摘要
为了解决目前利用Wiener过程对产品进行寿命预测时,由于考虑个体差异至少需要对模型的三个参数进行估计从而导致计算量较大的问题,提出了利用个体方差和总体方差的相关关系以减少待估计参数个数的寿命预测方法。利用历史寿命数据和实时退化数据,采用Bayes估计和EM(expectation maximization)算法得到性能退化信息的参数值,从而得出防喷阀的剩余寿命的概率密度函数及相关分布。仿真结果表明所提方法不仅减少了待估计参数的个数,使计算过程更简单有效,而且有效提高了剩余寿命的预测精度。
To reduce the number of the parameters which need to be estimated in the Wiener model, a new lifetimepredication approach was proposed. In virtue of the relevant relation between the population variance and the indi-vidual variance, the number of the estimated parameters could be reduced by mthe aid of the history data and real-time degradation data, Bayesian estimation and expectationalgorithm were used to obtain parameters that reflect the performance degradation informationty density function and distribution of the remaining lifetime were obtained for the blowout preventer valve according to the estimated parameters. The emulation results show that the proposed approach can not only reduce the num-ber of the estimated parameters and make the calculation smpler, but t can also improve the prediction accuracy ofthe remaining lifetime.
出处
《山东科技大学学报（自然科学版）》
CAS
2017年第5期23-28,共6页
Journal of Shandong University of Science and Technology(Natural Science)
基金
国家自然科学基金项目(41674037
61374126
61379029
61403223)
山东省自然科学基金项目(ZR2013FM021)
