基于图像分形的故障特征提取方法

A Fault Feature Extraction Method Based on Fractal Image

提出了一种图像分形的故障特征提取方法。介绍了分形维数的理论计算公式,分析了计盒维数的计算方法。对现场测试的柴油机爆燃阶段振动信号波形图像和切片Wigner高阶矩谱图像进行了分形维数计算。研究结果表明柴油机缸套活塞不同磨损状态的分形维数差异明显,能够表征不同状态的特征参数。

In this paper the fault feature extraction method based on fractal image is put forward. Theoretical calculation formulas of fractal dimension are introduced. The calculation method about box-counting dimension based on image processing is analyzed. The vibration signals measured from diesel engine in the stage of deflagrate are analyzed with slide Wigner higher order moment spectrum. Then the dimensions of fractal image extracted from the signal waves and the wigner higher order moment spectrums are calculated by fractal dimension calculation program. The result indicates that different wearing stage of piston -cylinder can be discriminated depending on feature parameters of fractal dimension because the fractal dimensions of images are different apparently.