Łukasz Sztangret

Effective strategies of metamodelling of industrial metallurgical processes

The main objective of the metamodelling is replacing the model of analysed process by its simple (with respect to the computation time) approximation. Metamodel gives a significant reduction of computation time of considered process simulation, as well as its further analysis (sensitivity analysis, optimization, etc.). The paper discusses the idea of metamodelling and compares the effectiveness of three techniques: Response Surface Methodology (RSM), Kriging method and Artificial Neural Network (ANN) applied to the benchmark functions. An example of the use of the considered metamodelling techniques in optimization of the problem of laminar cooling of rolled Dual Phase (DP) steel strips is presented. Metamodelling and optimization of a real industrial metal forming problems seems a novel approach in the field of research on Artificial Intelligence and Optimization practical applications.

Application of inverse analysis with metamodelling for identification of metal flow stress

Abstract The problem of effectiveness of the inverse algorithms used for identification of material model is investigated in the paper. Identification of flow stress models in metal forming processes is considered. This identification is usually performed by coupling the Finite element (FE) model with optimisation techniques which leads to long computing times. A proposition of application of the metamodel in the inverse analysis is presented in the paper. Metamodel is an alternative for the FE model. Artificial neural network was used as a metamodel of the axisymmetrical compression test. Experiments were performed on the Gleeble 3800 simulator for various materials and inverse calculations with the metamodel were performed. Validation of the results confirmed with higher degree of accuracy of the proposed approach. Dans cet article, on examine le problème d’efficacité des algorithmes inverses utilisés dans l’identification de modèle de matériau. On considère l’identification de modèles de contrainte d’écoulement dans les procédés de traitement du métal. Cette identification est habituellement effectuée en couplant le modèle d’EF à des techniques d’optimisation, ce qui mène à de longues durées de calculs. Dans cet article, on propose l’application du métamodèle dans l’analyse inverse. Le métamodèle est une substitution du modèle d’EF. On a utilisé le réseau neuronal artificiel comme métamodèle de l’essai de compression axisymétrique. On a effectué des expériences avec le simulateur Gleeble 3800 pour des matériaux variés et l’on a effectué des calculs inverses à l’aide du métamodèle. La validation des résultats a confirmé le très bon degré d’exactitude de cette approche.

A Methodology for Optimization in Multistage Industrial Processes: A Pilot Study

The paper introduces a methodology for optimization in multistage industrial processes with multiple quality criteria. Two ways of formulation of optimization problem and four different approaches to solve the problem are considered. Proposed methodologies were tested first on a virtual process described by benchmark functions and next were applied in optimization of multistage lead refining process.

Modelling techniques for optimizing metal forming processes

Abstract: This chapter presents optimization techniques and strategies, and their applications to solving problems associated with metal forming processes. Most of the classical optimization models for such processes are strongly non-linear and demand long computing times for complex numerical simulations. More robust and time-effective optimization methods have been intensively researched. Probabilistic, nature-inspired optimization techniques belonging to this group of robust methods, as well as metamodel-driven and approximation-based optimization strategies, are discussed here. Some case studies of the application of these methods to particular metal forming problems are presented.

On Aggregation of Stages in Multi-criteria Optimization of Chain Structured Processes

This work is concerned with complex optimization problems which can be divided into multiple, multi-dimensional problems arranged linearly (as can be observed in the multi-stage industrial processes). The relations between complexity of the problem, level of aggregation of stages into larger groups, and efficiency of search for optimal solution were investigated.

Optimization as a support for design of hot rolling technology of dual phase steel strips

The objective of the paper was performing of the sensitivity analysis of the model used for design of manufacturing technology for auto body parts made of the Advanced High Strength Steels (AHSS). Dual phase steel was considered as an example. The sensitivity analysis was performed to evaluate the importance of all variables as far as their influence on the finishing rolling temperature and grain size. The phase composition after cooling was also considered. An arbitrary hot rolling process characterized only by a number of passes and cooling conditions between passes, as well as by laminar cooling parameters, was selected for the analysis. Metamodel of the rolling cycle was developed to decrease the computing costs for the optimization task. Modified Avrami equation was used for modelling phase transformations during cooling. Such process parameters as the initial temperature, interpass times, heat exchange coefficients and rolling velocities were selected as optimization variables for the rolling proces...

Modified approximation based optimization strategy

The paper presents the Approximation Based Optimization (ABO) strategy and its modification, which allows decrease the optimization computing time through the reduction of a number of objective function calls. It also gives the acceleration of a convergence of the optimization procedure. Elaborated strategy was validated using the Rastrigins benchmark function and in optimizing of the real industrial metallurgical process.

Substituting of a Thermodynamic Simulation with a Metamodel in the Scope of Multiscale Modelling

A typical multiscale simulation consists of numerous fine scale models, usually one for each computational point of a coarse scale model. One of possible ways of limiting computing power requirements is replacing fine scale models with some simplified and speeded up ersatz ones. In this paper, the authors attempt to develop a metamodel, replacing direct thermodynamic computations of precipitation kinetic with an advanced approximating model. MatCalc simulator has been used for thermodynamic modelling of precipitation kinetic. Typical heat treatment of P91 steel grade was examined. Selected variables were chosen to be modelled with approximating models. Several attempts with various approximation variants (interpolation algorithms and Artificial Neural Networks) have been investigated and its comparison is included in the paper.

Application of Metamodels to Identification of Metallic Materials Models

Improvement of the efficiency of the inverse analysis (IA) for various material tests was the objective of the paper. Flow stress models and microstructure evolution models of various complexity of mathematical formulation were considered. Different types of experiments were performed and the results were used for the identification of models. Sensitivity analysis was performed for all the models and the importance of parameters in these models was evaluated. Metamodels based on artificial neural network were proposed to simulate experiments in the inverse solution. Performed analysis has shown that significant decrease of the computing times could be achieved when metamodels substitute finite element model in the inverse analysis, which is the case in the identification of flow stress models. Application of metamodels gave good results for flow stress models based on closed form equations accounting for an influence of temperature, strain, and strain rate (4 coefficients) and additionally for softening due to recrystallization (5 coefficients) and for softening and saturation (7 coefficients). Good accuracy and high efficiency of the IA were confirmed. On the contrary, identification of microstructure evolution models, including phase transformation models, did not give noticeable reduction of the computing time.

Optimization of parameters of feed-back pulse coupled neural network applied to the segmentation of material microstructure images

The paper presents application of bio-inspired optimization procedures to the problem of image segmentation of material microstructures. The method used for image processing was Feed-Back Pulse Coupled Neural Network (FBPCNN), which is very flexible in the case of highly diversified images, offering interesting results of segmentation. However, six input parameters of FBPCNN have to be adjusted dependently on image content to obtain optimal results. This was the main objective of the paper. Therefore, the procedure of image segmentation assessment was proposed on the basis of number of segments, their size, entropy and fractal dimension. The proposed evaluation was used as objective function in optimization algorithms. The results obtained for Simple Genetic Algorithms, Particle Swarm Optimization and Simulated Annealing are presented.