摘要
根据腐蚀疲劳裂纹在扩展过程中受到多种环境因素影响,裂纹扩展预测难精确的特点,本文提出了基于遗传算法参数优化的最小二乘支持向量机方法来预测结构腐蚀疲劳裂纹扩展。该算法采用遗传算法优化最小二乘支持向量机的模型参数,从而避免了算法陷入局部最优解,实现了精确度高、泛化能力强的裂纹扩展预测模型。最后通过对已有文献的某试件裂纹扩展的实验数据进行建模分析。结果表明:基于遗传算法的最小二乘支持向量机预测方法优于神经网络算法、蚁群算法,预测误差较小,具有很好的预测能力。
Since crack growth is complicated and difficult to measure, a method of parameter optimized least square support vector machine is presented to predict crack growth. Genetic algorithm is used to optimize the parameters of the least square support vector machine （LS-SVM） for avoiding local optimal solution. Therefore, our optimized model of crack growth is more accurate and comprehensive in crack growth prediction. The training and measuring data of crack growth used in this paper is obtained from Reference[ 14]. Finally, the prediction results of the optimized LS-SVM are compared with neural network and ant colony optimization. The comparison shows that the LSSVM based on genetic algorithm model is more accurate for the prediction of crack growth.
出处
《机械科学与技术》
CSCD
北大核心
2008年第11期1346-1350,共5页
Mechanical Science and Technology for Aerospace Engineering
基金
国家高技术研究发展计划项目(863计划)(2006AA04Z427)
国家自然科学基金委员会与中国民用航空总局联合项目(60672164)资助
关键词
疲劳
裂纹扩展
最小二乘支持向量机
遗传算法
优化
fatigue
crack growth
least square support vector machine
genetic algorithm
optimization
