Hot Rolling Process Of Key Variables' Analysis And Extraction Based On Fuzzy Decision Tree

With the rapid development of Chinese economy, the demand for steel in many fields tends to expand considerably and the quality for steel is requested to a higher level. Hot rolled strip is one of the main steel products, and its microstructures and properties are significantly affected by the temperature of finish rolling. Therefore, we need to set up a reliable finish-rolling temperature model to perform a real-time and effective control of finish-rolling temperature. The premise of establishing a reliable temperature control model needs to analyze all of the finishing process variables and extract several key characteristic variables which affect finish-rolling temperature.In the existing methods for analyzing critical factors, which have important influence on finish rolling temperature, every factors is analyzed individually based on the given finish rolling temperature control model. The analyzing process costs plenty of time, ignores the link between influencing factors, and the analyzing results are not accurate. Furthermore, it can not determine the degree of effect on finish rolling temperature quickly and effectively. Therefore, it is very important to find appropriate methods and tools for extracting critical factors which have significant influence on finish rolling temperature.Based on fuzzy decision tree, an extraction method of hot rolling key variables is presented in this paper. By studying and analyzing the hot rolling mechanism, several key characteristic variables can be extracted which have critical influence on finishing temperature, and the influence degree of every variable on the hot rolling finishing temperature can also be defined.Firstly, this paper illustrates the relations between temperature of hot rolled strip and relevant variables. By comparing and analysis the advantages and disadvantages of decision tree and fuzzy decision algorithm, the construction process of fuzzy decision tree is determined. Then a comparative analysis on the advantages and disadvantages of the decision tree algorithm and fuzzy decision algorithm is conducted also. The actual process data is preprocessed and some attributes are reduced based on the characteristics of the hot rolling production so as to lighten the redundancy among attribute factors. Finally, the fuzzy decision tree algorithm is improved and a key variables extraction system is developed, which is more easily embedded into online system and achieves functional integration.With the key variables extraction system, the most critical variables and factors to hot rolling finishing temperature can be extracted from a number of variables in the hot rolling process. By analyzing these variables, the correlation between the temperature and some critical variables are revealed quantitatively, which not only provides a reference for further causal analysis, but also establishes a solid foundation for the building of a reliable temperature control model for finishing rolling.