混凝土重力坝抗震可靠度的SVM预测模型

Support vector machine model for prediction of seismic reliability of large concrete dams

研究混凝土大坝抗震可靠性意义重大。混凝土大坝可靠性分析中,大多数功能函数不一定能显式给出,目前一般使用几何法求解,由于存在各种假定和近似,几何法不是一种完备的求解方法,对非线性问题求解繁琐甚至困难。由于支持向量机(SVM)的非线性映射能力及对残缺不全或模糊随机的不确定信息较强的容错能力,提出采用优化的SVM模型来预测大坝的抗震可靠度。并应用于采集到的22个混凝土重力坝断面数据分析中,研究结果表明:该模型是合理可靠的,用SVM测大坝抗震可靠度是可行的。

Because it is impossible to get analytical expression of performance function in the seismic reliability analysis of large concrete dams, geometric method is used.But there are hypothesis and approximate in geometric method, so it isn＇t a maturity method.The SVM model is proposed to predict the seismic reliability of large concrete dams.Based on collected data from 22 concrete gravity dams, practical effectiveness of SVM model to predict the seismic reliability of large concrete dam is discussed.Results show that the predicted and measured results are consis- tent, showing that the model can provide a basis for prediction of seismic reliability of large concrete dams.