基于AR模型和小波变换的松动件定位方法

Location estimation method for loose part based on AR model and wavelet transform

针对检测到的松动件冲击信号含有许多噪声信号,为了有效提取有用信号,将基于AR模型和小波变换相结合的多重滤波方法应用于松动件冲击信号的提取.采用移动均值法滤除低频噪声,再利用AR模型对高频噪声进行白化处理,通过小波非线性阈值去噪,实现松动件跌落信号的提取.利用时域均方差获取信号到达各个传感器的时间差,并通过双曲线交汇法计算松动件跌落位置.在钢板试验台对所提出的方法进行试验验证,实验结果表明,所提方法对不同形状和不同质量的松动件都具有比较好的定位效果,且稳定性好.

To estimate the location of loose part,a method based on AR model and wavelet transform was used to extract the impact signal.Low-frequency noise was filtered through the moving average operation.At the same time,high-frequency noise was whitened with AR model.Then the impact signal was extracted by the wavelet nonlinear threshold value filtering method.Time difference of sensors and impact location of loose part can be got respectively using root mean square and hyperbola intersection method.The proposed test method was validated on the steel plate test-bed.And analysis indicates that the proposed method for loose parts with different shapes and different masses have a relatively good position effect and a stable performance.