基于四阶累积量的供水管道泄漏振动信号自适应时延估计

Adaptive time delay estimation of water supply pipeline leakage vibration signal based on the fourth-order cumulant

在相关高斯噪声干扰下,特别是在低信噪比时基于互相关的供水管道泄漏振动信号时延估计性能严重下降。因此,提出基于四阶累积量及递归最小二乘(RLS)自适应滤波的供水管道泄漏振动信号时延估计方法。首先,通过自、互四阶累积量分割信号频带,获得含有泄漏振动信号特征的一维切片;然后,引入可变步长的RLS算法,提高收敛速度;最后,利用RLS算法对上述一维切片进行迭代计算,分析滤波器权系数曲线及误差曲线得出时延估计结果。仿真与实验结果表明,该方法能够有效的实现供水管道泄漏定位:在低信噪比与相关高斯噪声背景下,互相关方法失效;与四阶累积量-互相关方法相比,所提方法的平均相对定位误差减少了2.56倍,标准差减小了1.736倍。

Under the interference of associated Gaussian noise, the performance of vibration signal delay estimation based on cross-correlation for water supply pipeline leakage degrades seriously, especially at low signal noise ratio. Therefore, this study proposes a delay estimation method based on the fourth-order cumulant and recursive least square(RLS) adaptive filtering for water supply pipeline leakage vibration signal. Firstly, a one-dimensional slice with characteristics of leakage vibration signal is achieved by segmenting the signal band with self-fourth-order cumulant and inter-fourth-order cumulant. Then, the variable step size RLS algorithm is introduced to improve the convergence speed. Finally, the RLS algorithm is used to calculate the aforementioned one-dimensional slice iteratively, and the filter weight coefficient curve and error curve are analyzed to realize the delay estimation results. Simulation and experimental results show that this method can effectively locate the leakage of water supply pipeline. Compared with the fourth-order cumulant-cross correlation method, the mean relative positioning error and standard deviation of the proposed method are decreased by 2.56 times and 1.736 times, respectively.