旋转机械装备故障特征统计量和健康指数在状态监测、故障诊断和寿命预测中起着重要的支撑作用。当前故障特征统计量和健康指数的研究还停留在实验验证阶段，制约了故障诊断理论基础的完善。针对这一问题，本项目拟开展从实验结合理论的角度来研究故障特征统计量和健康指数：(1)研究峭度、平滑指数、熵及基尼指数理论基础，推导出时域统计量和频域包络谱统计量的权重来同时量化脉冲性和循环平稳性，并建立变工况与上述统计量间的数学关系，且进一步研究健康指数和故障程度间的数学联系，揭示造成健康指数强波动性的原因；(2)研究上述统计量的统一数学表达式，揭示量化脉冲性或循环平稳性的一类统计量，并建立与深度神经网络的平行关系。通过对统一数学表达式参数优化，为不同研究对象设计新一代统计量；(3)基于预期新理论，研究故障特征提取与健康指数构造新方法。研究成果利于完善故障诊断理论基础，为设计出变工况下状态监测和故障诊断新方法提供指引。

Fault features and health indicators of rotating machinery play an important role in condition monitoring, fault diagnosis and prognostics. Current investigations on fault features and health indicators of rotating machinery are mainly supported by using experimental results, which lacks of thoroughly theoretical supports. To solve this problem, this project will conduct experimental and theoretical investigations on fault features and health indicators of rotating machinery, including:(1) to theoretically investigate kurtosis, smoothness index, entropy and Gini index for quantification of the impulsiveness and cyclostationarity of repetitive transients caused by early rotating machine faults and then to derive mathematical weights for balancing fault features in a time domain and fault features of squared envelope spectra in a frequency domain; to theoretically investigate how speeds and loads influence changes of the aforementioned fault features and then to build their mathematical relationship for fault features working at varying operating conditions; to theoretically investigate the relationship between health indicators and fault levels so as to discover the causes that result in strong fluctuations of health indicators; (2) to theoretically investigate an unified framework of the aforementioned fault features and then to reveal a class of fault features that can be used to quantify impulsiveness or cyclostationarity of repetitive transients, and then to investigate its parallel relationship with deep Neural networks; to investigate parameters optimization of the unified framework so as to design a new generation of fault feature for a specific object; (3) based on expected new theories, new methods for condition monitoring and fault diagnosis at varying operating conditions will be designed. Expected outcomes from this project will enrich theoretical foundations for early fault diagnosis of rotating machinery, will guide researchers to design more and more new condition monitoring and fault diagnosis methods, and will provide strong and solid foundations for enhancing the reliability and safety of rotating machinery.