基于小波包分解和PSO-BPNN的滚动轴承故障诊断

Rolling bearing fault diagnosis based on wavelet packet decomposition and PSO-BPNN

针对现有煤矿旋转机械滚动轴承故障诊断方法存在信号有效特征提取不完全、故障诊断精度不高及效率低等问题,提出了一种基于小波包分解和粒子群优化BP神经网络的滚动轴承故障诊断方法。该方法包括信号特征提取和故障类型识别两部分:在信号特征提取部分,对采集的滚动轴承振动信号进行小波包分解,得到各子频带能量及信号总能量,经归一化处理后获得表征滚动轴承状态的特征向量;在故障类型识别部分,通过粒子群优化算法优化BP神经网络的初始权值和阈值,以加速网络收敛速度,避免陷入局部极小值。实验结果表明,该方法提高了滚动轴承故障诊断效率和准确率。

In view of problems in existing rolling bearing fault diagnosis methods for coal mine rotating machinery,such as incomplete signal feature extraction,low fault diagnosis accuracy and low efficiency,a rolling bearing fault diagnosis method based on wavelet packet decomposition and particle swarm optimization BP neural network was proposed.The method includes signal feature extraction and fault type recognition.In the signal feature extraction part,the collected vibration signals of rolling bearing are decomposed by wavelet packet to obtain energy of each sub-frequency band and total energy of the signal.After normalization processing,feature vector representing state of rolling bearing is obtained.In the fault type recognition part,initial weight and threshold of BP neural network are optimized by particle swarm optimization to accelerate convergence speed of the network and avoid falling into local minimum.The experimental results show that the method improves fault diagnosis efficiency and accuracy of rolling bearing.