Learning and evolution are two fundamental forms of adaptation. There has been a great interest in combining learning and evolution with artificial neural networks (ANNs) in recent years. This paper: 1) reviews different combinations between ANNs and evolutionary algorithms (EAs), including using EAs to evolve ANN connection weights, architectures, learning rules, and input features; 2) discusses different search operators which have been used in various EAs; and 3) points out possible future research directions. It is shown, through a considerably large literature review, that combinations between ANNs and EAs can lead to significantly better intelligent systems than relying on ANNs or EAs alone.

/pdf/evolving-artificial-neural-networks-4kzrbu7pkn.pdf

Evolving artificial neural networks

A genetic algorithms based multi-objective optimization technique was utilized in the training process of a feed forward neural network, using noisy data from an industrial iron blast furnace. The number of nodes in the hidden layer, the architecture of the lower part of the network, as well as the weights used in them were kept as variables, and a Pareto front was effectively constructed by minimizing the training error along with the network size. A predator-prey algorithm efficiently performed the optimization task and several important trends were observed.

A genetic algorithms based multi-objective neural net applied to noisy blast furnace data

Genetic algorithms (GAs) are biologically inspired computing techniques, which tend to mimic the basic Darwinian concepts of natural selection. They are highly robust and efficient for most engineering optimising studies. Although a late entrant in the materials arena, GAs based studies are increasingly making their presence felt in many different aspects of this discipline. In recent times, GAs have been successfully used in numerous problems in the areas of atomistic material design, alloy design, polymer processing, powder compaction and sintering, ferrous production metallurgy, continuous casting, metal rolling, metal cutting, welding, and so on. The present review attempts to present the state of the art in this area. It includes three broad sections given as: fundamentals of genetic algorithms, genetic algorithms in materials design, and genetic algorithms in materials processing. The first section provides the reader with the basic concepts and the intricacies associated with this novel tec...

Genetic algorithms in materials design and processing

More than 90% of the world's steel is produced using the continuous casting process, a method that has seen enormous advances over the last forty years. Mould fluxes play an important part in this process. These fluxes contain fluorides, which can volatilize at operational temperatures polluting both the plant air and cooling water. Airborne fluoride could potentially be a health and safety issue. Waterborne fluoride forms hydrofluoric acid (HF), which can cause plant corrosion, and may lead to contamination of watercourses necessitating water treatment schemes. This adds to production costs and may present potential environmental hazards. These concerns could be reduced or eliminated by removing fluoride from mould fluxes.The present study examines the effect of different fluxing agents upon key mould flux properties. When substituting fluorides for alternative fluxing agents the key design properties of the fluoride-containing flux must be replicated; namely, (i) flux viscosity at 1300°C, (ii) break temperature and (iii) percentage of crystallinity in the solid slag layer. This is to ensure ‘optimal casting’ where operational problems, such as sticker breakouts and defects such as longitudinal cracking, are minimized. In addition, the quality of the steel should not be affected by the substitution. Therefore, any substitute/additive or combination of additives would have to possess the capacity to replicate the effects that fluorine has on mould flux behaviour.This study focuses on B2O3 and Na2O as alternative substitutes for CaF2 in billet fluxes. The new flux has been successfully tested in a plant trial on a continuous casting plant.

http://www.jstage.jst.go.jp/article/isijinternational/45/7/45_7_1051/_pdf

Development of Fluoride-Free Fluxes for Billet Casting

A thermodynamic model for calculating the phosphorus distribution ratio between top–bottom combined blown converter steelmaking slags and molten steel has been developed by coupling with a developed thermodynamic model for calculating mass action concentrations of structural units in the slags, i.e., CaO-SiO2-MgO-FeO-Fe2O3-MnO-Al2O3-P2O5 slags, based on the ion and molecule coexistence theory (IMCT). Not only the total phosphorus distribution ratio but also the respective phosphorus distribution ratio among four basic oxides as components, i.e., CaO, MgO, FeO, and MnO, in the slags and molten steel can be predicted theoretically by the developed IMCT phosphorus distribution ratio prediction model after knowing the oxygen activity of molten steel at the slag–metal interface or the Fe
 t
 O activity in the slags and the related mass action concentrations of structural units or ion couples in the slags. The calculated mass action concentrations of structural units or ion couples in the slags equilibrated or reacted with molten steel show that the calculated equilibrium mole numbers or mass action concentrations of structural units or ion couples, rather than the mass percentage of components, can present the reaction ability of the components in the slags. The predicted total phosphorus distribution ratio by the developed IMCT model shows a reliable agreement with the measured phosphorus distribution ratio by using the calculated mass action concentrations of iron oxides as presentation of slag oxidation ability. Meanwhile, the developed thermodynamic model for calculating the phosphorus distribution ratio can determine quantitatively the respective dephosphorization contribution ratio of Fe
 t
 O, CaO + Fe
 t
 O, MgO + Fe
 t
 O, and MnO + Fe
 t
 O in the slags. A significant difference of dephosphorization ability among Fe
 t
 O, CaO + Fe
 t
 O, MgO + Fe
 t
 O, and MnO + Fe
 t
 O has been found as approximately 0.0 pct, 99.996 pct, 0.0 pct, and 0.0 pct during a combined blown converter steelmaking process, respectively. There is a great gradient of oxygen activity of molten steel at the slag–metal interface and in a metal bath when carbon content in a metal bath is larger than 0.036 pct. The phosphorus in molten steel beneath the slag–metal interface can be extracted effectively by the comprehensive effect of CaO and Fe
 t
 O in slags to form 3CaO·P2O5 and 4CaO·P2O5 until the carbon content is less than 0.036 pct during a top–bottom combined blown steelmaking process.

