Research Of Twin Roll Casting And Rolling Force Based On CMAC Neural Network

In recent years, the twin roll strip casting technology has made remarkable achievements, which has received wide attention from the world steel industry, so it is necessary to improve the process to achieve industrial production as soon as possible. In our country, although twin roll strip casting has been studied for many years, there is still a large gap between the level of foreign technology and foreign technology. Twin roll thin strip casting and rolling technology system structure is very complex, the requirements of the control with high speed, high precision characteristics. So far, the process is not very mature, the design of the control system is not a complete set of control theory. So, so far, are still in the groping.The control of twin roll casting and rolling force is an important part in the whole process of casting process, and is an important subject in the process of twin roll casting. Owing to the particularity of continuous casting process, the cast metal rolling in the molten pool will experience from the high temperature liquid to the changes of semi-solid and solid, will relate to the temperature field, velocity field, therefore mathematical modeling of casting and rolling force is more trouble. Common methods of modeling is the twin roll casting zone can be divided into liquid phase area, mushy zone and solid zone, calculated respectively according to three regions with different physical characteristics, material characteristics. Finally, according to certain rules for the calculation of the obtained total casting and rolling force. But due to its complexity, the mathematical model does not fully reflect the essence of the casting and rolling force. After analyzing the external factors which affect the casting rolling force, this paper presents a calculation model of cast rolling force based on CMAC neural network. And put forward a kind of CMAC neural network and PID compound control method for casting rolling force.This paper is to Liaoning University of science and technology of magnesium alloy casting rolling engineering center project as the research background. Firstly, the paper introduces the magnesium alloy twin roll thin strip casting process of rolling process and describes the research status at home and abroad, then the casting and rolling force calculation of the traditional mathematical modeling, and briefly explain the problem of this method. After analyzing the influence factors of twin roll cast rolling force, the CMAC neural network and the cast rolling force calculation model based on CMAC neural network are introduced. Under this premise, the CMAC neural network and PID composite control is introduced to realize the simulation experiment of roll force control. Matlab simulation results show that the CMAC neural network calculation model of cast rolling force has good accuracy, and the error of the actual value is not more than six percent. And the casting rolling force of CMAC neural network and PID composite control method is effective. Compared with common PID control, the advantages of robustness and dynamic performance are good. All the experimental data used in Matlab simulation experiment are from the center of magnesium alloy casting and rolling engineering of University of Science and Technology Liaoning.