针对传统功率谱在频率概念上的局限性及傅氏变换的固有缺陷,提出一种新的广义局部频率概念,在白适应峰值分解方法的基础上,研究周期激励下Duffing系统随阻尼参数r变化的频域动力学特征,发现了频率分岔现象,并且不同参数r下的混沌时间序列在中心频率附近出现连续频段,其形状具有相似性.通过厄米解调分析,总结出混沌时间序列具有频率调制特性和频率调制的相似性.上述研究表明:提出的基于自适应峰值分解的广义局部频率方法,能够有效提取Duffing系统的频域特征,为观察非线性系统混沌状态下频率连续分布规律提供一种新方法.
Owing to the limitations of the concept of frequency for power spectrum and the inherent defects of Fourier transform, a novel concept of general local frequency is proposed. Based on a approach to adaptive peak decomposition, the dynamic feature in frequency domain varying with parameter r of Duffing system driven by periodic signal is investigated. And a phenomenon of frequency bi- furcation is found. Moreover, coninuous frequency bands exist near the central frequency of chaos time seriers at different values of parameter r and their shapes are similar. By demodulation analysis of Hilbert transform, the modulation characteristic and modulation similarity of chaos time seriers are summarized. The above study shows that the proposed approach to general local frequency based on adaptive peak is effective for freature extraction in frequency domain of Duffing system. It provides a new method to observe a continunous distribution of frequency bands for the non-linear system in chaotic state.