Soft-sensor design via task transferred just-in-time-learning coupled transductive moving window learner

Alakent, Burak

doi:10.1016/j.jprocont.2021.03.006

Cited by 27 publications

(15 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In such a situation, a large measurement delay occurs, which is not beneficial to process quality control and optimization. Alternatively, soft sensing methods are developed to alleviate the problem. − The concerned hard-to-measure quality variables are estimated with the help of process variables. As known, adequate training samples are a key factor in reliable model construction.…”

Section: Introductionmentioning

confidence: 99%

Adversarial Transferred Data-Assisted Soft Sensor for Enhanced Multigrade Quality Prediction

et al. 2023

View full text Add to dashboard Cite

Although recent transfer learning soft sensors show promising applications in multigrade chemical processes, good prediction performance mainly relies on available target domain data, which is difficult to achieve for a start-up grade. Additionally, only employing a single global model is inadequate to characterize the inner relationship of process variables. A just-in-time adversarial transfer learning (JATL) soft sensing method is developed to enhance multigrade process prediction performance. The distribution discrepancies of process variables between two different operating grades are first reduced by the ATL strategy. Subsequently, by applying the just-in-time learning approach, a similar data set is selected from the transferred source data for reliable model construction. Consequently, with the JATL-based soft sensor, quality prediction of a new target grade is implemented without its own labeled data. Experimental results on two multigrade chemical processes validate that the JATL method can give rise to the improvement of model performance.

show abstract

Section: Introductionmentioning

confidence: 99%

Adversarial Transferred Data-Assisted Soft Sensor for Enhanced Multigrade Quality Prediction

et al. 2023

View full text Add to dashboard Cite

show abstract

“…Liu et al (2019) proposed a soft sensor based on domain adaptive extreme learning machine (DAELM), which solved the problem of limited data in multi‐grade chemical processes. Alakent (2021) developed a data‐driven method for online prediction of quality variables based on transfer learning by combining the JITL model with the MW learner. Liu et al (2021) displayed a general model training framework based on transfer learning for industrial Internet of Things systems, which can cut down training time and improve accuracy.…”

Section: Introductionmentioning

confidence: 99%

Hybrid data‐driven modeling for accurate prediction of quality parameters in sugarcane milling with fine‐tuning

Yang,

Mao,

Yin

et al. 2023

J Food Process Engineering

View full text Add to dashboard Cite

The data‐driven model for predicting sugarcane milling quality parameters is crucial for optimization. Current challenges include real‐time data acquisition and limited labeled data, reducing model accuracy and raising data collection costs. There is an urgent need to utilize unlabeled data to establish data‐driven models for sugarcane milling. This study introduces novel hybrid strategies, namely combining Convolutional Neural Network (CNN) with Kernel Extreme Learning Machine (KELM), utilizing pre‐training and fine‐tuning techniques. These strategies include CNN‐KELM and Transfer Learning CNN‐KELM (TL‐CNN‐KELM). Results demonstrate superior accuracy compared to single algorithms. TL‐CNN‐KELM, leveraging unlabeled data through transfer learning, achieves reductions of 0.007 in mean squared error and 0.005 in mean absolute error for extraction rate predictions. For initial juice sugar content predictions, reductions of 0.027 in mean squared error and 0.012 in mean absolute error are attained. This approach demonstrates enhanced accuracy in predicting quality parameters while maintaining robustness to data size.Practical ApplicationsThis work can reduce the data acquisition cost for quality parameter prediction and accelerate the intellectualization of sugarcane milling. In production practice, there are many process industries similar to sugarcane milling, so the novel algorithms also provide an idea for the quality parameter optimization and prediction of other process industries.

show abstract

“…6 In addition, Alakent proposed the combination of a moving window (MW) 6 and just-in-time learning (JIT) 7 using transductive inference, 8 while they later proposed a combination of a task-transferred JIT model with an MW learner in a transductive learning setting. 9 Negative transfer, 10 which is a phenomenon in which the prediction accuracy is reduced by transferring information, should be considered in TL. More specifically, the negative transfer is an important issue in TL and is considered to occur when the SD is completely unrelated to the TD, and when training in SD interferes with training in TD.…”

Section: Introductionmentioning

confidence: 99%

“…developed a domain‐adaptation extreme learning machine to construct soft sensors for multiple processes, wherein only small samples are produced or available 6 . In addition, Alakent proposed the combination of a moving window (MW) 6 and just‐in‐time learning (JIT) 7 using transductive inference, 8 while they later proposed a combination of a task‐transferred JIT model with an MW learner in a transductive learning setting 9 …”

Section: Introductionmentioning

confidence: 99%

Adaptive soft sensor based on transfer learning and ensemble learning for multiple process states

Yamada

Kaneko

2022

Analytical Science Advances

View full text Add to dashboard Cite

The objective of this study is to develop an adaptive software sensor technique that can predict objective process variables for a target grade in a plant while also considering information related to various other grades. We use a dataset of the target grade as the target domain and those of the other grades as source domains to perform transfer learning. Multiple models or sub-models are constructed by setting a source domain for each grade and changing the number of samples used as the source domain. Furthermore, to prevent the negative transfer, the use of a source domain is automatically judged. In this study, we constructed sub-models using the locally weighted partial least squares approach as an adaptive soft sensor technique. The values of an objective variable were predicted with ensemble learning using sub-models. The effectiveness of the proposed method was verified using a dataset measured in an actual incineration plant, and the proposed method was able to accurately predict the product quality even when the plant was operated in five grades and when a new grade was produced.

show abstract

Soft-sensor design via task transferred just-in-time-learning coupled transductive moving window learner

Cited by 27 publications

References 32 publications

Adversarial Transferred Data-Assisted Soft Sensor for Enhanced Multigrade Quality Prediction

Adversarial Transferred Data-Assisted Soft Sensor for Enhanced Multigrade Quality Prediction

Hybrid data‐driven modeling for accurate prediction of quality parameters in sugarcane milling with fine‐tuning

Adaptive soft sensor based on transfer learning and ensemble learning for multiple process states

Contact Info

Product

Resources

About