (x u , h 2 (x u )) into L 1 by cross-positioning and (x u , h 1 (x u )) into L 2 at the same time.…”

“…Additionally, as the operating conditions change, the prediction model performance usually degrades. To adapt to new conditions, the popular solution is to use the adaptive method, such as recursive models, time differential models, moving window methods, and just-in-time learning . The most widely used method is the recursive model , because this can not only allow the model to capture the information on recent data but also renew the original recursive model with less computational intensity.…”

Development of an Adversarial Transfer Learning-Based Soft Sensor in Industrial Systems

“…However, neither the offline analysis methods nor online measurement devices can meet the real-time prediction requirements totally [4][5][6]. Soft sensor technology provides an alternative way to address this issue [7,8]. Soft sensor modeling methods are mainly categorized into data- driven modeling [9], mechanism modeling [10] and hybrid modeling [11].…”