A novel empirical mode decomposition strategy based on Fourier transform and band-pass filter techniques, contributing to efficient instantaneous vibration analyses, is developed in this study. Two key improvements are proposed. The first is associated with the adoption of a band-pass filter technique for intrinsic mode function sifting. The primary characteristic of decomposed components is that their bandwidths do not overlap in the frequency domain. The second improvement concerns an attempt to design narrowband constraints as the essential requirements for intrinsic mode function to make it physically meaningful. Because all decomposed components are generated with respect to their intrinsic narrow bandwidth and strict sifting from high to low frequencies successively, they are orthogonal to each other and are thus suitable for an instantaneous frequency analysis. The direct Hilbert spectrum is employed to illustrate the instantaneous time-frequency-energy distribution. Commendable agreement between the illustrations of the proposed direct Hilbert spectrum and the traditional Fourier spectrum was observed. This method provides robust identifications of vibration modes embedded in vibration processes, deemed to be an efficient means to obtain valuable instantaneous information.

本文研究了一种基于傅立叶变换和带通滤波技术的经验模态分解新策略，有助于进行有效的瞬时振动分析。提出了两个关键的改进。第一个与采用带通滤波器技术进行固有模式函数筛选有关。分解分量的主要特征是它们的带宽在频域中不重叠。第二个改进涉及尝试将窄带约束设计为固有模式功能的基本要求，以使其具有物理意义。由于所有分解成分都是根据其固有的窄带宽以及从高到低依次严格筛选产生的，它们彼此正交，因此适用于瞬时频率分析。直接希尔伯特频谱用于说明瞬时时频能量分布。观察到拟议的希尔伯特直接谱图和传统傅立叶谱图之间的可观协议。该方法提供了嵌入在振动过程中的振动模式的可靠标识，这被认为是获得有价值的瞬时信息的有效手段。