Continuous Casting Mold Level Control System Based On Fuzzy Neural Network

The mold is one of the major equipments of continuous casting production line in steel rolling industry which possesses the capability to affect the quality, quantity and cost of the steel directly makes the mold level control becoming a crucial part in the continuous casting process. There are some distinctive significant advantages of mold level control system such as stabilizing the level of mold, promoting the efficiency of the primary cooling, increasing the casting output/surface quality, and reduce the labor intensity as well as the accident happening, etc. However, uncomfortable uncertainties and nonlinearities of certain model (eg. erosion of the tundish stopper, casting speed changing, mold oscillation, bad working environment) lead considerable challenges and difficulties for a traditional PID controller. It is therefore essential to develop a practical control strategy with better performance for realistic project.In this thesis, detailed field investigations was dedicated on the liquid level control system of continuous casting mold including the stopper position control loop and the liquid level control loop based on the mold of casting production line in Da-Steel Corporation. In the meantime, the model of AC servo electric cylinder, stopper flow characteristics, characteristics of the casting speed, mold and liquid level detection device were established.Then, a liquid level control system based on fuzzy neural work was designed with a distinct network structure and on line parameters revising capability. Applicable initial parameters of the controller result from the fuzzy neural work controller off-line training by utilizing field data makes the controller we designed reflecting the basic request. Eventually, the simulation is launched under the environment of MATLAB, and it gets the conclusion that the liquid level system of mold based on fuzzy neural network has better performance than the system based on PID in the following aspects:better following performance for furnace temperature varying, shorter overshoot, shorter adjusting time and greater robustness.