吴亚伦,张凤玲,艾延廷.基于核主成分分析的高温动态应变计疲劳寿命预测[J].航空发动机,2024,50(4):162-168 |
基于核主成分分析的高温动态应变计疲劳寿命预测 |
Life Prediction of High-temperature Dynamic Strain Gauge based on KPCA |
|
DOI: |
中文关键词: 高温动态应变计 核主成分分析 遗传算法 神经网络 航空发动机 |
英文关键词:high-temperature dynamic strain gauge kernel principal component analysis genetic algorithms neural networks aeroengine |
基金项目:沈阳航空航天大学博士启动基金(120421004)、大学生创新创业项目(Z202110143019)资助 |
|
摘要点击次数: 1786 |
全文下载次数: 1555 |
中文摘要: |
高温动态应变计作为航空发动机部件应力、应变检测的重要工具,一旦发生疲劳破坏会直接影响其测试结果的可靠性。
针对目前应变计测试耗时长、使用寿命离散程度高等问题,对高温动态应变计敏感栅结构参数进行基于多类型核函数的核主成分
分析(KPCA)。采用最佳的核函数对应变计疲劳寿命影响因素进行降维,得出栅丝直径、弯数、涂层厚度为主要影响因素;为解决
降维后应变计疲劳寿命预测精度差、收敛速度慢等问题,运用遗传算法(GA)优化反向传递(BP)神经网络,即通过遗传算法对神经
网络中权值和阈值进行参数寻优,应用于高温动态应变计疲劳寿命的预测,并与几种传统的预测方法进行了比较。结果表明:GA
优化后的BP神经网络预测的绝对误差(MAE)、均方误差(MSE)、平均绝对百分比误差(MAPE)均有所减小,对于高温应变计疲劳
寿命的预测效果更可靠。 |
英文摘要: |
As an important tool for stress and strain inspection of aeroengine components, high-temperature dynamic strain gauges
will directly affect the reliability of its test results once fatigue failure occurs. In view of the problems of long time consumption and high
dispersion of service life of strain gage testing, the kernel principal component analysis (KPCA) method based on multi-type kernel func?
tion was carried out for sensitive grid structural parameters of high-temperature dynamic strain gage. At first, the optimal kernel function
was used to reduce the dimensionality of the fatigue life. The grid wire diameter, number of bends, and coating thickness are the main
influencing factors. Then, to solve the problems of poor accuracy and slow convergence speed of fatigue life prediction of strain gages after
dimension reduction, the genetic algorithm (GA) was used to optimize the parameters of the weights and thresholds in the BP neural
network, which was applied to the prediction of fatigue life of high-temperature dynamic strain gages, and compared with several
traditional prediction methods. The results show that the mean absolute error (MAE), mean squared error (MSE), and mean absolute
percentage error (MAPE) of the prediction by BP neural network after GA optimization are all reduced, making it more reliable for strain
gauge life prediction. |
查看全文 查看/发表评论 下载PDF阅读器 |