摘要
针对焦化鼓风机系统具有非线性时变、多变量、强耦合及存在随机干扰的特点,通过采用基于最近邻聚类方法的RBF神经网络快速学习算法,实时在线辨识,建立被控对象的精确逆模型并用于控制,实现了将具有强耦合特性的多输入多输出(MIMO)系统解耦成单个独立的伪线性对象,并提出一种基于RBF神经网络逆控制与非线性比例积分微分(PID)控制相结合的智能控制策略,保证了系统稳定的同时改善了控制系统性能。仿真和应用结果证实了该控制策略具有快速适应对象和过程变化的能力及较强的鲁棒性。
A precise inverse model controller was constructed, which is in accordance with the characteristics of blast blower system in coke oven. The system has some special characteristics such as nonlinearity, time-variant, uncertainty, stochastic disturbance, multiple-input-multiple-output （MIMO） and strong coupling, and the blast blower model can realize precise control and on-line identification. The nonlinear MIMO system is decoupled into isolated dynamic pseudo linear objects using radial basis function （RBF） neural network based on nearest neighbor clustering algorithm. An intelligent control strategy is presented, which is based on the proposed RBF neural network inverse controller and a nonlinear PID controller. Simulation and application results demonstrate that stability and improved system performance can be achieved simultaneously. This strategy has the ability to adapt process and object changing quickly and good robust performance in simulation and practical applications.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2009年第11期2309-2315,共7页
Chinese Journal of Scientific Instrument
基金
安徽省教育厅自然科学研究项目(KJ2008B104)资助
关键词
焦炉
鼓风机系统
集气管压力
神经网络
逆控制
解耦
coke-oven
blast blower system
collector pressure
neural network
inverse control
decoupling
