Machine Learning Techniques in Predicting Hot Deformation Behavior of Metallic Materials

In engineering practice,it is often necessary to determine functional relationships between dependent and independent variables.These relationships can be highly nonlinear,and classical regression approaches cannot always provide sufficiently reliable solutions.Nevertheless,Machine Learning(ML)techniques,which offer advanced regression tools to address complicated engineering issues,have been developed and widely explored.This study investigates the selected ML techniques to evaluate their suitability for application in the hot deformation behavior of metallic materials.The ML-based regression methods of Artificial Neural Networks(ANNs),Support Vector Machine(SVM),Decision Tree Regression(DTR),and Gaussian Process Regression(GPR)are applied to mathematically describe hot flow stress curve datasets acquired experimentally for a medium-carbon steel.Although the GPR method has not been used for such a regression task before,the results showed that its performance is the most favorable and practically unrivaled;neither the ANN method nor the other studied ML techniques provide such precise results of the solved regression analysis.

Improvement on Conventional Load Distribution Algorithm in Hot Tandem Mills

Load distribution is a key technology in strip hot rolling process, which influences the coil＇s mierostrueture and performance. Currently, Newton-Raphson algorithm is applied to load distribution of hot tandem mills in many hot rolling plants and has some serious defects such as having a strict restriction on initial iterative calculation value and requiring coefficient matrix of nonlinear equations to be nonsingular. To eliminate these defects and improve the online performance of the process control computer, Newton descendent numeric algorithm is introduced to this field to widen the initial value range and a new model named error conversion algorithm is put forth to deal with special conditions when the coefficient matrix is singular. Furthermore, considering the characteristics of load distribution, a condition of strip thickness distribution abnormality and corresponding solutions are provided which ensure that rolling parameters can be calculated normally. Simulation results show that the improved algorithm has overcome the defects of the Newton-Raphson algorithm and is suitable for online application.

Opti mization and Model of Laminar Cooling Control Systemfor Hot Strip Mills

The structure of laminar cooling control system for hot rolling was introduced and the control mode, cooling strategy, segment tracking and model recalculation were analyzed. The parameters of air/water cooling models were optimized by regressing the data gathering in situ, and satisfactory effect was obtained. The coiling temperature can be controlled within ±15℃.

PROFILE AND FLATNESS CONTROL SYSTEM IN 1700mm HOT STRIP MILLS OF ANSTEEL

Varying contact-length backup roll and linearly variable crown work roll are provided for improving the mill performance of profile and flatness control. Integrated with theses technologies, relevant profile and flatness control models are developed for hot strip mills on the basis of large amount of finite element calculation. These models include shape setup control model in process control system, bending force feedforward control model, crown feedback control model and flatness feedback control model in basis automation system. Such a profile and flatness control system with full functions is applied in 1 700 mm industrial hot strip mills of Ansteel. Large amount of production data shows that the crown precision with the tolerance of±18 μm is over 90%, the strip percentage which the actual flatness is within ±25 I-unit surpasses 96%, and general roll consume is reduced by 28% by using the profile and fiatness control system. In addition, schedule-free rolling is realized.

Advanced run-out table cooling technology based on ultra fast cooling and laminar cooling in hot strip mill

In order to meet the severe requirements of market and reduce production costs of high quality steels,advanced run-out table cooling based on ultra fast cooling(UFC) and laminar cooling(LC) was proposed and applied to industrial production.Cooling mechanism of UFC and LC was introduced first,and then the control system and control models were described.By using UFC and LC,low-cost Q345B strips had been produced in a large scale,and industrial trials of producing low-cost dual phase strips were completed successfully.Application results show that the ultra fast cooling is uniform along the strip width and length,and does not affect the flatness of strips.The run-out table cooling system runs stably with a high precision,and makes it possible for the user to develop more high quality steels with low costs.

