提出了一种基于模拟退火算法的第二心音(S2)主动脉瓣分量(A分量)和肺动脉瓣分量(P分量)的提取方法.假设A分量和P分量服从包络调制的chirp模型,并用18个参数来表达.在最小均方误差准则下,通过模拟退火算法为A分量和P分量估计出最优参数,从而重构出A分量和P分量.仿真结果表明:在无噪声条件下,重构A分量和P分量的均方误差分别在1%和8%以内;在- 15 dB的加性高斯白噪声条件下,重构A分量和P分量的均方误差分别在1.5%和10%以内.进一步提取了10位志愿者S2信号的A分量和P分量,并用两位志愿者的颈动脉重搏切迹验证了重构的A分量和P分量的分界时刻,证明了方法的有效性.%An extraction method based on simulated annealing (SA) algorithm was proposed to estimate the aortic valve component (A component) and pulmonary valve component (P component) from a second heart sound ( S2). It was assumed that both A and F components are subject to envelope modulated chirp model. Each component was characterized by 9 parameters. A S2 signal is thus represented by 18 parameters in total. Under mean squared error ( MSE) criterion,the optimal parameters of A component and P component were estimated by using SA algorithm. The A and P components were then reconstructed. Simulation results showed that,the MSEs of reconstructed A and P components are less than 1% and 8% for noise-free condition; the MSEs were less than 1. 5% and 10% for - 15 dB additive Gaussian white noise condition. In further applications,A and P components are extracted from S2 signals of 10 subjects. The end timing and start timing of reconstructed A and P components were further verified by the dicrotic notches for 2 subjects.
展开▼