多源数据融合建模的寿命预测方法

A Life Prediction Method for Multi-source Data Fusion Modeling

针对单一传感器在设备状态监测期间不能很好地进行退化建模和剩余使用寿命预测的问题,提出了一种多源数据融合建模的寿命预测方法。首先,根据设备的退化性能,构造了复合健康指标;其次,使用非线性漂移维纳过程对设备进行退化建模,通过使用极大似然法估计模型参数后,推导出设备剩余寿命概率密度函数;最后,对所提出的方法进行了验证,并与单一传感器预测结果进行了对比,结果表明此方法具有较高的准确性。

Aiming at the problem that a single sensor cannot perform degradation modeling and remaining useful life prediction well during equipment condition monitoring,a method based on multi-sensor data is proposed for predicting the remaining useful life.First,a composite health index is constructed based on the degradation performance of the device.Then,the degradation of the device is modeled by using a nonlinear drift Wiener process.After estimating the model parameters by using the maximum likelihood method,the Probability Density Function(PDF)of the remaining useful life of the device is derived.Finally,the proposed method is evaluated and compared with the single sensor prediction results.The results show that this method has higher accuracy.