船撞桥最小二乘支持向量机预测方法

为了提高桥梁与桥区通航船舶的安全性,提出了一种船撞桥概率智能预测方法.以桥墩跨径、水流速度、水流方向与桥墩连线法线方向夹角以及航道弯曲度为系统输入,以单航次船撞桥事故率为系统输出,应用最小二乘支持向量机进行了船撞桥概率估算.结合实际航道,选择了长江和黑龙江上12座桥梁的洪水期、中水期和枯水期3个时段的样本数据进行验算,并与神经网络船撞桥概率估算结果进行对比.对比结果表明:支持向量机方法能准确地预报船撞桥概率,具有全局最优解,并且收敛性和学习效率均优于神经网络.

Prediction method of ship impacting bridge based on least square support vector machines

In order to improve the safety of bridge and ship,an intelligent prediction method of the probability on ship impacting bridge was proposed based on least square support vector machines.The transverse span of piers,current velocity,current direction relative to bridge and waterway curvature were taken as the inputs of the method,while the output is the accident probability of pervoyage,some data samples of 12 bridges over Yangtze River and Heilongjiang River during the periods of flood,normal and low water respectively were chosen,the method was validated by the data,and the prediction results of neural network method and the proposed method were compared.Comparison result shows that the method can accurately predict the probability of ship impacting bridge,its prediction probability is a globally optimal resolution,and its convergent velocity and learning efficiency are better than that of neural network method.1 tab,3 figs,13 refs.