基于非线性预测滤波算法的疲劳裂纹扩展预测

针对传统Paris疲劳裂纹扩展模型预测精度低、无法考虑裂纹扩展过程中各种不确定因素影响的问题，提出一种基于非线性预测滤波算法的疲劳裂纹扩展预测方法。使用基于Paris公式的状态空间方程表征裂纹扩展过程，采用基于Lamb波的监测技术构建观测空间方程，利用实时观测信息修正模型预测值。最后通过Q235钢试件的单边疲劳裂纹扩展实验验证了该预测方法的有效性。实验结果表明，非线性预测滤波算法在疲劳裂纹扩展预测中可以有效地修正Paris公式的预测误差，其预测精度高于扩展卡尔曼滤波和粒子滤波算法的预测精度,同时算法效率较粒子滤波算法有明显提高。

Fatigue Crack Growth Prediction Based on NPF Algorithm

A fatigue crack growth prediction method was proposed based on NPF algorithm, aiming at solving the problems that traditional Paris fatigue crack growth model had low accuracy and could not take the influences of various uncertain factors in the crack growth processes into consideration. State space equations were used to characterize crack growth processes based on Paris formula. Observation space equations were established using monitoring technology based on Lamb waves.The real-time observation information was used to modify the predicted values of the model. Finally, the effectiveness of the method herein was verified by the fatigue crack growth experiments of Q235 steel specimens. Experimental results show that the NPF algorithm may effectively correct prediction errors of Paris formula in fatigue crack growth prediction. The prediction accuracy is better than that of the extended Kalman filter and particle filter(PF) algorithms, and the algorithm efficiency is significantly better than that of PF algorithm.