作　　者： ; ;

机构地区： 华南理工大学化学与化工学院传热强化与过程节能教育部重点实验室

出　　处： 《海洋工程》 2011年第4期87-91,共5页

摘　　要： 奥里油盛产于委内瑞拉,是一种煤、石油等潜在的替代燃烧物,近年来已经应用于多个国家的电厂锅炉、工业炉等燃油场合。随着应用量的增大,其海上运输产生的溢油风险也越来越大。针对奥里油溢油现象并基于图像处理及边缘检测提出了一种奥里油溢油扩散面积预测的方法。在Matlab环境下,对图像进行边界提取,面积计算,最后利用BP神经网络进行面积预测。通过与实验进行比较,结果表明,采用Sobel算子进行边缘检测时提取的边缘较细、对边缘定位较准、对灰度渐变和噪声处理较好。采用BP神经网络获得的面积预测结果与实验测得的结果最大相对误差为0.88%,吻合度较好。 Orimulsion is originally developed in Venezuela.It is a potential fuel which can replace coal and petroleum.In recent years,Orimulsion has been applied to many fuel places such as boilers in power plants and industrial furnaces in severam countries.With the increase of its quantity and fields of application,the risk of Orimulsion spill caused by marine transport is also increasing.Focusing on this phenomenon of Orimulsion spill,a nonlinear processing is put forward to forecast the area of Orimulsion spill based on image processing and edge detection.It means that with the help of MATLAB,the image edges of Orimulsion spill are detected,the area of Orimulsion spill is calculated,and the next time area of Orimulsion spill is forecasted using BP neural network.At the same time,the data from these methods is compared with the data obtained from the experiments.The results show that the edge extracted by Sobel operator is thinner,more precise than any other operators and it is better to deal with the noise and gray changes using Sobel operator.The results also show that forecasted area data by BP neural network closely match the experimental area data with the maximum relative error being only 0.88% between them.

关 键 词： 奥里油 溢油扩散 面积预测 神经网络

领　　域： [环境科学与工程]