机构地区: 东南大学机械工程学院设备监控与故障诊断研究所
出 处: 《制造技术与机床》 2004年第7期24-28,共5页
摘 要: 时频分布从时域特征与频域特征的结合途径揭示了信号的构成本质。文章介绍了基于Wigner -Ville分布 (WVD)的故障诊断方法 ,包括基于核函数抑制交叉项 ,时频分布与人工神经网络相结合 ,以及WVD的高阶谱。机械系统故障信号往往是非平稳的 ,联合时频分布是对故障信号分析的有力工具。WVD很高的能量聚集性和很好的时频分辨率 ,极大地提高了故障信号特征提取的准确度。 Quadratic time frequency uncover the feature of non-stationary signal on the basis of the combination of time domain and frequency domain. In this paper,a new fault diagnosing method based on WVD (Wigner-Ville Distribution ) was introduced,including cross term suppression based on kernel function, time-frequency distribution combining with Artificial Neural Network and Higher Order Spectrum of WVD. Machine fault signal was usually composed of non-stationary signals. Joint time frequency was a powerful tool to analyze fault signal. Due to high temporal and frequency resolution and high energy focus,the accuracy of fault signal feature extraction is greatly improved by WVD.
关 键 词: 机械 故障诊断 时频分布 信号处理 神经网络 高阶谱 分布
领 域: [机械工程]