基于监督学习的轨道车辆车轴裂纹扩展预测方法研究

传统裂纹扩展分析需要精确的数学模型以及校准参数,但现实中的复杂形状裂纹扩展过程难以进行精准建模,而通过有限元软件对裂纹扩展过程进行分析需大量的运算时间。因此,文章提出了基于监督学习的轨道车辆车轴裂纹扩展方法,该方法无需进行数学建模,可通过结构健康监测数据以及初始裂纹数据直接预测裂纹的扩展趋势。经验证,该方法预测准确率较高,能够较好地拟合裂纹扩展趋势。

Research on Prediction Method of Rail Vehicle Axle Crack Propagation Based on Supervised Learning

The traditional crack propagation analysis requires accurate mathematical model and calibration parameters,however,it is difficult to accurately model the complex shape crack propagation process in reality.In addition,to change the crack propagation process by FE software requires a lot of computing time.Therefore,this paper proposes a crack propagation method of rail vehicle axle based on supervised learning.This method does not require mathematical modeling,and can directly predict the crack propagation trend through structural health monitoring data and initial crack data.It is proved that the prediction accuracy of this method is high,and it can better fit the crack propagation trend.