Double-Level Locally Weighted Extreme Learning Machine for Soft Sensor Modeling of Complex Nonlinear Industrial Processes

Zhang, Xiangrui; Deng, Xiaogang; Wang, Ping

doi:10.1109/jsen.2020.3018716

Cited by 27 publications

(7 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…According to the equations above, estimating the probability density functions (PDFs) is necessary to compute MI values. Both parametric and non-parametric procedures can calculate the PDFs, but the PDF estimation is not a simple task in practical applications [ 54 ]. K-nearest neighbor (K-NN) non-parametric method to calculate MI was proposed in [ 51 ].…”

Section: Preliminaresmentioning

confidence: 99%

Industrial Semi-Supervised Dynamic Soft-Sensor Modeling Approach Based on Deep Relevant Representation Learning

Lima

Araújo

2021

Sensors

View full text Add to dashboard Cite

Soft sensors based on deep learning have been growing in industrial process applications, inferring hard-to-measure but crucial quality-related variables. However, applications may present strong non-linearity, dynamicity, and a lack of labeled data. To deal with the above-cited problems, the extraction of relevant features is becoming a field of interest in soft-sensing. A novel deep representative learning soft-sensor modeling approach is proposed based on stacked autoencoder (SAE), mutual information (MI), and long-short term memory (LSTM). SAE is trained layer by layer with MI evaluation performed between extracted features and targeted output to evaluate the relevance of learned representation in each layer. This approach highlights relevant information and eliminates irrelevant information from the current layer. Thus, deep output-related representative features are retrieved. In the supervised fine-tuning stage, an LSTM is coupled to the tail of the SAE to address system inherent dynamic behavior. Also, a k-fold cross-validation ensemble strategy is applied to enhance the soft-sensor reliability. Two real-world industrial non-linear processes are employed to evaluate the proposed method performance. The obtained results show improved prediction performance in comparison to other traditional and state-of-art methods. Compared to the other methods, the proposed model can generate more than 38.6% and 39.4% improvement of RMSE for the two analyzed industrial cases.

show abstract

Section: Preliminaresmentioning

confidence: 99%

Industrial Semi-Supervised Dynamic Soft-Sensor Modeling Approach Based on Deep Relevant Representation Learning

Lima

Araújo

2021

Sensors

View full text Add to dashboard Cite

show abstract

“…based on advanced machine learning were further witnessed, such as support vector regression (SVR) [10], Extreme Learning Machine (ELM) [11]. For example, Zhang et al [12] proposed a Double-level Locally Weighted Extreme Learning Machine (DLWELM) based soft sensor for online prediction of quality variables, which shows better performance compared to conventional statistical methods.…”

Section: Introductionmentioning

confidence: 99%

Modeling Task Relationships in Multivariate Soft Sensor With Balanced Mixture-of-Experts

Huang

Wang

Liu

et al. 2023

IEEE Trans. Ind. Inf.

View full text Add to dashboard Cite

Accurate estimation of multiple quality variables is critical for building industrial soft sensor models, which have long been confronted with data efficiency and negative transfer issues. Methods sharing backbone parameters among tasks address the data efficiency issue; however, they still fail to mitigate the negative transfer problem. To address this issue, a balanced Mixture-of-Experts (BMoE) is proposed in this work, which consists of a multi-gate mixture of experts (MMoE) module and a task gradient balancing (TGB) module. The MoE module aims to portray task relationships, while the TGB module balances the gradients among tasks dynamically. Both of them cooperate to mitigate the negative transfer problem. Experiments on the typical sulfur recovery unit demonstrate that BMoE models task relationship and balances the training process effectively, and achieves better performance than baseline models significantly.

show abstract

“…By absorbing the similar labeled and the unlabeled data into prediction model, Zheng et al 25 proposed a JITL semi‐supervised ELM modeling framework to predict the Mooney viscosity of industrial rubber mixing process. Zhang et al 26 studied double‐level similarity based local weighted JITL‐ELM soft sensor strategy for nonlinear industrial process.…”

Section: Introductionmentioning

confidence: 99%

“…Currently, most of the reported local dynamic JITL-ELM methods are used to nonlinear processes modeling. [24][25][26][27][28] To the author's knowledge, the related reports about the local dynamic JITL-ELM model based optimal ILC control methods for nonlinear batch process have not previously been published. The main reason is that updating a local dynamic JITL-ELM model is more computationally expensive than updating a fixed global model, which leads to the control performance of optimal ILC strategy degradation because of the real-time problem of the JITL-ELM model.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A local dynamic extreme learning machine based iterative learning control of nonlinear batch process

Zhou

Jia

2021

Optim Control Appl Methods

View full text Add to dashboard Cite

This article deals with the optimal control issue of nonlinear batch process. First, in order to derive high efficiency and accuracy process model, a novel hierarchical searching mechanism local dynamic nonlinear model is constructed which is composed of just‐in‐time learning and extreme learning machine (JITL‐ELM). Then, based on the local dynamic JITL‐ELM model, an optimal quadratic‐criterion‐based iterative learning control (Q‐ILC) algorithm is presented, where the control input trajectory can be obtained by solving a quadratic programming problem. Moreover, on the basis of inverse model system, the initial batch control input trajectory of the Q‐ILC algorithm can be obtained by the use of JITL method. As a result, not only the issue of model‐plant mismatch and real‐time disturbance can be solved, but also obtain faster system convergence rate and smaller tracking error. Besides, the convergence properties of control input and tracking error are analyzed. Finally, a typical batch process is presented to demonstrate the feasibility and superiority.

show abstract

Double-Level Locally Weighted Extreme Learning Machine for Soft Sensor Modeling of Complex Nonlinear Industrial Processes

Cited by 27 publications

References 28 publications

Industrial Semi-Supervised Dynamic Soft-Sensor Modeling Approach Based on Deep Relevant Representation Learning

Industrial Semi-Supervised Dynamic Soft-Sensor Modeling Approach Based on Deep Relevant Representation Learning

Modeling Task Relationships in Multivariate Soft Sensor With Balanced Mixture-of-Experts

A local dynamic extreme learning machine based iterative learning control of nonlinear batch process

Contact Info

Product

Resources

About