Hilbert包络功率谱熵的曲轴轴系故障诊断

Fault Diagnosis of Crankshaft Systems Based on Hilbert Envelope Power Spectrum Entropy

曲轴轴系是往复机械传动系统的重要组件之一,其工作状态对整套机械设备的正常工作和使用具有直接的影响。研究针对振动信号分析的曲轴轴系的损伤模式识别问题,提出一种基于Hilbert包络功率谱熵和支持向量机相结合的往复机械曲轴轴系损伤识别方法。基于曲轴轴系损伤机理展开研究,分析曲轴轴系发生故障振动的原因,建立典型的机械故障动力学模型,并通过构建故障模拟与振动测试系统获取曲轴实时振动信号;基于CEEMDAN方法对信号进行分解,选取峭度较大的本征模式分量进行Hilbert包络解调分析获取包络矩阵,计算信号的瞬时包络功率谱熵,可以明显看出曲轴轴系故障表征;最后进一步用SVM完成模式识别。数值模拟与实验结果验证了方法的有效性。该方法对往复机械曲轴轴系早期故障模式的识别具有较好的参考意义。

As one of the important components of reciprocating machinery transmission system,the operating status of the crankshaft system directly affects the overall performance of the reciprocating machinery.In this paper,the damage pattern recognition of the crankshaft system based on vibration signal analysis is studied.A new method based on Hilbert envelope power spectrum entropy combined with support vector machine is proposed.Based on the damage mechanism of the crankshaft system,the cause of failure mode vibration of the crankshaft system is analyzed and a typical mechanical fault dynamics model is established.Meanwhile,the real-time vibration signals of the crankshaft are obtained by constructing fault simulation and vibration test platforms.The signals are decomposed by complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)method,from which the intrinsic mode function with higher kurtosis is selected for Hilbert demodulation envelope analysis to construct the envelop matrix.Then,the instantaneous envelope power spectrum entropy of the signals is calculated.The fault characterization of the crankshaft system can be clearly seen.Finally,support vector machine is further employed to complete the ultimate pattern recognition.Numerical simulation and experimental research demonstrate the effectiveness of the proposed method.This method provides a reference for the identification of early fault modes of crankshaft systems of reciprocating machinery.