基于BP神经网络的连铸结晶器液位模糊PID优化控制研究

Research on Fuzzy PID Optimization Control for Continuous Casting Crystallizer Liquid Level Based on BP Neural Network

针对塞棒和拉坯速度等因素对连铸结晶器液位在线控制的影响,提出了一种融合BP神经网络和模糊PID的优化算法。根据连铸结晶器液位控制模型,利用BP神经网络将金属液流入量影响因素量化为塞棒位置,再利用模糊PID关联结晶器内液位与塞棒位置修正量、拉坯速度,在线整定PID参数以实现结晶器内液位的优化控制。仿真结果表明,BP神经网络和模糊PID能避免连铸结晶器液位控制产生的超调,在较短时间内使液位趋于稳态,减小液位的控制偏差,降低塞棒和拉坯速度等不确定扰动对铸坯质量的影响。

According to the influence of stopper and drawing speed on the online control of continuous casting crystallizer liquid level,an optimization algorithm combining BP neural network and fuzzy PID was proposed.According to the continuous casting crystallizer liquid level control model,BP neural network was used to quantify the influencing factors of the molten metal inflow to the position of the stopper,and then fuzzy PID was used to correlate the liquid level in the crystallizer with the correction amount of the stopper position and the drawing speed.PID parameters can be adjusted online to achieve optimal control of the liquid level in the crystallizer.The simulation results show that the BP neural network and fuzzy PID can avoid the overshoot of the continuous casting mold level control,make the level fluctuation stabilize in a short time,reduce the control deviation of the liquid level and reduce the influence of uncertain disturbances such as stopper and drawing speed on the quality of the cast billet.