飞行器快速故障诊断与主动容错控制一体化设计方法研究

Great attentions have been paid on developing active fault-tolerant control technology to improve the safety of an aircraft in academia and research institutions. Active fault-tolerant control can change flight control according to the fault diagnosis to ensure the safety of the aircraft. However, because the available time for applying fault-tolerant control is very short after the aircraft suffering from a fault, the rapidity and accuracy of fault diagnosis are often contradictory, which becomes a major obstacle to the implementation of active fault tolerant control. Therefore, this proposal will focus on two aspects to solve this contradiction. On the one hand, the following methods will be studied to accelerate the speed of fault diagnosis. The CFT method will be improved to give fast fault warnings for the key parts of aircraft. With the aid of fault warning information, a fast online system identification method will be proposed. And the research on classification of fault models by using support vector machines will achieve rapid fault diagnosis. On the other hand, the partial ordered robust fault-tolerant control method will be studied for a set of fault models. The partial ordered robust fault-tolerant control method can active fault-tolerant control by relying on a set of fault models rather than precise fault model, so that it can relax the requirement for accuracy of fault diagnosis. The integrated design of the fast fault diagnosis and the partial ordered robust fault tolerant control can effectively solve the contradiction between the rapidity and accuracy of fault diagnosis, and can promote the application of the active fault-tolerant flight control technology to enhance the safety of the aircraft.

通过主动容错控制技术提高飞行器的安全性，已成为学术界和研究机构广泛关注的课题。主动容错控制是指根据故障诊断信息，自主改变飞行器控制器来提高飞行器的安全性。但是，由于飞行器发生故障后，可用的容错控制时间很短暂，而故障诊断的精确性与快速性往往自相矛盾，这成为主动容错控制实施的主要障碍。鉴于此，本项目将从两方面研究消除这一矛盾的方法。一方面研究提高故障诊断速度的方法，包括利用CFT方法给出飞行器部件的快速故障预警方法、在故障预警信息辅助下快速地在线辨识系统的方法、以及利用支持向量机对辨识得到的故障模型进行分类的方法；另一方面研究根据故障模型集合建立偏序鲁棒容错控制的方法，使主动容错控制可以依赖于故障模型集合而不是精确的故障模型，放宽对故障诊断精确性的要求。将快速故障诊断与偏序鲁棒容错控制进行一体化设计可有效消除故障诊断精确性与快速性的矛盾，促进飞行器主动容错控制技术的应用，提高安全性。

本项目采用理论设计与分析、数字仿真、实验验证等手段，围绕飞行器快速故障诊断与主动容错控制一体化设计的若干关键问题和既定研究目标取得了以下研究成果：（1）系统地研究了故障诊断与主动容错控制一体化设计方法，重点提出了基于辅助测量系统的故障诊断与主动容错控制的鲁棒一体化设计方法，解决了故障完全解耦和主动容错控制鲁棒性差的问题；（2）系统地研究了共形傅里叶变换（Conformal Fourier Transform，CFT）算法以及在故障检测和模型辨识方面的应用，提出了三维CFT快速算法、针对机翼微小故障的早期CFT故障检测方法以及基于CFT频谱分析的模型辨识方法；（3）研究了飞行器在线选择切换容错控制器的主动容错方法，提出了基于主动H∞容错控制器的多种在线优化选择容错控制策略，以及存在故障误诊断时的最优化容错控制在线选择策略；（4）研究了开源六旋翼飞行器的硬件设计与飞控程序，解决了飞行器电机和各种传感器的在线受控故障模拟与故障注入以及在线频谱分析故障检测算法的工程实现。另外针对飞行器制导与控制的基础性问题，我们提出了基于捕获区域的导弹拦截点火控制策略。上述结果表明，本项目研究对飞行器的快速故障诊断、主动容错控制以及它们的一体化设计都取得了一定成果，完成了研究目标。