An approach to model building for accelerated cooling process using instance-based learning

Authors:
Yi Zheng;Shaoyuan Li;Xiaobo Wang
Affiliations:
Department of Automation, Shanghai Jiao Tong University, 800 Dong Chuan Road, Shanghai 200240, China;Department of Automation, Shanghai Jiao Tong University, 800 Dong Chuan Road, Shanghai 200240, China;Institute of Research and Development, Baosteel Iron and Steel Ltd., Co., 600 Fu Jin Road, Shanghai 201900, China
Venue:
Expert Systems with Applications: An International Journal
Year:
2010

Citing 8
Cited 0

Matrix multiplication via arithmetic progressions

Journal of Symbolic Computation - Special issue on computational algebraic complexity
Trading MIPS and memory for knowledge engineering

Communications of the ACM
A Review and Empirical Evaluation of Feature Weighting Methods for aClass of Lazy Learning Algorithms

Artificial Intelligence Review - Special issue on lazy learning
A clustering-based method for unsupervised intrusion detections

Pattern Recognition Letters
Face recognition based on discriminant fractional Fourier feature extraction

Pattern Recognition Letters
Face recognition with local steerable phase feature

Pattern Recognition Letters
Locally linear reconstruction for instance-based learning

Pattern Recognition
Applying instance-based techniques to prediction of final outcome in acute stroke

Artificial Intelligence in Medicine

Quantified Score

Hi-index	12.05

Visualization

Abstract

Precise mathematical modelling of accelerated cooling process (ACC) is not accessible or worthwhile due to the various compositions and gauges of plates and its high system specificity. An online modelling method for ACC is developed through combining instance-based learning (IBL) with the physical dynamical process model. When a plate comes, k plates whose material composition and operating conditions are nearest to that of current plate are selected from historical instances. Then, an approach based on locally linear reconstruction is extended to be suitable for MIMO system first, and is applied to structure the parameters of current plate's dynamical model according to the selected k plates, due to that LLR could be able to preserve locally linear topology surrounding the new pattern and it is robust to the number of historical data. To guarantee the accurate of historical instance, a correction method is developed to modify the parameters of current plate when the cooling process ends. The resulting model can predict the temperature evolutions of the moving plates with various compositions and gauges during the cooling process. Experimental studies with real industrial data in one steel company show the effectiveness of the proposed modelling approach.