【作者】 高大路； 王玉； 常瑞生；

【Author】 GAO Da-Lu, WANG Yu, CHAND Ruei-Sheng(Northwestern Polytechnial University, Xi’an,Shanxi 710072)

【机构】 西北工业大学材料学院；

【摘要】 针对摩擦焊接头“伪结合”缺陷信号的信噪比低、难以准确检测的特点,在对超声波C扫描采集到缺陷信号进行分析预处理的基础上,利用离散二进小波变换对缺陷信号进行去噪。该方法首先选择一个合适的小波函数,计算噪声信号在尺度J下的小波分解;然后估计不同尺度上高频系数的噪声水平,设置阈值,并利用阈值法来修改小波系数;利用低频系数和修改后的高频系数来重构信号,得到去噪后的信号。实验证明,此方法对提高缺陷信号的信嗓比效果较好。更多还原

【Abstract】 Aiming at the characteristic of friction welding pseudo defects signal, such as low signal noise ratio (SNR), and cannot be detected effectively, this paper proposes discrete dyadic wavelet transform (DDWT) method for the friction welding defects signal de-noising , which is based on the analysis and preprocess of the signal collected by ultrasonic C-scanning system. The first step is to choose an appropriate wavelet function, and then calculate the wavelet decomposition on the scale J. The second step is to estimate the noise level of the high-frequency coefficient on different scales, then set threshold value and modify wavelet coefficient using threshold value method. The third step is to reconstruct the signal used by low-frequency coefficient and high-frequency coefficient which has been modified, and then we can get the denoised signal. Examples show that it greatly improves signal noise ratio of defects signal.更多还原