摘要
提出了一种偏最小二乘回归(PLSR)与人工神经网络(ANN)相结合的故障诊断方法,并将此方法应用于齿轮箱的故障诊断中。首先建立齿轮箱运行状态的PLSR模型,然后建立ANN模型,利用PLSR模型残差和系统参数对ANN进行训练,最后,运用此ANN对齿轮箱实施故障诊断。实验结果表明,该方法能有效地诊断出齿轮箱故障。此外,还将该方法与基于主成分分析(PCA)和ANN的故障诊断方法进行了比较。结果表明,二者诊断精度相同,但作者提出的方法具有更高的模型精度。
A fault diagnosis approach based on partial least squares regression （PLSR） and artificial neural network （ANN） was proposed and applied to fault diagnosis of a gearbox. A PLSR model with regard to working conditions of the gearbox was established. An ANN model was built and trained with PLSR model residual and system parameter. The ANN was tested in recognizing gearbox faults. Experimental results show that gearbox faults are effectively identified by this approach. The diagnosis result based on this method was compared with that based on principle component analysis （PCA） and ANN and the result shows that both methods have the same identification accuracy while the one proposed in this paper has higher model accuracy.
出处
《机床与液压》
北大核心
2009年第12期212-214,218,共4页
Machine Tool & Hydraulics
基金
国家自然科学基金项目(50775035)
国家自然科学基金项目(50875048)
国家高技术研究发展计划(2007AA04Z421)
江苏省自然科学基金项目(BK2007115)
关键词
齿轮箱
故障诊断
偏最小二乘回归
人工神经网络
Gearbox
Fault diagnosis
Partial least squares regression
Artificial neural network
