基于减聚类ANFIS模型的船舶横摇运动实时预测  

Real time prediction of ship rolling motion based on subtractive clustering ANFIS model

作　　者：张泽国[1] 尹建川[1] 胡江强[1] 李可[1] 皇甫国光 董显利 

机构地区：[1]大连海事大学航海学院,辽宁大连116026

出　　处：《上海海事大学学报》2017年第1期7-11,67,共6页Journal of Shanghai Maritime University

基　　金：国家自然科学基金(51279106;51379002);交通运输部应用基础研究项目(2014329225010);中央高校基本科研业务经费项目(3132016116;3132016314);辽宁省教育厅一般项目(L2014214)

摘　　要：为准确高效地预测船舶在波浪中的航行状态以保证人员、货物和船舶的安全,提出一种基于减聚类的自适应神经模糊推理系统(Subtractive Clustering based Adaptive Neural-Fuzzy Inference System,SC-ANFIS)模型.SC-ANFIS模型使用减聚类算法对输入样本进行聚类分析,得到模糊规则数,并建立神经模糊推理系统,再使用结合BP算法与最小二乘估计算法的混合算法对建立的预测系统进行优化训练,得到最优的预测系统模型,并使用自相关分析确定预测系统模型的输入.运用该模型对大连海事大学科研教学船"育鲲"号的横摇运动进行实时预测,结果验证了该方法可行、有效,并具有较高的预测精度.In order to predict navigation states of ships in wind and waves accurately and efficiently and ensure the safety of personnel, cargoes and ships, a Subtractive Clustering based Adaptive Neural-Fuzzy Inference System （SC-ANFIS） model is proposed. In the model, the subtractive clustering algorithm is used to carry out clustering analysis on input samples so as to obtain the fuzzy rule number, and then the neural-fuzzy inference system is built. The hybrid algorithm combining BP algorithm with the least squares estimation algorithm is used to train and optimize the system so as to establish an optimal predic-tion system model. The input of the prediction system model is determined by auto-correlation analysis.The rolling motion of the ship Yukun （a scientific research and training ship of Dalian Maritime Universi-ty） is predicted in real time by the model. The prediction results demonstrate that the proposed method is feasible, effective and of high prediction accuracy.

关 键 词：船舶横摇运动 时间序列预测 自适应神经模糊推理系统 减聚类 

分 类 号：U661.321[交通运输工程—船舶及航道工程] TP183[交通运输工程—船舶与海洋工程]