Prediction of Rolling Load in Hot Strip Mill byInnovations Feedback Neural Networks

Because the structure of the classical mathematical model of rolling load is simple, even with the self-adapting technology, it is difficult to accommodate the increasing dimensional accuracy. Motivated by this fact, an Innovations Feedback Neural Networks （IFNN） was presented based on the idea of Kalman prediction. The neural networks used the Back Propagation （BP） algorithm and applied it to the prediction of rolling load in hot strip mill. The theoretical results and the off-line simulation show that the prediction capability of IFNN is better than that of normal BP networks, namely, for the prediction of the rolling load in hot strip mill, the prediction precision of IFNN is higher than that of normal BP networks. Finally, a relative complete rolling load prediction system was developed on Windows 2003/XP platform using the OOP programming method and the SQL server2000 database. With this sys- tem, the rolling load of a 1700 strip mill was calculated, and the prediction results obtained correspond well with the field data. It shows that IFNN is valid for rolling load prediction.

Dynamic matrix predictive control for a hydraulic looper system in hot strip mills

Controlling the looper height and strip tension is important in hot strip mills because these variables affect both the strip quality and strip threading. Many researchers have proposed and applied a variety of control schemes for this problem, but the increasingly strict market demand for strip quality requires further improvements. This work describes a dynamic matrix predictive control(DMC) strategy that realizes the optimal control of a hydraulic looper multivariable system. Simulation experiments for a traditional controller and the proposed DMC controller were conducted using MATLAB/Simulink software. The simulation results show that both controllers acquire good control effects with model matching. However, when the model is mismatched, the traditional controller produces an overshoot of 32.4% and a rising time of up to 2120.2 ms, which is unacceptable in a hydraulic looper system. The DMC controller restricts the overshoot to less than 0.08%, and the rising time is less than 48.6 ms in all cases.

Comprehensive contour prediction model of work rolls in hot wide strip mill

The predictive calculation of comprehensive contour of work rolls in the on-line strip shape control model during hot rolling consists of two important parts of wear contour calculation and thermal contour calculation, which have a direct influence on the ac- curacy of shape control. A statistical wear model and a finite difference thermal contour model of work rolls were described. The comprehensive contour is the equivalence treatment of the sum of grinding, wear, and thermal contours. This comprehensive contour calculation model has been applied successfully in the real on-line strip shape control model. Its high precision has been proved through the large amounts of actual roll profile measurements and theoretical analyses. The hit rates （percent of shape index satisfying re- quirement） of crown and head flatness of the strips rolled, by using the shape control model, which includes the comprehensive contour calculation model, have about 16% and 10% increase respectively, compared to those of strips rolled by using manual operation.

VCR and ASR technology for profile and flatness control in hot strip mills

The roll contour pattern and variety of work and backup rolls in service and its effect on profile and flatness control performance in 1 700 mm hot strip mill at Wuhan Iron and Steel(Group) Corporation were tested and analyzed by the developed finite element models of different typical roll contours configurations.A rather smooth local work roll contour near strip edges and an increase in rolled length can be obtained by application of long stroke work roll shifting system with conventional work roll contours that is incapable of the crown control.In comparison with the conventional backup and work roll contours configuration,the crown control range by the roll bending force enhances by 12.79% and the roll gap stiffness increases by 25.26% with the developed asymmetry self-compensating work rolls(ASR) and varying contact backup rolls(VCR).A better strip profile and flatness quality,an increase in coil numbers within the rolling campaign and a significant alleviated effect of severe work roll wear contours on performance of edge drop control are achieved by the application of ASR with crown control and wear control ability in downstream stand F5 and VCR in all stands of 1 700 mm hot strip mill.

Remaining Useful Life Prediction for a Roller in a Hot Strip Mill Based on Deep Recurrent Neural Networks

Accurate estimation of the remaining useful life(RUL)and health state for rollers is of great significance to hot rolling production.It can provide decision support for roller management so as to improve the productivity of the hot rolling process.In addition,the RUL prediction for rollers is helpful in transitioning from the current regular maintenance strategy to conditional-based maintenance.Therefore,a new method that can extract coarse-grained and fine-grained features from batch data to predict the RUL of the rollers is proposed in this paper.Firstly,a new deep learning network architecture based on recurrent neural networks that can make full use of the extracted coarsegrained fine-grained features to estimate the heath indicator(HI)is developed,where the HI is able to indicate the health state of the roller.Following that,a state-space model is constructed to describe the HI,and the probabilistic distribution of RUL can be estimated by extrapolating the HI degradation model to a predefined failure threshold.Finally,application to a hot strip mill is given to verify the effectiveness of the proposed methods using data collected from an industrial site,and the relatively low RMSE and MAE values demonstrate its advantages compared with some other popular deep learning methods.

