CAJ | 학술논문

针对FastICA算法容易陷入局部最优,导致提取的胎儿心电往往含有较多噪声的问题.本文将修正BFGS法(MB?FGS)和混沌优化算法相结合来代替传统的牛顿迭代法,提出一种新的独立分量分析方法,并用于胎儿心电信号的提取.分别用合成信号和临床信号对该算法进行验证,实验结果表明本文提出的算法能提取出清晰并不含母体心电的胎儿心电信号,而且算法性能更优于FastICA.
침대FastICA산법용역함입국부최우,도치제취적태인심전왕왕함유교다조성적문제.본문장수정BFGS법(MB?FGS)화혼돈우화산법상결합래대체전통적우돈질대법,제출일충신적독립분량분석방법,병용우태인심전신호적제취.분별용합성신호화림상신호대해산법진행험증,실험결과표명본문제출적산법능제취출청석병불함모체심전적태인심전신호,이차산법성능경우우FastICA.
Due to fast independent component analysis(FastICA)method is easy to strap into local optimum values,the extracted fetal electrocardiogram(ECG)is often mixed with noise especially maternal ECG. To solve those problem ,this paper presents a novel method for extracting the fetal ECG signal,which combines modified BFGS(MBFGS)method and chaos optimization algorithm instead of Newton iteration method. The experiments are conducted by the analog signals and clinical signals respectively. The results indicate that the proposed method in this paper could well extract fetal ECG and there is almost no maternal ECG in fetal ECG,also our method is superior to FastICA.