基于LightGBM的航空发动机剩余寿命预测

Remaining life Prognostics of Aeroengine Based on LightGBM

机械设备的安全与维护长期以来一直是工业界重点关注的问题,航空发动机作为高精密度的机械设备,对其进行准确的剩余寿命预测是保障航空安全的重要前提。传统的剩余寿命预测方法受模型单一的选择影响明显,在预测结果上的准确性上比较低,并且处理数据的能力有限。因此本文将采用先进的机器学习里的LightGBM算法来进行剩余寿命预测,并且通过贝叶斯算法实现模型超参数的自动优化。实验结果表明,与MLP、XGBoost等算法相比,LightGBM在处理大量数据时有更强大的能力,并且在模型结构上更有深度,对发动机的剩余寿命预测表现出更高的准确性和预测精度。

The safety and maintenance of aero-engine has always been the focus of the industry, and its health status prediction is an important part. The traditional health status prediction method is obviously affected by the single choice of model, the accuracy of prediction results is relatively low, and the ability to process data is limited. Therefore, this paper will use the LightGBM algorithm of advanced machine learning to predict the health state, and realize the parameter optimization of the model through Bayesian algorithm.Experimental results show that, compared with MLP, XGBoost and other algorithms, LightGBM has more powerful ability in processing a large number of data, and has more depth in model structure, showing higher accuracy and prediction accuracy for engine health state prediction.