Mechanical and wear properties of Al-Al_3Mg_2 nanocomposites prepared by mechanical milling and hot pressing

β-A13Mg2 intermetallic was used as a reinforcing agent to improve the mechanical properties of an aluminum matrix. Different amounts of A13Mg2 nanoparticles （ranging from 0wt% to 20wt%） were milled with aluminum powders in a planetary ball mill for 10 h. Consolidation was conducted by uniaxial pressing at 400β under a pressure of 600 MPa for 2 h. Microstructural characterization confirms the uniform distribution of A13Mg2 nanoparticles within the matrix. The effects of nano-sized A13Mg2 content on the wear and mechanical properties of the composites were also investigated. The results show that as the A13Mg2 content increases to higher levels, the hardness, compressive strength, and wear resistance of the nanocomposites increase significantly, whereas the relative density and ductility decrease. Scanning electron microscopy （SEM） analysis of worn surfaces reveals that a transition in wear mechanisms occurs from delamination to abrasive wear by the addition of A13Mg2 nanoparticles to the matrix.

CFD-Based Optimization of Hot Primary-Air Pipe Networks in Power Plant Milling Systems

A hot primary-air pipe system is the bridge connecting an air-preheater with a coal mill in power generation stations.The effective geometrical configuration of the pipe network greatly affects the air flow distribution and consequently influences the safe and economic operation of milling systems in power stations.In order to improve the properties of the air flow,in the present work the SIMPLEC method is used to simulate numerically the flow field for the original layout of the system.As a result,the internal mechanisms influencing the uneven pressure drop in each branch are explored and three optimization schemes are proposed accordingly.The numerical results indicate that,for the original layout,the local pressure drop of the tee section accounts for approximately 74%of the total drop of the system,with other pressure drops depending on the specific branch considered.It is shown that after optimization,a roughly balanced flow resistance and flow rate can be obtained.Compared with the original layout,the pressure drop relating to different branches is significantly reduced.

Roll Shape Setting Technology of Hot Galvanizing and Planishing Mill

A roll shape setting model was built for the hot galvanizing and planishing mill. The uniform transversal distributions of the front tension in the exit and the unit pressure were considered as the objective function. At the same time, the quality of the products, the stability of zinc layer, and the homogeneity of spangles were of considerable significance in the planishing process. The model was applied to the roll shape setting of the 1800 cold rolling 3# CGL hot galvanizing and planishing mill of Baosteel Co Ltd. After being planished, the flatness of a strip that was less than 6 I was more than 97%, and the flatness of others were less than 10 I; the pass percentage of the zinc lay- ers reached 100%.

Application of Cooling Water in Controlled Runout Table Cooling on Hot Strip Mill

The controlled runout table cooling is essential in determining the final mechanical properties and flatness of steel strip.The heat of a hot steel strip is mainly extracted by cooling water during runout.In order to study the heat transfer by water jet impingement boiling during runout,apilot facility was constructed at the University of British Columbia.On this pilot facility,the water jet impingement tests were carried out under various cooling conditions to investigate the effect of processing parameters,such as cooling water temperature,water jet impingement velocity,initial strip temperature,water flow rate,water nozzle diameter and array of water nozzles,on the heat transfer of heated strip.The results obtained contribute to the optimization of cooling water during runout.

Roll shifting strategy with varying stroke and step in hot strip mill

A new roll shifting strategy with varying stroke and varying step was investigated,Two characteristic parameters including cat ear height and gap contour smoothness were introduced to assess the effect of shifting strategy on roll wear,and the relation between characteristic parameters and shifting strategy was established.Both varying stroke and varying step can reduce cat ear height and gap contour smoothness,so the shifting strategy with varying stroke and varying step is better than the one with either varying stroke or varying step,Based on the effect of shifting control parameters on characteristic parameters of roll wear,the selection principle of these shifting control parameters was gained.A case study was conducted to validate the proposed roll shifting strategy,reducing uncontrollable quartic loading gap contour,improving strip profile and extending rolling length of a rolling campaign.

