基于VMD和BPNN-GA的齿轮裂纹故障诊断

Fault Diagnosis of Gear Crack Based on VMD and BPNN-GA

齿轮裂纹是机械传动机构容易出现的故障之一,严重的裂纹直接影响齿轮的使用寿命及整个传动系统的安全。基于变分模态分解(Variational Mode Decomposition,VMD)和改进的BP神经网络模型(BPNN),本文提出了一种齿轮裂纹故障诊断方法。首先对齿轮箱振动信号进行VMD分解,得到内禀模式函数(Intrinsic Mode Function,IMF);然后计算各个IMF的均方根和峭度,并选择与齿轮裂纹长度密切相关的IMF的峭度和均方根作为故障特征;最后通过基于遗传算法(Genetic Algorithm,GA)的BPNN模型对得到的齿轮裂纹故障特征进行分类。结果表明,这里提出的故障诊断方法能够准确识别无裂纹、1/4裂纹、1/2裂纹和3/4裂纹的齿轮,在识别精度和计算效率方面具有优异的综合性能。

Gear crack is one of the faults of mechanical transmission mechanism,which directly affects the service life of gear and the safety of the whole transmission system.Based on the empirical mode decomposition(VMD)and the improved BP neural network model(BPNN),a fault diagnosis method of gear crack is proposed in this paper.Firstly,the vibration signal of gearbox is decomposed by VMD to get the intrinsic mode component IMF;then,the RMS and kurtosis of each IMF component are calculated,and the RMS and kurtosis of the intrinsic mode component closely related to the crack length of the gear are selected as the fault characteristics;finally,based on genetic algorithm genetic algorithm The BPNN model of algorithm(GA)is used to classify the fault features of gear cracks.The results show that the fault diagnosis method proposed in this paper can accurately identify the gears without crack,1/4 crack,1/2 crack and 3/4 crack,and has excellent comprehensive performance in recognition accuracy and calculation efficiency.