机器学习分析和预测橡胶圈老化性能

Study and prediction on aging of rubber rings by machine learning

采用热空气老化试验分析丙烯酸酯橡胶O形密封圈、氟橡胶O形密封圈以及丁腈橡胶O形密封圈断裂伸长率的变化规律.基于试验数据,利用机器学习方法建立预测橡胶圈断裂伸长率随老化时间变化规律的模型.研究结果表明,Savitzky-Golay滤波法和分段线性插值法的数据增强方法与极限梯度提升算法结合的机器学习技术,能够准确预测橡胶圈的老化规律和长期老化性能;经试验检测数据验证,与传统的拟合函数相比,机器学习预测模型在橡胶圈断裂伸长率的老化预测上具有更显著的准确性和适用性.

The hot air aging test was used to analyze the change rules of elongation at break of acrylate rubberO-ring,fluorinerubberO-ring and nitrile rubberO-ring.Based on the experimental data,models were established to predict the change rules of elongation at break of the rubber rings with aging time by using machine learning method.The results showed that the machine learning technologycombined with the data augmentation method of Savitzky-Golay filtering method and Piecewise Linear Interpolation method and eXtreme Gradient Boosting algorithm could accurately predict the aging law and long-term aging performance of rubber rings.As verified by experimental test data,the machine learning prediction models have more significant accuracy and applicability in the aging prediction of rubber rings elongation at break compared with the traditional fitting functions.