退化数据驱动的气缸剩余寿命在线预测

针对气缸可靠性研究中剩余寿命预测方面的问题,提出了一种基于退化路径的气缸剩余寿命在线预测方法。在建立了基于维纳过程的气缸退化模型基础上,推导了退化路径决定下的气缸剩余寿命的概率密度函数解析表达式,提出了一种融合Bayes估计和期望最大化算法的参数在线估计方法,实现了气缸剩余寿命在线预测,并通过气缸性能退化实验数据验证了方法的有效性。通过与同类方法对比结果表明,在小样本情况下,所提方法能更准确地预测气缸剩余寿命且预测的不确定性更低。

On-Line Residual Life Prediction for Cylinder Based on Degradation Data

In the light of the residual life prediction in cylinder reliability, an on-line residual life prediction method of cylinder based on degradation path is proposed. First, a degradation model of cylinder based on Wiener process is constructed and the analytical expression of probability density function of residual life determined by the degradation path is deduced. Then, a parameter estimation method based on Bayesian paradigm and expectation maximization algorithm is proposed to achieve the on-line residual life prediction. Finally, the validity of the proposed residual life prediction method is verified by the cylinder performance degradation experimental data. The results show that the method proposed in this paper can predict the residual life of the cylinder more accurately with a lower prediction uncertainty compared with the existing similar methods in the case of the small sample.