遗传极限学习计算法在船舶碰撞危险度确定中的应用  

Application of genetic limit learning algorithm in determination of ship collision risk

作　　者：罗捷 王德岭 LUO Jie;WANG De-ling(Shanghai Maritime University,Shanghai 201306,China)

机构地区：[1]上海海事大学,上海201306

出　　处：《舰船科学技术》2021年第18期34-36,共3页Ship Science and Technology

基　　金：华北电力大学中央高校基金项目(2016MS125)

摘　　要：为防止船舶航行过程中出现碰撞问题,研究遗传极限学习计算法在船舶碰撞危险度确定中的应用。以船员视角、两船间的安全距离和航行速度为基础,利用最小安全时间直观描述船舶碰撞危险度;根据船舶航行样本数据,利用极限学习机预测最小安全时间,采用遗传算法优化极限学习机,通过初始化种群、适应度运算、交叉变异等过程确定最优染色体,提升最小安全时间预测精度。实验结果显示所研究方法遗传迭代25次后误差平方和降至2.5,不同海域的防碰撞操作控制成功率达到97.5%,说明该方法具有较快的迭代速度与较好的应用效果,可有效保障船舶航行过程中的安全性。In order to prevent collision and ensure safe navigation of ships,the application of genetic limit learning calculation method in determining collision risk of ships is studied.Based on the crew’s perspective,the safe distance between the two ships and the sailing speed,the collision risk of the ship is intuitively described with the minimum safe time.According to the ship voyage sample data,the ultimate learning machine was used to predict the minimum safe time,and the genetic algorithm was used to optimize the ultimate learning machine.The optimal chromosome was determined through the process of initialization population,fitness operation and crossover mutation,so as to improve the accuracy of the minimum safe time prediction.The experimental results show that the sum of squares of errors decreases to 2.5 after 25 genetic iterations,and the success rate of anti-collision operation control in different sea areas reaches 97.5%,which indicates that the method has a faster iteration speed and better application effect,and can effectively guarantee the safety of ships in the process of navigation.

关 键 词：遗传极限学习 船舶碰撞 安全距离 最小安全时间 神经元 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]