基于局域波降噪和双谱分析的自动机故障诊断研究

特征提取是机械故障诊断的关键，能否准确地提取出反映机械设备工作状态的特征信息，直接影响到故障诊断的准确性和早期预报的可靠性。自动机表面的振动信号成分复杂，除了含有丰富的零部件运动状态信息外，也存在着大量的噪声成分。只有有效地去除干扰信息，才能对信号做出正确地评估和分析。自动机表面的振动信号具有明显的短时冲击特性，是一种典型的非高斯、非线性信号。高阶谱分析，特别是双谱分析，在处理非高斯信号和识别非线性系统故障等方面具有一定的优越性。将局域波理论和双谱分析相结合，对自动机振动信号进行降噪处理和分析，结果表明其应用在自动机故障诊断中具有较好的识别效果。

Fault Diagnosis for Automata Based on Local Wave Noise Reduction and Bispectral Analysis

Feature extraction is a key of mechanical fault diagnosis, which directly affects the accuracy of fault diagnosis and the reliability of early prediction. The vibration signal components of automata surface are complex, including rich information on the motion states of components and parts, and a lot of noise components. Interference information should be effectively removed in order to make a correct assessment and analysis of the signal. The vibration signal of automata surface has obvious short impact characteristics, and is a typical non-Gaussian, nonlinear signal. Bispectral analysis has certain advantages especially in dealing with non-Gaussian signal and identifying nonlinear system failures. Automata vibration signal is noise-reduced and analyzed by the local wave theory and the Bispectral analysis.