基于小波神经网络弧齿锥齿轮故障诊断及实验研究

Study on Spiral Bevel Gear Fault Diagnosis based on Artificial Neural Networks and Wavelet Transform Technique

提出了一种离散小波变换结合神经网络的故障状态识别方法,运用信号特征提取机理对航空用弧齿锥齿轮故障诊断及状态识别进行了研究。建立了弧齿锥齿轮传动系统振动测试试验台,对正常结构和故障结构的齿轮传动进行了试验测试,通过小波阈值去除掉齿轮箱的振动数据信号系统噪声的影响;采用离散小波变换提取信号的能量特征,利用带有反馈算法的神经网络对齿轮系统的故障状态进行了分类识别。结果表明,该方法齿轮故障识别结果的有效率可达100%,为齿轮系统的故障分析提供了一种有效途径。

A fault diagnosis method of discrete wavelet transform and neural networks is proposed,by using signal feature extraction mechanism,aviation spiral bevel gear fault diagnosis and state recognition is studied.A test rig of spiral bevel gear system vibration testing is built,the testing of normal and defective gear transmission is carried out.The influence of noise on gearbox vibration data signal system is removed through the wavelet threshold.By using discrete wavelet transform,the signal energy features is extracted,gear system fault state classification recognition is carried out by using neural networks with feedback algorithm.The results show that the gear fault recognition result effective rate is to 100%.An effective way for gear system fault analysis is provided.