摘要
为了提高船舶交通流预测的效率和准确率,分析了船舶流量预测中的影响因素多、非线性、随机性等问题,建立了ELM(极限学习机)预测模型。同时为了避免极限学习机算法受输入权值矩阵和隐含层偏差随机性的影响,算法又采用GA(遗传算法)对极限学习机的输入权值矩阵和隐含层偏差进行优化,建立GA-ELM船舶交通流预测模型。利用上海洋山港船舶流量对该模型进行了实例分析,通过MATLAB仿真进行预测,将GA-ELM模型与单纯的BP模型、ELM模型进行对比和分析,结果表明:GA-ELM模型具有更高的预测精度和效率,从而能够相对准确、高效地对船舶交通流量进行预测。
In order to improve the efficiency and accuracy of vessel traffic flow prediction,by analyzing the influence factlinear and randomness problems and so on,establishing ELM （ Extreme Learning Machine） prediction modvoid ELM influenced by input weight matrix and the randomness of hidden layer deviation, using the genetic algorithm to optimize the inputweight matrix and hidden layer deviation of the ELM,establishing GA-ELM vessel traffic flow prediction model. Using Shanghai Yangshan Portas case analysis for GA-ELM vessel traffic flow prediction model, by MATLAB simulation to make a comparison and analysis of GA-ELM model and simple BP model, ELM model. The results show that GA-ELM model has higher prediction accuracy and beter generalization ability so that it can be used to predict the vessel traffic flow relatively accurately and eficiently.
作者
崔翔鹏
黄洪琼
Cui Xiangpeng Huang Hongqiong(College of Information Engineering, Shanghai Maritime University, Shanghai 201306, China)
出处
《微型机与应用》
2017年第9期15-17,21,共4页
Microcomputer & Its Applications
基金
国家自然科学基金(61673260)
关键词
船舶交通流量
遗传算法
极限学习机
预测
ship traffic flow
genetic algoritjims
extreme learning machine
prediction