A Thermodynamic Model of Phosphorus Distribution Ratio between CaO-SiO2-MgO-FeO-Fe2O3-MnO-Al2O3-P2O5 Slags and Molten Steel during a Top–Bottom Combined Blown Converter Steelmaking Process Based on the Ion and Molecule Coexistence Theory

Simultaneous Dephosphorization And Desulphurization Of Molten Pig Iron

Operational data of BOF and the slag samples for different starting conditions of phosphorus (0.06–0.26%P) and silicon content (0.3–1.2%Si) of hot metal have been analysed. The contribution of parameters which are well known to affect phosphorus distribution at tap, such as basicity, temperature, FeO content of slag, slag mass etc., is investigated through models of the ionic theory of slag, optical basicity, regular solution approach, and molecular theory of slag. The best overall results are obtained by the model based on the molecular theory of slag in which several operational parameters are also incorporated. Investigations of different slag samples, based on optical, SEM, EPMA and X-ray studies, reveal the effect of MgO and Al2 O3 on slag morphology and phosphorus distribution in different phases. It is important to consider the phosphorus distribution ratio in the solid and liquid part of the slag. The solid part of the slag, which is mostly dicalciumsilicate, can contain up to 5% phosphoru...

Effect of MgO and Al2O3 variations in oxygen steelmaking (BOF) slag on slag morphology and phosphorus distribution

The scrap dissolution in an actual process like the BOF is affected both by mass transfer and heat transfer. In this paper, the mass transfer of carbon in liquid melt is considered along with heat transfer. The approaches used in this paper to model the scrap dissolution phenomenon include the application of Green’s function, quasi-static, integral profile, and the finite difference approach for different Biot numbers. Mass transfer coefficients are calculated using the Chilton–Colburn’s analogy for the case of forced convection. Since the quasi-static approach requires the least computational time, it is used for a detailed parametric study, including the effect of other parameters like different scrap ratios and heating rates of liquid melt. The region of control of heat transfer vs mass transfer is also identified. The dissolution of mixed scrap (light and heavy scrap) is investigated for different scrap ratios and the autogenous heating rates of liquid melt, with the help of mathematical models. The heat transfer coefficient is estimated as a function of mixing energy and the mass transfer coefficient by invoking the Chilton–Colburn analogy. The permissible limits of light scrap, which can be charged into the BOF, are also suggested from the results of this model. The Artificial Neural Network (ANN) model is trained on the dataset (patterns) generated by the coupled heat and mass transfer model. The accuracy of the results obtained using different ANN topologies is discussed followed by a recommendation for selecting the best approach.

Scrap Dissolution in Molten Iron Containing Carbon for the Case of Coupled Heat and Mass Transfer Control

Multicomponent Mixed-Transport-Control Theory for Kinetics of Coupled Slag/Metal and Slag/Metal/Gas Reactions: Application to Desulphurization of Molten Iron

Conventional models for prediction of silicon content of blast furnace hot metal are briefly reviewed. Four different artificial neural net (ANN) models, namely, back propagation algorithm (BPA), dynamic learning rate algorithm, functional link network (FLN) and fuzzy neural network (FNN), are trained and tested on operational data from blast furnace (BF1) at Visakhapatnam Steel Plant. FNN can predict silicon mass content of hot metal with a standard error (actual versus predicted) of 0.09% and correlation coefficient of 0.86; standard back propagation predicts with a standard error of 0.08 % and correlation coefficient of 0.79.

Brahma Deo

Papers

Simultaneous Dephosphorization And Desulphurization Of Molten Pig Iron

Effect of MgO and Al2O3 variations in oxygen steelmaking (BOF) slag on slag morphology and phosphorus distribution

Scrap Dissolution in Molten Iron Containing Carbon for the Case of Coupled Heat and Mass Transfer Control

Multicomponent Mixed-Transport-Control Theory for Kinetics of Coupled Slag/Metal and Slag/Metal/Gas Reactions: Application to Desulphurization of Molten Iron

Artificial neural nets for prediction of silicon content of blast furnace hot metal