摘要
针对失效率函数可变的非线性动态系统,基于系统状态建立参数未知的动态失效率模型。将历史状态和Nelson-Aalen估计的失效率数据作为训练样本,利用具有快速且精确学习能力的径向基网络逼近动态失效率函数。针对系统的非线性,通过粒子滤波方法对运行中的状态进行在线估计,并由此确定随状态改变的系统失效率,从而对可靠性进行实时评估与预测。通过对带有疲劳裂纹增长的机械系统进行仿真,充分验证了这种可靠性预测方法的可行性和有效性。
For the nonlinear dynamic system with variable failure rate function, a dynamic failure rate model with unknown parameters is built based on system states. Taking the history states and the failure rate data obtained from Nelson-Aalen estimator as the training samples, the radial basis function network is used for approximating the dynamic failure rate function because it has the fast and accurate learning ability. Using the particle filtering method, the system states are on-line estimated, and the relative system failure rate function is determined. Based on the estimated failure rate, the real-time reliability can be predicted. Numerical experiment is performed for a mechanical system with fatigue crack prapagation, and results verify the feasibility and effectiveness of the real-time reliability prediction.
出处
《南京航空航天大学学报》
EI
CAS
CSCD
北大核心
2011年第B07期70-73,共4页
Journal of Nanjing University of Aeronautics & Astronautics
基金
国家自然科学基金(60931160440)资助项目
关键词
可靠性实时预测
动态失效率
神经网络
粒子滤波
real-time reliability prediction
dynamic failure rate
neural network
particle filtering