A Study On Prediction Of Pilotage Quantity Based On ABC-BP Neural Network

The quantity of pilotage is one important index for comprehensive assessment of port construction and.development.Pilotage duties are important parts of port safety&service and international shipping.It is also the first image of countries which famous for its water transportation.Pilotage safety associates with international reputation,the safeguarding of sovereignty,the image of government,the security of life properties as well as the security of marine.The forecasting research of pilotage quantity provides reasonable data about development projects and human resource distribution,and decision support for each pilot station.Thus,the development of port could be safe and rapid.Within the application of artificial neural network,BP is the most widely used neural network.However,owing to some reasons of the system itself,problems like local minimum and low convergence rate could be created during the learning process of BP neural network.Artificial bee colony algorithm simulates the process that a bee collects nectar.This new intelligent algorithm is proposed based on principles of nectar collecting and communication methods between bees.Artificial bee colony algorithm is good at local search and overall search.Taking use of this advantage,the process of searching for optimal weights and threshold of each layer by using BP neural network could be transferred into the process that a bee looking for the best nectar,and this could be applied for the forecasting of pilotage quantity.This paper introduces artificial neural network,artificial bee colony algorithm firstly.Secondly,it determines factors that do effects for the quantity of pilotage based on references and Delphi method.By comparing the predicted values and actual values that came out from those two models,predicated value error done by ABC-BP neural network is smaller and the convergent rate is faster,proves the superiority of the optimization algorithm..Taking pilot station in Shenzhen port as an example,it uses ABC-BP neural network to establish forecast models and set up parameters(take the quantity of pilotage for Shenzhen pilot station during 2001-2003 as training group;take the quantity of pilotage during 2014-2016 as the validation group)for the pilot station in Shenzhen port respectively.As a result,ABC-BP neural network could be used to establish forecasting models for the quantity of pilotage.be used to establish forecasting models for the quantity of pilotage.