基于WOA-AdaBoost的碳纤维复合材料疲劳损伤预测
Fatigue damage prediction of carbon fiber reinforced polymer based on WOA-AdaBoost
段启明 1张杰 1单喆煜 2王加红 3李斌4
作者信息
- 1. 云南交通职业技术学院交通信息工程学院,云南 昆明 650500
- 2. 云南交通职业技术学院交通运输工程学院,云南 昆明 650500
- 3. 华能澜沧江股份有限公司小湾电厂,云南 昆明 650214
- 4. 华能龙开口水电有限公司,云南 昆明 650214
- 折叠
摘要
为了更准确预测碳纤维复合材料疲劳损伤扩展,提出一种基于WOA-AdaBoost的碳纤维复合材料疲劳损伤预测方法.提取传感信号中时域波形、时域统计、频域特征作损伤面积值构建数据集,引入AdaBoost集成学习预测其疲劳损伤,采用鲸鱼优化算法对AdaBoost中弱学习器的学习率与数量进行寻优处理,构建WOA-AdaBoost的预测模型实现碳纤维复合材料疲劳损伤预测.实验结果表明,相较于AdaBoost、SVM等预测方法,建立的WOA-AdaBoost预测方法相关系数为 0.949,RMSE、MAE参数值更小,对碳纤维复合材料的损伤具有更好的预测效果.
Abstract
In order to predict the fatigue damage extension of carbon fiber reinforced polymer more accurately,a fatigue damage prediction method for carbon fiber reinforced polymer was proposed based on WOA-AdaBoost.The time-domain waveform features,time-domain statistical features,and frequency-domain features in the sensing signals were extracted as inputs,and the damage area values as outputs,and AdaBoost integrated learning was introduced to predict their fatigue damage.In order to further improve the prediction accuracy,the whale optimization algorithm was used to optimize the learning rate and the number of weak learners in AdaBoost,and the prediction model of WOA-AdaBoost was constructed to achieve the fatigue damage prediction of carbon fiber reinforced polymer.The experimental results showed that the correlation coefficient of the established prediction method of WOA-AdaBoost was 0.949 and the values of its RMSE and MAE parameters were smaller compared to the prediction methods such as AdaBoost and SVM,the proposed method had a better prediction of the damage of carbon fiber reinforced polymer.
关键词
结构健康监测/碳纤维复合材料/疲劳损伤预测/WOA-AdaBoostKey words
structural health monitoring/carbon fiber reinforced polymer/fatigue damage prediction/WOA-AdaBoost引用本文复制引用
基金项目
云南省哲学社会科学规划教育学项目(2023)(AC23010)
出版年
2024
国家科技期刊平台
NSTL
万方数据