摘要
针对当前舰船电机轴承异常检测正确率低、检测自动化程度低、检测过程十分耗时等难题,为了提高舰船电机轴承异常检测效果,设计了基于神经网络的舰船电机轴承异常检测方法。首先提取舰船电机轴承状态信号,采用小波包分析去除舰船电机轴承状态信号中的噪声,然后采用Hilbert变换提取电机轴承异常状态的特征,将特征作为神经网络的输入,电机轴承异常作为神经网络的输出,建立舰船电机轴承异常检测模型,最后进行舰船电机轴承异常检测的仿真实验,本文方法的舰船电机轴承异常检测正确率超过95%,能够很好检测到舰船电机轴承异常现象,而舰船电机轴承异常检测时间要少于当前其他舰船电机轴承异常检测方法,能够满足舰船电机轴承异常检测的实际要求。
In order to improve the effect of abnormal detection of ship motor bearing,a neural network-based abnormal detection method of ship motor bearing was designed.First extract ship motor bearing vibration signal,using wavelet packet analysis eliminate the noise in the ship motor bearing vibration signal,and then using the Hilbert transform to extract the features in the motor bearing vibration signal,the characteristics as neural network input,motor bearing abnormal as the output of neural network,ship motor bearing anomaly detection model is established,finally has carried on the ship motor bearing anomaly detection experiments,the method of ship anomaly detection accuracy of motor bearing more than 95%,can detect the ship motor bearing anomalies,However,the abnormal detection time of ship motor bearing is less than that of other current abnormal detection methods,which can meet the actual requirements of abnormal detection of ship motor bearing.
作者
蔡新梅
CAI Xin-mei(BoHaiShipbuilding Vocational College,Huludao 125000,China)
出处
《舰船科学技术》
北大核心
2019年第14期118-120,共3页
Ship Science and Technology
关键词
舰船电机
轴承异常
检测模型
信号去噪
异常特征向量
ship motor
abnormal bearing
detection model
signal denoising
abnormal eigenvector
