摘要
采用具有更好的仿生效果的径向基函数 (RBF)网络对单处损伤结构及多处损伤结构的损伤程度、位置、区域、处数进行识别 ,网络学习方法选择了简单易行、精度高且运算速度快的正交最小二乘 (OLS)法 .通过实例对该方法进行了测试 ,并与BP网进行了比较 .测试结果可验证 :RBF网络及其OLS学习方法可以快速、有效、高精度地识别结构损伤状况 .
The method of structural damage recognition based on neural network has attracted much attention recently. This paper identifies structural damage extent, location, area and numbers for unilateral damage structures and multiple damage structures by radial basis function neural network (RBFN). The training algorithm is Orthogonal Least Squares (OLS) method which has great computation precision and convergent speed. Some examples are given to demonstrate this method. The test results show that RBFN and it OLS training algorithm can identify the damage information quickly and effectively, with high computation precision.
出处
《固体力学学报》
CAS
CSCD
北大核心
2002年第4期477-482,共6页
Chinese Journal of Solid Mechanics
基金
湖北省自然科学基金 (2 0 0 1ABB0 78)
武汉市青年科技晨光计划项目 (2 0 0 15 0 0 5 0 39)资助
