摘要
提出一种基于神经网络的航空发动机全包线P ID控制器参数整定方法,在全包线内选定若干离线整定点,在这些点离线整定P ID控制器参数kp,ki,kd。以离线整定点参数为训练样本,离线训练BP神经网络,该网络可映射高度H,马赫数M a与kp,ki,kd的非线性关系,便可用该网络在线整定包线内任意点的kp,ki,kd。用发动机非线性部件级模型为被控对象的数字仿真表明,用上述方法设计的发动机P ID控制器在全包线内,都能获得理想的动静态品质。该方法简单易行,效果好,具有实用价值。
The tuning method for aeroengine PID control parameters over the whole envelope is proposed based on neural net＇work（NN）. The off-line tuning points in the envelope are selected for off-line tuning PID parameters, kp,ki,kd. The parameters at off-line tuning points are used for training samples to train BP NN, which maps H and Ma on kp,ki,kd. The trained NN can be used for tuning PID parameters on the whole envelope online. The digital simulation by nonlinear aeroengine component level model used as the controlled plant shows that the steady and dynamic behavior of the control system designed by the method is satisfied. The method is simple, effective and practical.
出处
《南京航空航天大学学报》
EI
CAS
CSCD
北大核心
2007年第2期236-239,共4页
Journal of Nanjing University of Aeronautics & Astronautics
基金
国家自然科学基金(50576033)资助项目
航空科学基金(04C52019)资助项目
