基于信号特征提取和卷积神经网络的轴承故障诊断研究

轴承是机械设备主要零部件之一,也是机械设备主要故障零部件之一。轴承故障问题为机械设备的重点,机械设备的使用受到故障轴承的直接影响。针对传统的卷积神经网络算法轴承故障诊断效率低下问题,本文提出了一种基于信号特征提取和卷积神经网络的优化方法。首先对原始数据信号进行时域和频域的信号特征提取,获得有效的故障特征值。之后,使用卷积神经网络对提取的特征值进行故障诊断,完成故障分类。本文使用美国凯斯西储大学的滚动轴承振动加速度信号作为数据集,对提出的方法进行验证,得到的故障诊断平均准确率为74.37%,准确率的方差为0.0001；传统的卷积神经网络算法故障诊断平均准确率为65.6%；准确率的方差为0.0019。实验结果表明,相比传统的卷积神经网络,提出的方法对轴承故障诊断的准确率有显著的提高,并且该方法的稳定性更佳,计算时间更少,综合性能更佳。

Research on Bearing Fault Diagnosis Based on Signal Feature Extraction and Convolutional Neural Network

Bearing is one of the important parts of mechanical equipment, and it is also one of the main fault parts of mechanical equipment. Bearing failure is the focus of mechanical equipment, and faulty bearings directly affect the use of mechanical equipment. Aiming at the problem of low diagnosis accuracy of bearing fault diagnosis based on traditional convolutional neural network algorithms, this paper proposes an optimization method based on signal feature extraction and convolutional neural network. Firstly, the signal characteristics in the time domain and frequency domain are extracted from the original data signal to obtain the effective fault characteristic values. Then, the convolutional neural network is used to diagnose the extracted feature values and complete the fault classification. In this paper, the rolling bearing vibration acceleration signal of Case Western Reserve University is used as a data set to verify the proposed method, and the average accuracy of the fault diagnosis is 74.37%, the variance of the accuracy is 0.0001, and the average fault diagnosis accuracy of the unoptimized algorithm is 65.6%. The variance of the accuracy is 0.0019. Experimental results show that compared with the traditional convolutional neural network, the proposed method has a significant improvement in the accuracy of bearing fault diagnosis, and the method has better stability, less calculation time and better comprehensive performance.