航空发动机智能容错数字控制

Intelligent Fault-tolerant Digital Control Systems of Aeroengines

研究航空发动机智能容错控制方案,使控制系统在传感器故障时正常工作,执行机构故障时安全工作.在自联想网络的特征提取以及过滤噪声和故障能力的基础上,提出自联想网络的最优估计、轴向故障征兆和三级阈值的综合逻辑以区分软故障和发动机性能蜕化、安装制造公差等引起的不确定性.仿真结果表明,这种方法不需要发动机数学模型,对不确定性鲁棒性好.

Faulttolerant digital control strategy of aeroengine is discussed. The control system can maintain normal operation when sensor fails and maintain safe operation when actuator fails. The autoassociative neural network （AANN） is adopted which has the property of feature extraction, noise and fault filtering. Integrated logic of optimal evaluation of AANN, axial fault signature and 3-level threshold developed has the ability of distinguishing soft fault from uncertainties such as deterioration or manufacturing tolerances. Digital simulation shows that the scheme do not need engine model, and robust to uncertainties.