摘要
为增强轴承退化特征信息,提高广义回归神经网络(Generalized Regression Neural Network, GRNN)的预测精度,提出了一种基于小波包能量谱和改进FOA-GRNN的轴承剩余使用寿命预测方法。首先,为提取和增强轴承退化特征,采取小波包能量谱对轴承振动信号进行分解,生成频带能量谱,以能量谱信息构建轴承退化特征;其次,为提高果蝇优化算法(Fruit Fly Optimization Algorithm, FOA)的寻优能力和寻优效率,提出了一种多种群自适应果蝇优化算法,引入自适应惯性权重,并应用于广义回归神经网络参数优化;实验结果表明,基于文中退化特征相比时域、频域特征,提高了预测精度,改进FOA-GRNN与FOA-GRNN、MFOA-GRNN、IFOA-GRNN相比具有较高的寻优精度和寻优效率。
In order to enhance the degradation characteristics information of bearings and improve the prediction accuracy of the generalized regression neural network(GRNN), a prediction method of bearing remaining useful life based on wavelet packet energy spectrum and improved FOA-GRNN is proposed. Firstly, to extract and enhance the degradation characteristics of bearing, the wavelet packet energy spectrum is used to decompose the bearing vibration signal to generate the frequency band energy spectrum and energy spectrum information is used to construct degradation characteristics of bearing. Secondly, in order to improve the optimization ability and efficiency of fruit fly optimization algorithm, an adaptive multi-population fruit fly optimization algorithm is designed, and it is applied to the parameter optimization of generalized regression neural network. Finally, the experimental results show that it has higher prediction accuracy based on the degradation characteristics of the paper compare with the time domain and frequency domain. The improved FOA-GRNN has higher optimization accuracy and efficiency than FOA-GRNN, MFOA-GRNN, IFOA-GRNN.
作者
张成龙
郑凯
刘杰
ZHANG Cheng-long;ZHENG Kai;LIU Jie(Department of Brewing Engineering Automation,Moutai Institute,Renhuai Guizhou 564500,China;School of Mines anf Civil Engineering,Liupanshui Normal University,Liupanshui Guizhou 553004,China)
出处
《组合机床与自动化加工技术》
北大核心
2020年第7期73-76,80,共5页
Modular Machine Tool & Automatic Manufacturing Technique
基金
贵州省教育厅青年科技人才成长项目(黔教合KY字[2018]458)
六盘水师范学院校级科研项目(LPSSY201909)。
关键词
轴承
剩余使用寿命
小波包能量谱
广义回归神经网络
果蝇优化算法
预测
bearing
remaining useful life
wavelet packet energy spectrum
generalized regression neural network
fruit fly optimization algorithm
prediction
