摘要
针对如何有效利用设备故障监测过程中获取的有限的直接状态信息和大量的间接状态信息,进行故障预测的问题。首先建立了Gamma退化过程状态空间模型;进而在EM中嵌入PF技术,实现了参数求解,并通过仿真方法验证了估计精度;最后根据UH60直升机主减速器行星架的振动特征和裂纹长度数据,将预测模型应用于行星架裂纹增长过程,对任意时刻的裂纹长度进行了预测。实例表明该预测模型可以以较高的精度估计行星架实际裂纹长度。
According to how to predict items failures using the little direct eonditton mtormauon anu abundant indirect condition information that obtained during condition monitoring process, firstly establishes a State Space Model basis on Gamma degradation process. Secondly to introduce particle filtering （PF） into Experience Maximization （EM） algorithm to estimate unknown parameters, moreover, the precision of EM-PF is validated by simulation testing. Finally using the vibration feature and crack length data of the main gear-box planetary plate on UH-60 helicopter, the prediction model into crack growth process is adopted and the crack length in any time point has been predicted. The example analysis shows that the model above has a preferable precision in prediction of real crack length.
出处
《火力与指挥控制》
CSCD
北大核心
2012年第11期69-72,共4页
Fire Control & Command Control
基金
总装重点预研基金资助项目(9140A27020308JB34)
关键词
裂纹增长
参数估计
EM算法
粒子滤波
creak growth,parameter estimation,Experience Maximization,particle filtering
