论文部分内容阅读
为了改善小波神经网络(WNN)存在收敛速度慢、易陷入局部极小值的缺陷,将WNN和遗传算法(GA)相结合,提出一种基于遗传小波网络(GA—WNN)的多模噪声中确定信号的滤噪方法。由于该方法融合了WNN良好的时频局部分析能力和GA自适应全局快速寻优的特点,因此,GA—WNN不仅克服了WNN存在的不足,而且可以使WNN参数最优化,从而进一步提高WNN的滤噪性能。仿真表明,在多模噪声背景下,GA—WNN能有效地从含噪信号中提取确定信号,并比传统的WNN滤噪效果好。
In order to improve the WNN’s disadvantage of slow convergence and easy falling into local minima, WNN is combined with genetic algorithm (GA) to propose a multi-mode noise based on genetic wavelet network (GA-WNN) Determine the signal noise filtering method. Because of the merging of WNN’s good ability of local time-frequency analysis and GA’s rapid global optimization, GA-WNN can not only overcome the shortcomings of WNN, but also optimize WNN parameters and further improve WNN Noise filtering performance. The simulation shows that GA-WNN can effectively extract the deterministic signal from the noisy signal under the background of multi-mode noise and has better effect than the traditional WNN filtering.