Neural Network Control for Steel Rolling Mills

Martinetz, Thomas; Protzel, Peter; Gramckow, Otto; Sörgel, Günter

doi:10.1007/978-1-4471-3087-1_55

Cited by 13 publications

(6 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As for the expectation value-based updates discussed above, samples can be used to either create discrete estimates ∇ k p(a t |s t ; h j ) ≈ [s(h j + δ jk δh k ) − s(h j )]/2 for the components k of the local gradient ∇ t = ∇p(a t |s t ; h t ), where s = s(h t ) = ±1 depending on whether the outcome of the binary measurement (a t |s t ; h t ) is positive or not. Alternatively, for finite difference updates, one may consider a neural gas [25] inspired approach depicted in Internally generated random cloud of sample controls h k around a given control vector ht at cycle t for which binary measurements "given st, detect at or not" are carried out between external cycles, yielding positive (h + k ) or negative (h − k ) outcomes.…”

Section: B Sample-based Updatesmentioning

confidence: 99%

Quantum machine learning with glow for episodic tasks and decision games

Clausen

Briegel

2018

Phys. Rev. A

View full text Add to dashboard Cite

We consider a general class of models, where a reinforcement learning (RL) agent learns from cyclic interactions with an external environment via classical signals. Perceptual inputs are encoded as quantum states, which are subsequently transformed by a quantum channel representing the agent's memory, while the outcomes of measurements performed at the channel's output determine the agent's actions. The learning takes place via stepwise modifications of the channel properties. They are described by an update rule that is inspired by the projective simulation (PS) model and equipped with a glow mechanism that allows for a backpropagation of policy changes, analogous to the eligibility traces in RL and edge glow in PS. In this way, the model combines features of PS with the ability for generalization, offered by its physical embodiment as a quantum system. We apply the agent to various setups of an invasion game and a grid world, which serve as elementary model tasks allowing a direct comparison with a basic classical PS agent.

show abstract

Section: B Sample-based Updatesmentioning

confidence: 99%

Quantum machine learning with glow for episodic tasks and decision games

Clausen

Briegel

2018

Phys. Rev. A

View full text Add to dashboard Cite

show abstract

“…Currently, in order to achieve the head-end quality requirements, there are automation systems based on physical modelling, particularly in the reheat furnace, roughing mill (RM), finishing mill (FM) and the runout cooling zone. 7,8 Figure 1 depicts a simplified and general diagram of a HSM, from its initial stage, the reheat furnace entry, to the final stage, the coilers.…”

Section: Hot Strip Millmentioning

confidence: 99%

“…Although coiler entry (CLE) temperature prediction y is a critical issue in a hot strip mill (HSM), the problem has not been fully addressed by interval type-2 fuzzy logic control systems. 7 …”

Section: Introductionmentioning

confidence: 99%

Modelling and control of coiling entry temperature using interval type-2 fuzzy logic systems

Mendez

Leduc-Lezama²,

Colás

et al. 2010

Ironmaking & Steelmaking

View full text Add to dashboard Cite

The set-up of the cooling water applied to the strip as it traverses the runout table in order to achieve the coiler entry temperature was made by an intelligent model implemented using interval type-2 fuzzy logic systems. The model uses as inputs the targets for coiling entry temperature, strip thickness, finish mill exit temperature and finishing mill exit speed. The experiments of this application were carried out for three different types of coil in a real hot strip mill. The results proved the feasibility of the system developed for coiler entry temperature prediction. Comparison with the online type-1 fuzzy logic based model shows that the proposed interval type-2 fuzzy logic system improves performance in coiler entry temperature prediction under the tested condition.

show abstract

“…For six years, neural networks from Siemens AG have been applied in steel process control (Röscheisen, et al, 1992;Poppe and Martinetz, 1993;Poppe, et al, 1995;Martinetz et al, 1995;Schlang et al, 1996;Jansen et al, 1999, Döll et al, 1999. Current neural modelling applications for predictive control of strip rolling mills include prediction of the temperature of the rolling stock, prediction of the spread in the roughing mill and the finishing mill, prediction of the rolling forces, prediction of mechanical properties, and others.…”

Section: Examples Of Current Applicationsmentioning

confidence: 99%

Neural computation in steel industry

Schlang¹,

Lang²,

Poppe³

et al. 1999

1999 European Control Conference (ECC)

View full text Add to dashboard Cite

A rolling mill process control system calculates the setup for the mill's actuators based on models of the technological process. Neural networks are applied as components of hybrid neuro/analytical process models. They are the key to fit the general physical models to the needs of the automation of a specific mill. Present applications include the calculation of the rolling force and strip temperature (hot and cold rolling); prediction of width-spread in the finishing mill; control of strip width shape; and control of the coiling temperature in a cooling train (hot rolling). The authors outline how significant benefits are achieved in rolling mill technology by using neural networks. The work presented here is the result of a close cooperation between Siemens Corporate Technology in Munich and the Industrial Projects and Technical Services Group in Erlangen.

show abstract

Neural Network Control for Steel Rolling Mills

Cited by 13 publications

References 1 publication

Quantum machine learning with glow for episodic tasks and decision games

Quantum machine learning with glow for episodic tasks and decision games

Modelling and control of coiling entry temperature using interval type-2 fuzzy logic systems

Neural computation in steel industry

Contact Info

Product

Resources

About