正则化参数自动优选的RAKSVD方法在地震弱信号去噪中的应用

Application of RAKSVD method with automatic optimization of regularization parameters in seismic weak signal denoising

K均值-奇异值分解(KSVD)去噪方法存在病态问题,因此需要引入正则项进行改进。通过改进正则化近似参数设定,利用AKSVD(近似KSVD)方法对弱信号的识别优势,提出正则化参数自动优选的RAKSVD去噪方法,并进行模型测试和实际资料处理。结果表明,该方法不仅取得了预期的去噪效果,而且更加注重对弱信号的保护,去噪后地震弱信号没有发生畸变,从而有利于对弱信号的提取和识别,同时计算效率还得到了提升。

Due to the ill conditioned problem of K-means singular value decomposition(KSVD)denoising method,it is necessary to introduce a regularization term to improve the stability of the method.In this paper,by improving the regularization parameter setting,and taking advantage of the approximate KSVD(AKSVD)method for weak signal recognition,a regularized approximate K-means singular value decomposition(RAKSVD)denoising method with automatic optimization of regularization parameters is proposed.Model test and practical application show that this method not only achieves the expected denoising effect,but also pays more attention to the protection of weak signal.After denoising,the weak seismic signal has no distortion,which is beneficial to the extraction and recognition of weak signal.In addition,the calculation efficiency is also improved.Therefore,this method has certain practical application value.