用于涡轴发动机传感器故障诊断的未知输入观测器干扰矩阵估算

针对处理涡轴发动机控制系统传感器故障诊断问题时有着良好性能的未知输入观测器展开研究,分析了与系统建模误差相关的干扰矩阵对未知输入观测器观测出的系统状态准确度的影响,提出了一种基于全阶状态观测器的涡轴发动机系统模型干扰矩阵的估算方法,该方法对某一固定工况下的干扰矩阵估算十分准确,使得根据此干扰矩阵设计出的未知输入观测器可以完全不受系统建模误差的影响.在此基础上,又给出了一种适应于更广泛工况范围的干扰矩阵近似方法,使得变工况下的未知输入观测器一样不受建模误差的影响.仿真结果表明:基于该方法估算出的干扰分布矩阵可以保证观测器具有良好的性能,在变工况下对发动机输出量估计残差的二范数不大于2. 

Estimation of unknown input observer disturbance matrix for fault diagnosis of turboshaft engine sensors

Focusing on the research of unknown input observer of good performance to deal with the problem of control system''s sensors fault, a method using a full-order state observer to estimate the disturbance matrix of turboshaft engine system model was proposed based on the analysis of relationship between the disturbance matrix reflecting modeling uncertainty and the accuracy of states observed by UIO. This method contributed to an accurate disturbance matrix so that the results of UIO had no combination of modeling uncertainty when the operating state of turboshaft was fixed. Furthermore, this method was developed to meet the requirements of variable operating states of the engine so that the modeling uncertainty had no influence on the design of UIO in this situation. Finally, some simulations were made to verify the proposed methods, which contribute to obtain a better matrix to ensure that the observed output residual of UIO is less than 2.