基于广义互相关时延估计算法的声发射定位技术

Acoustic Emission Localization Technique Based on Generalized Cross-Correlation Time Difference Estimation Algorithm

利用互相关分析算法估计出声发射信号到达不同传感器时由于传输距离不同而引起的时间差,是实现声发射源定位的关键技术。在理论分析互相关时差定位算法的基础上,提出了一种基于加权函数的广义互相关声发射定位方法。分析比较了GCC、PHAT、Roth和SCOT 4种不同加权函数用于声发射信号时差估计的效果,选定信号波动少、噪声抑制能力高的PHAT加权函数进行声发射源定位实验,并与互相关算法在不同噪声环境下进行比较。在800 mm×800 mm×5 mm的大理石板上,采用断铅作为声发射信号激励源进行定位实验研究,选取6个测点,每个测点测试10组。结果表明,在实验室环境下,60组测试数据中55组数据相对于测点的定位绝对误差均小于37 mm,相对误差(相对于两传感器间的距离750 mm)小于5%(约占90%)。单测点十次平均后绝对误差均小于20 mm,相对误差小于3%。

The key technique in acoustic emission （AE）localization is to determine the time difference estimation （TDE） between different sensors due to different distances from each sensor to the AE source. GCC （generalized cross-correlation） method based on weighted functions is represented after a theoretical analysis of cross-correlation TDE algorithm. By comparing different effects of four weighted functions, PHAT was chosen to conduct AE source location experiments for its small fluctuations and strong anti-noise ability. And the results were compared with CC （cross-correlation） method under different SNR （Signal Noise Ratio）conditions. Experiments were conducted on a 800 mm×800 mm×5 mm marble slab, and AE source was generated by lead breaks at six different points with each tested 10 times. Results show under the lab environment,among 60 groups test data,55 have absolute errors below 37 mm and relative errors （according to distance between the two sensors, which is 750 mm）less than 5% （almost 90% ）. The average absolute error of each point is less than 20 mm while the average relative error is less than 3%.