基于非抽样小波镜像变换的电弧特征提取

Arc feature extraction based on undecimated wavelet with mirror image transformation

为准确度量电弧信号突变信号特征,运用非抽样小波镜像变换算法提取电弧信号特征。首先将截取到的电弧信号作镜像合成变换;其次,采用非线性小波阈值和非抽样小波分解与重构方法相结合滤除镜像信号噪声,减少消噪算法运算过程的计算量;最后通过信号镜像逆变换裁剪出真实的电弧波形。实验结果表明,与传统方法相比,新方法在降低信号噪声的同时,有效抑制了信号相位偏移,完整捕获电弧波形,避免信号两侧边缘发生畸变,满足焊接过程实时控制要求。

For accurately measuring the characteristics of jump signal,a undecimated wavelet mirror image transformation algorithm is used to extract the arc signal feature. First,the captured arc signal is transformed by mirror image. Secondly,the minror image signal is denoised by nonlinear wavelet threshold value and decomposition and reconstruction of undecimated wavelet to reduce the calculated amount in operation of denoising algorithm. Finally,real waveform is cut out by inverse transformation of signal mirror image. The result shows that comparing with a traditional method,this new method can effectively restrain the signal phase offset,capture complete arc waveform,avoid distortion on both sides of tile signal edge, and meet the requirements of real-time control during welding process.