基于HPSO优化LS-SVM的混凝土梁损伤诊断模型

Damage Diagnosis Model for Concrete Girder Based on LS-SVM Optimized by HPSO

针对传统结构损伤智能诊断方法所存在的缺陷,提出一种基于混合粒子群算法(HPSO)优化最小二乘支持向量机(LS-SVM)的结构损伤识别模型,并以混凝土梁为例,以模态参数及其相关量作为输入量,以损伤位置和程度作为输出量,建立起适应的映射模型。此外,为提高LS-SVM的泛化能力,应用HPSO对其核函数参数进行了优化。结果表明:应用HPSO优化LS-SVM所构建的模型具有识别精度高的特点。

Here, a sort of recognition model about structural damage diagnosis is put forward based on the least square support vector machines （LS-SVM） optimized by HI＇SO so as to overcome the shortcomings of the traditional intelligence diagnosis model. Then taking the concrete girder for example, the modal parameter and correlated parameter are regarded as the input and the damage location and damage degree are regarded as the output to build the adaptive mapping model. Meanwhile, the kernel function parameters are optimized by using HPSO in order to improve the generalization ability of LS-SVM. Finally, the result of example shows that LS-SVM has the ability of high identification precision.