超声无损检测中基于支持向量机的自适应裂谱分析法

传统裂谱分析（SSP）方法对滤波器类型及其参数选择过于敏感，优化处理算法的信噪分离规则不能根据应用场合、信号和噪声的性质进行自适应调整.为了提高超声无损检测（UNDT）和无损评价（UNDE）中基础数据的信噪比（SNR），提出了一种基于支持向量机（SVM）模式识别理论的自适应裂谱分析方法.采用以高斯函数为核函数的SVM所构成的信噪分离器，对信号和噪声进行识别和分离，从而消除噪声，得到高信噪比的超声回波信号.实验结果表明，与传统裂谱分析方法相比，该方法提高了消噪性能的稳定性，增强了湮没晶粒（或其他散射体）散射中缺陷回波信号的能力.

Adaptive split spectrum processing technique based on support vector machine for ultrasonic nondestructive testing

The classical split spectrum processing(SSP) algorithms are sensitive to the filter's type and parameters,and the signal and noise separate rules in the optimization processing cannot adaptively adjust to application environments and characteristics of signal and noise.A novel adaptive SSP technique based on the pattern recognition theory of support vector machine(SVM) was presented to enhance the signal to noise ratio(SNR) of fundamental ultrasonic echo signals for ultrasonic nondestructive testing(UNDT) and ultrasonic nondestructive evaluation(UNDE).A signal and noise separator based on SVM whose kernel function is Gauss function was used to distinguish the target signals from noises,and enhance the SNR by removing noises.The experimental results indicate that compared with the classical SSP algorithm,the presented technique has higher reliability of denoising performance and improves the SNR enhancing ability for ultrasonic target echo signals contaminated by material noises.