李海涛¹, 刘泰麟¹, 邵泽东¹, 黄海广²
1.青岛科技大学信息科学技术学院, 山东 青岛 266061;2.温州科技职业学院, 浙江 温州 325000

摘要:
针对赤潮灾害等级预测难的现状，提出了一种基于C4.5决策树与二分分割算法优化的BP（反向传播）神经网络赤潮等级预测模型。该模型针对传统BP神经网络输入参数难以选择和隐含层节点数量难以确定的问题，通过决策树分类获取最优的属性组合，来解决输入参数难以选择的问题；通过“二分分割算法”，来解决隐含层节点数难以确定的问题。实验结果表明，该模型在青岛近海海域赤潮灾害等级预测中，预测结果的均方根误差（RMSE）小于传统BP神经网络的预测误差，并且在网络训练时间上有所缩短，预测精度上有所提高，能够获得良好的预测结果，可为赤潮等级预测提供新的解决方法。
关键词:赤潮灾害等级预测C4.5决策树算法二分分割算法BP神经网络
DOI：10.11759/hykx20181130001
分类号:X824
基金项目:青岛市创新创业领军人才（15-07-03-0030）；农业部水产养殖数字农业建设试点项目（2017-A2131-130209-K0104-004）

Application of a BP neural network based on decision tree and two partition algorithm in prediction of red tide severity
LI Hai-tao¹, LIU Tai-lin¹, SHAO Ze-dong¹, HUANG Hai-guang²
1.School of Information Science and Technology, Qingdao University of Science and Technology, Qingdao 266061, China;2.Wenzhou Vocational College of Science and Technology, Wenzhou 325000, China
Abstract:
To predictred tide severity, we propose a back propagation (BP) neural network model based on a C4.5 decision tree and a dichotomous algorithm. To solve the problems associated with selecting input parameters and determining the number of hidden layer nodes when usingthe traditional BP neural network, the proposed model uses decision-tree classification to obtain an optimal combination of attributes and a "two-partition algorithm" todetermine the number of hidden layer nodes. The experimental results showed that the root mean square error (RMSE) of the prediction results of the proposed model is less than that of the traditional BP neural network, and the training time of the network is shortened.The prediction accuracy is also improved, so accurate prediction results can be obtained.This research providesa novelapproach for the prediction of red tideseverity.
Key words:red tide disaster grade predictionC4.5 decision tree algorithmtwo-partition algorithmBP neural network





PDF全文下载地址:

点我下载PDF