复杂多变信号分析在机械故障诊断中具有普遍性和重要性。常规分析方法缺乏自适应性，难以准确全面提取复杂多变信号中的丰富信息。以具有典型代表意义的行星齿轮箱为研究对象，针对振动信号的成分复杂性和时变调制特点，通过理论分析、仿真计算和实验研究等手段，研究自适应信号分解方法及其在机械故障诊断中的应用，包括经验调幅调频分解、局部均值分解、本质时间尺度分解、Hilbert振动分解和基于时变滤波的改进拓展，以及自适应时频分析和复杂多分量调制信号解调分析等方法。挖掘发挥自适应分解方法在分析复杂多变信号方面的独特优势，克服常规方法的局限，有效分析复杂多分量时变调制信号中蕴含的丰富特征信息，提取幅值、频率、相位等关键动力学参数及其变化特征，解释故障振动调制机理，全方位多角度揭示机械设备的健康状态，识别故障原因。

Complex signal analysis is a common yet important topic in machinery fault diagnosis. However, most of the conventional signal processing methods lack adaptability, thus are unable to fully extract the meaningful information embedded in complex signals. In this project, with planetary gearboxes as the representative research target, and regarding the complexity and the nonstationary modulation characteristics of vibration signals, adaptive signal decomposition methods and their applications to machinery fault diagnosis will be investigated, by means of theoretical analysis, numerical simulation and experimental validation. Research topics include empirical AM-FM decomposition, local mean decomposition, intrinsic time-scale decomposition, Hilbert vibration decomposition, and their enhancement and generalization via time-varying filtering, as well as adaptive time-frequency analysis and demodulation analysis of complex nonstationary modulated signals. By exploiting the unique advantages of adaptive signal decomposition methods, the limitations inherent with most conventional methods will be overcome. Thus the rich characteristic information contained in complex nonstationary modulated vibration signals (e.g. the fundamental yet key parameters such as amplitude, frequency, and phase, as well as their instantaneous changes) is expected to be effectively extracted, and thereby the health status of machinery will be well identified.