一种最优正交小波基选择的遗传算法方法

An Optimal Choice Method of an Orthogonal Wavelet Basis Through Genetic Algorithms

本文提出一种利用遗传算法的最优正交小波基选择方法。通过将信号用小波级数展开后得到其在某个期望尺度上的近似表示，由此建立一个表达信号与其近似之间误差的代价函数，然后我们利用遗传算法最优化此代价函数以获得全局最优的正交小波基。实验结果表明本文的方法能够快速准确地得到全局最优的正交小波基，从而得到信号在期望尺度上的最佳近似。

An optimal choice method of an orthogonal Wads hots through genetic algorithms is Presental in paper. By expanding the signal in wavelet series, we get the its approximation up to a desired scale. And a cost function for representing the error between the signal and its approximation is established. We then optimize the cost function by genetic algorithms in order to obtain the global optimal orthogonal wavelet basis. The experimental result shows that our technique can aecure the global optimal orthoglnal wavelet hots quickly and accurately, and the bet approximation of signal is then achieved.