Research Of Slag Detecting Method Based On Vector-Quantization Theory In Continuous Casting

A lot of nonmetal oxide is produced during the course of steel-smelting because a lot of impurities in iron ore and auxiliary materials added to molten steel. By reason of less specific gravity, they float on the surface of molten steel and form a thick layer of slag. So measures must be taked to detect slag at the late stage of casting to improve the quality of steel products.In this dissertation, on the basis of investigations of existing slag detection methods, a new slag detection method is put forward, which combines vibration signal analysis and vector quantization technology. Basic principles of the application of VQ in casting state recognition are expounded. In addition, some key technologies was fully analyzed. There are six chapters in this paper.In chapter 1, the research and application status are introduced in detail, and then , the feasibility of vibration detection method is analysed, finally pattern recognition technology is introduced in casting states recognition.In chapter 2, vector quantization technology(VQ) is simply explained, and then, the basic principle of the application of VQ in recognizing casting status is expound. In the end of the chapter, three key problems is put forward: signal character acquisition; codebook training and fast search.In chapter 3, existing time and frequency analysis methods are used for signal processing to find the characteristic that can indicate casting states, in the end of the chapter, wavelet package analysis method for characteristic acquisition is put forward and testified by results.In chapter 4, familiar codebook design methods and fast search methods are introduced, and then, codebook training and fast search methods of codes for casting states are put forward.In chapter 5, experiments of codebooks' training and casting status recognition are introduced.In chapter 6, jobs completed are summarized and problems waiting for solving are put forward.