中文摘要：

针对纯电动轿车车窗防夹使用安全方面的需求,以解析模型的故障诊断理论为基础,提出了H_/H∞性能指标的最优鲁棒故障诊断车窗防夹控制算法。该算法将车窗防夹事件发生时的电机转矩变化率看做故障,通过构建鲁棒故障诊断观测器产生一个残差信号,当车窗防夹发生时残差会显著偏离零值,从而将纯电动轿车车窗防夹检测问题转换为H-/H∞性能指标的鲁棒故障诊断问题,进而利用线性矩阵不等式LMI的方法来求解。在MATLAB环境下建立了Simulink模型,仿真结果表明了该算法在阈值设计为0.4721的情况下,能够在故障发生后0.15 s及时地检测出防夹事件,同时对不确定扰动及噪声有较强的鲁棒性。

英文摘要：

According to the analysis of anti-pinch window control on pure electric vehicles,this paper proposed the algorithm based on analytical model of H-/H∞ fault detection.For detecting the pinched condition,the pinch torque was considered as a fault to generate a residual.In normal state,the residual was zero.When a fault occured,it would deviate from the constant.The robust fault detection for linear time invariant（LTI） system with both finite energy and noise disturbance is studied by using an H-/H∞ approach.To design optimal residual generator,presented a solution of the optimization problem via several LMI formulations.Based on MATLAB,the results show that the algorithm can detect the fault after it occurs 0.15 s in the threshold of 0.4721.In addition,simulation results show that the proposed algorithm has high sensitivity to fault and strong robustness to uncertainties.