基于广义维数的故障特征提取及诊断研究

从分形原理出发，以广义维数理论为基础，建立以振动信号的广义维数、广义维数谱图、敏感维数为特征组的模式空间样本库。一方面，通过计算实测信号的广义维数，提取其敏感维数，进而对实测信号的裂纹故障进行诊断识别。另一方面，通过对待检信号与不同裂纹深度信号的广义维数相关系数计算、分析、比较，可估计出待检信号的裂纹深度，说明广义维数相关系数分析方法，能对圆盘的裂纹故障进行较好地定量分析和识别。

STUDY ON EXTRACTING AND DIAGNOSING OF FAULT FEATURE BASED ON GENERAL DIMENSION

It is a practical important problem to the state monitoring and fault diagnosing of mechanical equipment. Many typical method have being continually improved, for example, time field analysis, FFT frequency field analysis and correlation function analysis. Many new method, gray correlative degree, fuzzy membership grade, expert system based on knowledge, wavelet transform and neural network, etc, have been extensively used in the control and diagnosis of mechanical equipment. From the fractal theory, the sample storage of pattern mode space, which is characterized by a group of vibration signal general dimension, general dimension chart and sensitive dimension, is established on the basis of general dimension theory. On the one hand, by calculating the general dimension of the measured signal and extracting its sensitive dimension, the crack fault of measured signal is diagnosed and identified. On the other hand, by calculating, analyzing and comparing the general dimension correlation coefficient of the waiting measured signal with the crack signal of different depth the crack deep of the waiting measured signal can be evaluated. It explains that the analysis method of general dimension coefficient of correlation can be better quantitatively analyzed the crack fault of circular plate.