除了信噪比、有效子波畸变等,稳健性(Robustness)也是度量滤波方法效果的一个重要的物理量,它刻画了滤波系统应对异常点值的能力.一般用影响函数作为评价稳健性的工具.支持向量机方法已较成功地应用于信号与图像的滤波中,尤其Ricker子波核方法更适于地震勘探信号处理.通过考察Ricker子波核最小二乘支持向量回归(LS-SVR:least squares support vector regression)滤波方法的影响函数,可以证明该方法的稳健性较差,本文用加权方法改善该方法的稳健性.经过大量理论实验得到一种改进的权函数,使加权之后的方法具有比较理想的稳健性.进一步用这个权函数辅助的加权Ricker子波LS-SVR处理含噪的合成与实际地震记录,都得到较好的效果.由具有平方损失函数的LS-SVR信号处理系统的无界影响函数出发,本文所提出的权函数可以有效地应用于具有相似损失函数的处理过程,如消噪、信号检测、提高分辨率与预测等问题.
In view of the fact that complex signals are often used in the digital processing of certain systems such as digital communication and radar systems,a new complex Duffing equation is proposed.In addition,the dynamical behaviors are analyzed.By calculating the maximal Lyapunov exponent and power spectrum,we prove that the proposed complex differential equation has a chaotic solution or a large-scale periodic one depending on different parameters.Based on the proposed equation,we present a complex chaotic oscillator detection system of the Duffing type.Such a dynamic system is sensitive to the initial conditions and highly immune to complex white Gaussian noise,so it can be used to detect a weak complex signal against a background of strong noise.Results of the Monte-Carlo simulation show that the proposed detection system can effectively detect complex single frequency signals and linear frequency modulation signals with a guaranteed low false alarm rate.
