摘要
针对结晶器液位控制的大惯性、时变、非线性等特点,对其模型和控制策略进行了研究。以最小液位偏差为目标,先用遗传算法离线优化对系统具有全局性影响的模糊控制器参数(cij,bj)和网络结构,再采用BP算法在线调节对系统具有局部性影响的控制输出权值Wi。最后的仿真结果表明,GA-FNC方法的应用大大提高了系统的自学习能力和鲁棒性,并显著改善了铸轧过程的稳定性,值得推广和应用。
Aiming at the features of level control for crystallizer, e.g. large inertia, time varying and non-linear, etc. , the level control model and control strategy have been researched. With obtaining minimum level deviation as the target, the parameters （cij,bj） of the fuzzy controller that affect the system globally and network structure are optimized off line first by using genetic algorithm; then the weight value Wi of control output that affects the system locally is regulated on line by adopting BP algorithm. The final simulation result shows that the application of GA-FNC greatly enhances the capability of self-learning and robustness of the system, and improves the stability of the casting and rolling processes ; so it is worth to be propagated.
出处
《自动化仪表》
CAS
北大核心
2010年第2期35-38,共4页
Process Automation Instrumentation
关键词
连铸结晶器
液位控制模型
遗传算法
BP算法
模糊控制
鲁棒性
Continuous casting crystallizer Level control model Genetic algorithm BP algorithm Fuzzy control Robustness
