闭环不确定系统基于反馈控制的故障诊断方法研究

Most fault diagnosis methods are developed for open-loop systems esentially while modern engineering systems usually operate in closed-loop. Due to closed-loop feedback, a fault can act on the control signal which is then propagated to the controlled variables,and finally makes the system extensively showing false faults under the effects of model uncertainty.The existing methods can not work for fault diagnosis of closed-loop uncertain systems. Moreover,the computation constraints make some classic fault diagnosis methods such as oberver-based one can not been directly applied for practical large-scale systems with multiple loops. With such backdrop,the proposal focuses on the following researches on fault diagnosis of closed-loop uncertain systems from the viewpoint of feedback control: 1) Analyzing the influence mechanisms of various faults to feedback control signals and then to residuals for closed-loop uncertain system. The fault evolution model is built and the fault diagnostic conditions is presented. 2) Based on the fault evolution model, the fault detection and isolation methods is proposed with help of feedback decoupling and parameter estimation and compensation techniques. 3) By use of the available signals in the feedback control loops, the residual access design method is presented and the corresponding residual evaluation scheme is developed. 4) Designing the feedback controller to enhance fault diagnosis ability and to faciliate residual access. The main results will be demonstrated by wind turbine plateform.

已有故障诊断方法本质上主要针对开环系统，而现代工程系统通常以闭环方式运行。闭环反馈使得故障作用于控制信号进而传播至被控变量，并在模型不确定性等因素影响下使系统广泛呈现虚假故障，现有方法已不再适应闭环不确定系统的故障诊断；另外，计算等资源的约束使得诸如基于观测器的经典故障诊断方法都无法推广应用于多闭环回路的实际大型系统。为此，拟开展如下闭环不确定系统基于反馈控制的故障诊断方法研究：1)分析各类型故障对反馈控制信号以及模型不确定情况下反馈控制信号对残差的影响机理，建立闭环不确定系统故障演化模型并发展故障可诊断性条件；2)基于演化模型，应用反馈解耦、参数估计与补偿等技术，研究闭环不确定系统的故障检测和隔离方法；3)基于反馈控制环路中的可用信号，研究残差直接提取设计方法并发展相应评估方案；4)开展面向增强故障诊断能力和便于残差提取的闭环不确定系统反馈控制器设计。上述主要成果针对风力发电机进行验证。

现代复杂工程系统通常以闭环方式运行，闭环反馈使得故障作用于控制信号进而传播至被控变量，并在系统不确定性等因素影响下广泛传播，常规故障诊断与容错控制方法效果欠佳；另外，计算或通讯等资源的约束使得自适应调节采样周期以降低资源消耗成为信息物理融合系统的优先选择，然而采样的时变和不确定性进一步增加了闭环系统故障诊断和容错控制的难度。为此，项目重点开展如下研究：1)通过权衡干扰、故障以及事件传输误差，发展闭环系统基于权衡的最优事件触发故障检测方案,其性能远优于现有广泛使用的基于最优辨识的事件触发故障检测方案；2)基于不确定性理论提出了闭环系统自触发采样、故障诊断和容错控制集成设计方案；其易于数字实现且能够尽可能地降低采样频率并保证故障系统输入到状态的稳定性；3)发展了事件触发采样和更新检测机制、故障诊断与故障调节的集成设计框架，同时从互联系统角度借助于小增益定理提出了分离设计框架，有效地降低设计复杂度；4）研究了动态系统基于约束信息的状态估计和故障诊断方法；5)分析了在不确定故障、丢包和时滞影响下闭环系统故障诊断单元和控制单元的耦合影响机理，并发展相应的集成设计方案。通过对风力发电机、四容水箱等基准案列的仿真实验验证了上述成果的可行性和有效性。在项目资助下，共发表/录用论文20篇，其中SCI检索8篇，授权发明专利4项，项目主持人获南通市226人才称号，获中国产学研合作创新成果一等奖和教育部科技进步二等奖各1项。