基于ABC-BP神经网络引航量预测研究

发布时间：2018-04-24 16:19

本文选题：BP神经网络 + 人工蜂群算法　；参考：《大连海事大学》2017年硕士论文

【摘要】：引航量是指到港船舶的引航艘次,是港口发展建设综合评估中的一个重要指标。引航工作是港口安全和服务不可或缺的重要环节,是国际航运的重要组成部分和水上国门第一形象。引航安全关系到国际声誉、主权维护和政府形象,关系到人命财产和水域环境的安全。引航量的预测研究能够为各个引航站的发展规划和人力资源布局提供合理的数据、决策支持,从而使得港口能够高速、安全的发展。在人工神经网络的应用中,BP神经网络是其中应用最为广泛的一种,但是在BP神经网络模型中由于其自身的原因造成BP神经网络在训练学习过程中存在容易陷入局部最小值、收敛速度慢等不足之处。人工蜂群算法是模拟蜜蜂采蜜的过程,该算法是根据蜜蜂采蜜的原理以及蜜蜂之间相互交流的方式而提出的一种新的智能算法,人工蜂群算法具有较强的全局和局部搜索能力,利用该优点,可以将BP神经网络求解各层最优的权值和阀值的这一过程转化为蜜蜂采蜜过程中搜寻最佳蜜源的过程,并应用于引航量的预测。本文首先介绍了人工神经网络以及人工蜂群算法,然后通过查阅参考文献和采用专家调查表法确定了影响引航量的因素,包括有:GDP、港口吞吐量、固定资产投资额。在此基础建立了 ABC-BP引航量预测模型,并与BP神经网络训练以及验证预测结果进行了分析对比,得出ABC-BP神经网络模型训练过程稳定,误差较小,并且训练过程中收敛速度较快,该模型稳定可靠。以深圳港引航站为实例,利用ABC-BP神经网络对深圳港进行引航量预测研究,通过调查问卷确定影响深圳港引航量因素,设置相关参数,以2001-2013年深圳港引航站引航量为训练样本,2014-2016年引航量为验证样本。最后利用已训练好的ABC-BP神经网络以及各因素平均增长率,对深圳港2020年和2030年的引航量进行了预测。
[Abstract]:Pilotage refers to the pilotage of ships arriving at port and is an important index in the comprehensive evaluation of port development and construction. Pilotage is an indispensable part of port security and service, an important part of international shipping and the first image of the waterfront. Pilotage safety is related to international reputation, sovereignty maintenance, government image, life and property, and the safety of water environment. The prediction of pilotage can provide reasonable data and decision support for the development planning and human resource layout of each pilot station, so that the port can develop at a high speed and safely. In the application of artificial neural network, BP neural network is one of the most widely used, but in the BP neural network model, because of its own reasons, the BP neural network is prone to fall into local minimum value in the process of training and learning. The convergence speed is slow and so on. Artificial bee colony algorithm is a new intelligent algorithm based on the principle of honeybee honey collection and the way of honeybee communication. Artificial bee colony algorithm has strong global and local search ability. By using this advantage, the process of solving the optimal weight and threshold value of each layer by BP neural network can be transformed into the process of searching for the best honey source in the honeybee honey harvesting process, and applied to the prediction of pilotage. In this paper, the artificial neural network and the artificial bee colony algorithm are introduced at first, and then the factors affecting pilotage are determined by reference and expert questionnaire, including the proportion of GDP, port throughput and the investment of fixed assets. On this basis, the prediction model of ABC-BP pilotage is established and compared with BP neural network training and verification and prediction results. It is concluded that the training process of ABC-BP neural network model is stable, the error is small, and the convergence rate of the model is faster during the training process. The model is stable and reliable. Taking the pilotage station of Shenzhen Port as an example, using ABC-BP neural network to study the pilotage prediction of Shenzhen Port, the factors affecting pilotage volume of Shenzhen Port are determined by questionnaire, and the relevant parameters are set up. The pilotage of Shenzhen Port from 2001 to 2013 is taken as the training sample and the pilotage from 2014-2016 as the verification sample. Finally, using the trained ABC-BP neural network and the average growth rate of various factors, the pilotage of Shenzhen Port in 2020 and 2030 is forecasted.
【学位授予单位】：大连海事大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U692

【参考文献】