Prediction Model And Optimization Method For Microstructure And Properties Of Hot Rolled Plate

Abstract:Hot-rolled steel sheet, as one of the most important species of steel, has a wide range of applications in various areas. Microstructure and properties-as one of the key performance indicators reflecting the quality of the hot-rolled sheet and very closely linked to that the manufacturing operations, has also been widespread concerned by steel plant and customers. Traditional methods get the microstructure and properties indicator through on-site destructive sampling inspection, which will spend a lot of manpower, material and time, while the optimization design of hot-rolled sheet only can be determined through a lot of physical experiments lacking scientific and rational guidance. Therefore, research microstructure and properties prediction techniques, and find relationships between indicators and process parameters, based on which providing guidance in terms of process system and process parameter setting for developing new type of steel, which has important implications for hot-rolled production.Hot rolling process is a complex system composed of several subsystems with characteristics of severe nonlinearity, uncertainty, time variation, large lag, strong coupling and multi-parameter, which makes it complicated to build a mathematical model. Based on the full analysis of the mechanism of the hot rolling process, this paper proposes a microstructure and properties prediction modeling method based on improved ELM model through the analysis and research of the data related to microstructure and properties; and on basis of the model, achieves the optimal design of the hot rolling process combining with PSO algorithm.The main work and content are as follows:(1) On the basis of analysis of the hot-rolling production process, through the gray relational analysis, the microstructure and properties and its influencing factors of hot rolled plate are studied and analyzed to determine the main influencing factors of hot-rolled plates microstructure and properties indicators including yield strength, tensile strength and elongation, building the foundation for the establishment of hot-rolled products microstructure and properties prediction model.(2) Based on the analysis of extreme learning machine, an improved method is proposed, and the microstructure and properties prediction model of hot-rolled plates is established based on the improved Extreme Learning Machine. The effective and reliability of the model established has been verified by simulation and analysis.(3) On the basis of building up the microstructure and properties model of hot-rolled plates, the optimization problem of its process parameters analysed and discussed in depth, and then the optimization model of hot rolling production process is established. The simulation of the model in terms of optimization is studied combining with particle swarm algorithm, the results of which reveal that this method can rationally set process parameters and provide guidance for the development of new steel types.