基于SVR的船舶溢油事故预测

船舶溢油事故的预测,是确定所需应急资源多少的重要依据;但船舶溢油事故具有小样本性,传统预测方法不再适用。针对船舶溢油事故的小样本性,该文选择SVR模型对其进行预测;SVR算法的关键是核函数的选择及自由参数的确定,针对上述问题该文分别采用交叉验证法、遗传算法;并在matlab软件上建立SVR模型,最后将该模型应用于上海港水域船舶溢油事故的预测。

SVR Model Forecast of Ship Oil Spill Accidents

The forecast of ship oil spill accidents was very important data which determining the quantity of emergency resources required. Based on Support Vector Regression （SVR） model, a new model was built to forecast ship oil spill accidents as these accidents had small sample size. The key points of SVR were the selection of kernel functions and free parameters. Therefore, cross validation method and genetic algorithm are applied ; And SVR model is constructed on Maflab for the forecast of ship oil spill accidents in Shanghai port.