A Gaussian Process Based Model for Air-Jet Cooling of Mild Steel Plate in Run Out Table

Abstract:

Controlled cooling rate is essential in steel production in order to obtain the desired grades for specific mechanical properties. Optimal control of cooling process parameters is important to obtain the desired cooling rate. The system level uncertainty around the cooling process, the model form error around the generative model for the cooling process as well as the measurement noise make the problem of optimal cooling even more challenging. Machine learning approaches have been used in the recent past to solve optimization and optimal control problems. The present study sets out to design an optimal and robust cooling rate controller using a data-driven approach within a machine learning framework which accounts for the uncertainties inherent in the system. A Gaussian process regression model is developed to predict the cooling rate using temperate-time data and two simulated latent parameters with a suitable confidence interval. The experiments have been undertaken on a laboratory scale Run Out Table setup. The results show the suitability of the proposed approach to obtain a robust response surface of the cooling rate with the process parameters.

• Periodical: Advances in Science and Technology (Volume 106)
• Pages: 137-142
• DOI: https://doi.org/10.4028/www.scientific.net/AST.106.137
• Online since: May 2021
• Authors: Prishat Bachhar, Abhishek Kundu, Pranibesh Mandal*
• Keywords: Cooling Rate, Gaussian Process, Generative Model, Machine Learning

References

[1] Mohapatra SS, Ravikumar SV, Ranjan R, Pal SK, Brat Singh S, Chakraborty S. Ultra fast cooling and its effect on the mechanical properties of steel,. Journal of Heat Transfer. 2014 Mar 1; 136(3).

DOI: 10.1115/1.4025638

[2] Ravikumar SV, Jha JM, Mohapatra SS, Sinha A, Pal SK, Chakraborty S. Experimental study of the effect of spray inclination on ultrafast cooling of a hot steel plate,. Heat and Mass Transfer. 2013 Oct 1; 49(10): 1509-22.

DOI: 10.1007/s00231-013-1190-3

[3] Zhang X, Wen Z, Dou R, Zhou G, Zhang F. Experimental study of the air-atomized spray cooling of high-temperature metal,. Applied Thermal Engineering. 2014 Oct 5; 71(1): 43-55.

DOI: 10.1016/j.applthermaleng.2014.06.026

[4] Jha JM, Ravikumar SV, Sarkar I, Pal SK, Chakraborty S. Jet impingement cooling of a hot moving steel plate: an experimental study,. Experimental Heat Transfer. 2016 Sep 2; 29(5): 615-31.

DOI: 10.1080/08916152.2015.1046019

[5] Laha D, Ren Y, Suganthan PN. Modeling of steelmaking process with effective machine learning techniques,. Expert systems with applications. 2015 Jun 15; 42(10): 4687-96.

DOI: 10.1016/j.eswa.2015.01.030

[6] Li F, Wu J, Dong F, Lin J, Sun G, Chen H, Shen J. Ensemble Machine Learning Systems for the Estimation of Steel Quality Control,. IEEE International Conference on Big Data. 2018 Dec 10: 2245-52.

DOI: 10.1109/bigdata.2018.8622583

[7] Kundu A, Matthies HG, Friswell MI. Probabilistic optimization of engineering system with prescribed target design in a reduced parameter space,. Computer Methods in Applied Mechanics and Engineering. 2018 Aug 1; 337: 281-304.

DOI: 10.1016/j.cma.2018.03.041

[8] Otto J, Paraschivoiu M, Yesilyurt S, Patera AT. Bayesian-validated computer-simulation surrogates for optimization and design: error estimates and applications,. Mathematics and Computers in Simulation. 1997 Nov 14; 44(4): 347-67.

DOI: 10.1016/s0378-4754(97)00061-x

[9] Kong D, Chen Y, Li N. Gaussian process regression for tool wear prediction,. Mechanical systems and signal processing. 2018 May 1; 104: 556-74.

DOI: 10.1016/j.ymssp.2017.11.021

[10] Fuentes R, Cross EJ, Halfpenny A, Barthorpe RJ, Worden K. Autoregressive Gaussian processes for structural damage detection,. Proceeding of the international conference on uncertainty in structural dynamics, Leuven, Belgium. 2014: 469-83.

[11] Zhou K, Liang G, Tang J. Vibration analysis of structure with uncertainty using two-level Gaussian processes and Bayesian inference,. Journal of Physics: Conference Series. 2016 Sep; 744: 012202.

DOI: 10.1088/1742-6596/744/1/012202

[12] Liao X, Zhang Z, Chen K, Li K, Ma J, Lu J. Modified Mechanistic Model Based on Gaussian Process Adjusting Technique for Cutting Force Prediction in Micro-End Milling,. Mathematical Problems in Engineering. 2019 Jan 1.

DOI: 10.1155/2019/7468698

[13] Ryan KM, Kristensen J, Ling Y, Ghosh S, Asher I, Wang L. A gaussian process modeling approach for fast robust design with uncertain inputs,. Turbo Expo: Power for Land, Sea, and Air. 2018 Jun 11; 51135.

DOI: 10.1115/gt2018-77007

[14] Sacks J, Welch WJ, Mitchell TJ, Wynn HP. Design and analysis of computer experiments,. Statistical Science. 1989 Nov 1: 409-23.

[15] Williams CK, Rasmussen CE. Gaussian processes for machine learning,. Cambridge, MA: MIT press; (2006).

[16] Haylock RG, O'Hagan A. On inference for outputs of computationally expensive algorithms with uncertainty on the inputs,. Bayesian statistics. 1996; 5: 629-37.

[17] Oakley J. Eliciting Gaussian process priors for complex computer codes,. Journal of the Royal Statistical Society: Series D (The Statistician). 2002 Mar; 51(1): 81-97.

DOI: 10.1111/1467-9884.00300

[18] Rougier J. Probabilistic inference for future climate using an ensemble of climate model evaluations,. Climatic Change. 2007 Apr 1; 81(3-4): 247-64.

DOI: 10.1007/s10584-006-9156-9