论文部分内容阅读
基于归一化滑动窗协方差算法导出一种新的递推最小二乘格形滤波器，并用它构造了一种新颖的故障检测器。首先，在Ｈｉｌｂｅｒｔ空间定义了信号向量和信号矩阵，描述了归一化残差向量和偏相关系数。然后，用向量空间的正交投影方法推导了ＮＳＷＣ格形滤波器算法方程，并通过对前向残差或后向残差的白色性检验，实现了实时故障检测。最后，用仿真例验证了这种故障检测方法的有效性。仿真结果表明这种故障检测器具有无需被监视系统的数学模型，计算量小、误报警率低等优点。
A new recursive least square lattice filter was derived based on the normalized sliding window covariance algorithm and used to construct a novel fault detector. First, the signal vector and signal matrix are defined in Hilbert space, and the normalized residual vector and partial correlation coefficient are described. Then, we use the orthogonal projection of vector space to derive the NSWC lattice filter algorithm equation, and achieve the real-time fault detection by the whiteness test of the forward or backward residuals. Finally, a simulation example is used to verify the effectiveness of this fault detection method. The simulation results show that this kind of fault detector has the advantages of no need of the mathematical model of the monitored system, small amount of calculation and low false alarm rate.