摘要
为建立轴承球磨损量数学模型并结合模型对样本进行可靠度评估,设计了BP神经网络,对磨损量曲线进行拟合。分析表明,拟合曲线能够很好地表达磨损量的变化趋势。通过提取神经网络的权值、阈值参数,建立了磨损量数学模型,并结合模型和蒙特卡洛法对小样本数据进行了处理,进而基于性能退化量分布评估了轴承球的可靠度。结果表明提出的方法针对基于单一性能退化的小样本可靠度评估问题提供了有效的解决途径。
A back propagation artificial neural network is designed to fit the curve of bearing balls' wear volumes in this paper.The analysis indicated that the fitting curve could express the trend of wear volumes' changing well.The model of wear volumes can be built through extracting weights and threshold from back propagation neural network,the data got from the model is processed by Monte Carlo methods.At last.the reliability of bearing balls was evaluated by the method which was based on the distribution of performance degradation.
出处
《工业控制计算机》
2019年第5期66-67,70,共3页
Industrial Control Computer
基金
国家"十二五"规划项目(D.50-0109-15-001)
关键词
轴承球
磨损量
BP神经网络
性能退化
小样本
可靠度
bearing ball
wear volumes
back propagation artificial neural network
performance degradation
small samples
reliability