Improvement of prediction model for work roll thermal contour in hot strip mill

Though high accuracy of the thermal contour was obtained while adopting finite element method,it could not meet the real-time requirement.One-dimensional finite difference method could realize real-time control in the case of neglecting both the circumferential and radical heat exchange,but the over-simplified modes resulted in poor accuracy.And two-dimensional full explicit difference was also limited in practical application since its time step was restricted to keep the model stable.Consequently,a new method of alternating direction finite difference was introduced and discussed in the model's stability,calculating speed and precision.Specific work roll after one real rolling unit was researched,The result shows that error of temperature on work roll surface between measured and calculated values is within 5 ℃.The influence of rolling rhythm and strip width on thermal crown was also studied.The conclusion is verified theoretically and practically that it can maintain absolutely stable and meet the online requirement.

Simulation-based multi-objective optimization for roll shifting strategy in hot strip mill

A simulation-based multi-objective optimization approach for roll shifting strategy in hot strip mills was presented. Firstly, the effect of roll shifting strategy on wear contour was investigated by mtmerical simulation, and two evaluation indexes including edge smoothness and body smoothness of wear contours were introduced. Secondly, the edge smoothness average and body smoothness average of all the strips in a rolling campaign were selected as objective functions, and shifting control parameters as decision variables, the multi-objective method of MODE/D as the optimizer, and then a simulation-based multi-objective optimization model for roll shifting strategy was built. The experimental result shows that MODE/D can obtain a good Pareto-optimal front, which suggests a series of alternative solutions to roll shifting strategy. Moreover, the conflicting relationship between two objectives can also be found, which indicates another advantage of multi-objective optimization. Finally, industrial test confirms the feasibility of the multi-objective approach for roll shifting strategy, and it can improve strip profile and extend same width rolling miles of a rolling campaign from 35 km to 70 km.

Complete Mill Simulation of the Rolling Process of 1660 mm Hot Strip Continuous Mills

The three-dimensional plastic deformations of strip are analyzed using the stream surface strip element method, the elastic deformations of rolls are analyzed using the influence coefficient method, the analyzing and computing model of shape and crown of 4-high mill was established by combining them, and the rolling process of 1660 mm hot strip continuous mills was simulated. The simulated results tally well with the experimental results. The modei and the method for simulation of shape analysis and control of hot strip mills were provided.

Work roll thermal contour prediction model of nonoriented electrical steel sheets in hot strip mills

The demands for profile and flatness of nonoriented electrical steels are becoming more and more severe. The temperature field and thermal contour of work rolls are the key factors that affect the profile and flatness control in the finishing trains of the hot rolling. A theoretic mathematical model was built by a two-dimensional finite difference to calculate the temperature field and thermal contour at any time within the entire rolling campaign in the hot rolling process. To improve the calculating speed and precision, some special solutions were introduced, including the development of this model, the simplification of boundary conditions, the computation of heat transfer coefficient, and the narrower mesh along the edge of the strip. The effects of rolling pace and work roll shifting on the temperature field and thermal contour of work rolls in the hot rolling process were demonstrated. The calculated results of the prediction model are in good agreement with the measured ones and can be applied to guiding profile and flatness control of nonoriented electrical steel sheets in hot strip mills.

Multi-objective optimization for draft scheduling of hot strip mill

A multi-objective optimization model for draft scheduling of hot strip mill was presented, rolling power minimizing, rolling force ratio distribution and good strip shape as the objective functions. A multi-objective differential evolution algorithm based on decomposition （MODE/D）. The two-objective and three-objective optimization experiments were performed respectively to demonstrate the optimal solutions of trade-off. The simulation results show that MODE/D can obtain a good Pareto-optimal front, which suggests a series of alternative solutions to draft scheduling. The extreme Pareto solutions are found feasible and the centres of the Pareto fronts give a good compromise. The conflict exists between each two ones of three objectives. The final optimal solution is selected from the Pareto-optimal front by the importance of objectives, and it can achieve a better performance in all objective dimensions than the empirical solutions. Finally, the practical application cases confirm the feasibility of the multi-objective approach, and the optimal solutions can gain a better rolling stability than the empirical solutions, and strip flatness decreases from （0± 63） IU to （0±45） IU in industrial production.