The Optimal Control Of The Molten Metal Level Of The Twin-roll Casting

As an advanced manufacturing technique of metal sheet, twin roll casting technology can help conduct rolling and moulding while the metal is crystalizing in the process of manufacturing. Comparing with traditional hot rolling process, this method simplified manufacturing process, eliminates investment and energy consumption and at the same time improves products quality, making this technology widely used in the field of metal manufacturing. The quality of metal sheet produced by twin roll casting technology is closely related with the location of metal freezing point. During production, the freezing point needs to be formed near the gap of roll casting while the liquid level of metallic molten pool produces effect directly on the forming of freezing point. Since the rapid freezing speed of molten metal in the process of producing sheet with this technology, it’s hard to detect and control artificially relevant parametric like rolling speed, roll-gap and stopper height. Besides, there is strong coupling relationship among parametric of liquid level control of molten pool. Therefore all these elements produce nonlinearity, hysteretic nature and time-varying characteristic. Focusing on the limitation of liquid level control, this thesis demonstrates that automation control can be realized through fuzzy control approach based on fuzzy math and control theory. However, the parametric of traditional fuzzy control majorly rely on expert experience, so there is subjectivity and uncertainty of this control process. Aiming at these problems, this thesis further proposes that based on GA, which features at overall optimization, the fuzzy control parametric can be optimized simultaneously. Optimizing target includes: subordinating degree parametric, control rule base parametric and fuzzy factor. Meanwhile, due to the “early-maturing” phenomenon of GA, adaption strategy needs to be used to modify genetic operator in design process to avoid those problems. This thesis firstly introduces the principle of twin roll casting technology, and then analysis the meaning of liquid level control and demonstrates the development of liquid control. On this foundation, the basic structure of liquid level of molten metal is clarified math model is accomplished and the effect to liquid level caused by parametric is also analyzed. After this part, this thesis focuses on the study of fuzzy control and discusses about both its advantages and limitation and confirms parametric of traditional fuzzy control according to the real control condition. At last, this thesis introduces the basic process of GA, the encoding of chromosome through the approach of real number encoding, the accomplishment of the design of individual fitness function and genetic operator and the realizing of revision function of genetic operator. It can be proved through MATLAB that comparing with traditional fuzzy control, the improved fuzzy control which is based on self-adaption GA has better dynamic property and is more concise and stable while controlling liquid level of molten metal of twin roll caster.