The Research Of Precipitation Prediction And Technical Optimization In Steel Based On Neural Network

In recent years, it has been a trend in the field of steel material to develop a new generation of steal material which is characterized by utrafine grain steel. According to the research, low-carbon utrafine grain steel is available through employing strain enhanced transformation. However, in the process of strain enhanced transformation, various understandings exist as to the mechanism of it. That is because the study of utrafine grain steel is still at the stage of experimental observation and generalized theoretical exploration, which can hardly lead to quantitative stimulated results and therefore raises the cost of the research and prolongs the experimental cycle as well.Making full use of the achievements of primary researches and combining the precipitates explorations of utrafine grain steel in China, this study, financed by Natural Science Foundation of Jiangsu Item, puts forward a structure control system of ultra-fine grain precipitiates preparation which suits the current situation of steel industry. The low-carbon steel was studied to investigate precipitation behavior and the effect of precipitation behavior on structure ultrafinement by chemical composition、deformation temperature、deflection and holding time under multi-pass and small deformation thermal simulation test, firstly put forword the method of precipitation measure and prediction and set up the system of precipitation prediction and structure control in low-carbon steel by precipitation analysis of imag processing and neural network(L-M and fuzzy algorithm), aiming at optimizing the structure and technics.Main contents and innovations:1. Precipitation mechanism model and L-M and fuzzy neural network model arecombined to set up the system of precipitiation prediction and structure control inlow-carbon steel, which improves the traditonal system from singel numeric driveto mechanism drive, which is more scientifically reasonable and more flexible sothat it is suitable to a larger number of steel grades.2. Based on knowledge of utrafine and the technics data of low-carbon utrafine grain steel preparation in thermal simulation of primary researches, experiment database and model parameter database has been established with SQL2000, which support the unified management and statistics analysis in the system. With variable databases ,this system is feasible to extensive system of steel. 3. The automatic classification based on morphological Characteristics, could easily and effectively classify the size and shape of precipitation, which is the preparation of prediction and also provide a reliable basis of precipitation quantitative microscopic analysis.4. With the input parameters for preparation technics parameters of low-carbon utrafine grain steel, and with the output parameters of precipitation size and shape to set up the system of precipitation prediction based on L-M and fuzzy algorithm neural network to predict structure and optimize process.With the above studies, the low-carbon steel precipitation prediction and structure control under thermal simulation process are successfully made sure,which will enrich Ultra-fine theory and cut short the R & D cycle, and provide theoretical and technical guidance for all types of ultra-fine grained steel research and development of new types of steels.