基于FPGA的超声信号自适应滤波与特征提取

Adaptive Filtering and Feature Extraction of Ultrasonic Signal Based on FPGA

针对电磁超声特征信号的非线性、非平稳特性,存在传统降噪丢失成分、特征难以提取的问题,该文提出一种用于电磁超声信号的自适应滤波和经验模态分解(EMD)方法相融合的数据处理算法。首先,对超声信号进行稳定性评估,在此基础上采用自适应滤波对电磁超声信号进行降噪处理,融入EMD的自适应滤波对特有频率噪声更敏感,利用EMD分解出不同时间尺度下波动时频信息及所包含的噪声频率成分,实现表征提取;然后,对EMD降噪后的超声信号进行重构,可消除频率混叠现象,并基于现场可编程门阵列(FPGA)实现了对电磁超声信号的实时降噪和特征提取,为进一步缺陷识别、缺陷评估便携化奠定了基础。最后,分别对带有微裂纹、塑性损伤的铝板进行实验研究,验证了该方法的有效性。该方法具有信噪比高、可实时提取时频信息和有效信息丢失少等特点,能对铝板中缺陷进行有效识别。

Aiming at the nonlinear and non-stationary characteristics of electromagnetic ultrasonic characteristic signals,there are problems that traditional noise reduction components and features are difficult to extract.A data processing algorithm for adaptive filtering of electromagnetic ultrasonic signals and the empirical mode decomposition(EMD)method is proposed.Firstly,the stability evaluation of the ultrasonic signal is carried out.On this basis,the adaptive ultrasonic filtering is used to denoise the electromagnetic ultrasonic signal.The adaptive filtering integrated into the EMDis more sensitive to the unique frequency noise.The EMD is used to decompose the fluctuating time and frequency at different time scales.The information and the noise frequency components involved are used to realize the feature extraction.The reconstruction of the ultrasonic signal after EMD denoising can eliminate the frequency aliasing phenomenon,and realize the real-time noise reduction and feature extraction of the electromagnetic ultrasonic signal based on FPGA.The basis for further defect identification and defect assessment and portability has been laid.Finally,the experimental study on aluminum plates with microcracks and plastic damage was carried out,and the effectiveness of the method was verified.The method has the characteristics of high signal-to-noise ratio,real-time extraction of time-frequency information and less loss of effective information,and can effectively identify defects in the aluminum plate.