基于最小二乘支持向量机模型的混凝土结构劣化评价新方法

Concrete Structure Deterioration Evaluation Based on LS-SVM Model

针对混凝土结构劣化评价的复杂性,提出了一种基于最小二乘支持向量机(LS_SVM)的混凝土结构劣化评价方法,并结合搜集到的既有混凝土结构劣化评价实例数据,选取其中的一部分作为训练样本对LS_SVM进行训练,利用网格搜索和交叉验证法确定LS_SVM模型的最优参数;另一部分作为检验样本进行测试。实例计算结果表明,该方法在劣化评价中耗时短、结果准确,适用于已有较多评价实例的同类混凝土结构或构件的劣化评价。

Aiming at the complexity of concrete structure deterioration evaluation,a evaluation method of concrete structure deterioration was proposed based on least-squares support vector machine（LS-SVM）.Combined with the instance data collected of deterioration evaluation,apart of collected evaluation data were used as the training sample to train the LS-SVM,and the grid-search and cross-validation were adopted to determine optimal parameters of LS-SVM model.The other part was used as the test sample.The calculation result from an actual case shows that the method takes a short time and has high precision in the evaluation of deterioration.It applies to deterioration evaluation of similar concrete structure and component which have many evaluation examples.