﻿ 基于人工智能技术的船舶航迹点生成算法
 舰船科学技术  2001, Vol. 44 Issue (6): 129-132    DOI: 10.3404/j.issn.1672-7649.2022.06.026 PDF

1. 北京师范大学 互联网教育智能技术及应用国家工程实验室，北京 100875;
2. 内蒙古化工职业学院，内蒙古 呼和浩特 010070

Ship track point generation algorithm based on artificial intelligence technology
QI Hu-chun1,2
1. National Engineering Laboratory of Internet Education Intelligent Technology and Application, Beijing Normal University, Beijing 100875, China;
2. Inner Mongolia Vocational College of Chemical Engineering, Hohhot 010070, China
Abstract: The marine environment data model is constructed by raster method as a simulation graph generated by ship track points. On this basis, the shortest sailing distance and the least sailing obstacles are taken as the objectives to build a mathematical model of ship track point generation. Using adaptive crossover probability and mutation probability to optimize the fish swarm algorithm, the optimized fish swarm algorithm is used to determine the final solution of the mathematical model of ship track point generation and obtain the best ship track point. The experimental results show that this algorithm can accurately generate ship track points, which has obvious advantages compared with similar algorithms, and still has good track generation effect under the influence of weather environment.
Key words: artificial intelligence technology     ship's track     track point generation     optimized fish swarm algorithm     crossover probability     mutation probability
0 引　言

1 人工智能技术的船舶航迹点生成算法构建 1.1 船舶航行环境模型构建

 $E = \left\{ {{g_{ij}}\left| {{g_{ij}} = 0{\kern 1pt} {\kern 1pt} {\kern 1pt} or{\kern 1pt} {\kern 1pt} {\kern 1pt} 1} \right.,i \in \frac{L}{b},j \in \frac{W}{b}} \right\}\text{。}$ (1)

1.2 船舶航迹点生成算法数学模型

 ${l_{\max }} \geqslant {l_i} \geqslant {l_{\min }} \text{，}$ (3)
 $\forall l\left( {{p_i},{p_{i + 1}}} \right) \cap {S_j} = \emptyset \text{，}$ (4)
 $\forall \theta {p_j}_{ \in \left( {1, \cdots ,k} \right)} \leqslant {\varphi _{\max }} \text{，}$ (5)
 ${R_{{P_{j \in \left( {1, \cdots ,k} \right)}}}} \geqslant \lambda L \text{，}$ (6)
 $D \geqslant {D_{\min }} \text{。}$ (7)

1.3 基于优化鱼群算法的模型求解

 ${Q_m} = \left\{ \begin{gathered} {q_{m1}},f \leqslant {f_{avg}} \text{，}\hfill \\ {q_{m1}} - \left[ {\left( {{q_{m1}} - {q_{m2}}} \right)*\left( {{f_{\max }} - f} \right)/\left( {{f_{\max }} - {f_{avg}}} \right)} \right] \text{，}\hfill \\ \end{gathered} \right.$ (8)
 ${Q_c} = \left\{ \begin{gathered} {q_{c1}},f' \leqslant {f_{avg}}\text{，} \hfill \\ {q_{c1}} - \left[ {\left( {{q_{c1}} - {q_{c2}}} \right)*\left( {f' - {f_{avg}}} \right)/\left( {{f_{\max }} - {f_{avg}}} \right)} \right]\text{。} \hfill \\ \end{gathered} \right.$ (9)

1）构建符合航行状态参数，将迭代次数确定，航迹点生成过程中需要利用迭代次数；

2）从迭代参数角度出发，设置求解船舶航迹点生成数学模型算法中的适应度函数；

3）构建原始计划，各船舶航迹点生成计划与各个个体对应；

4）从适应度函数角度出发，实现适应度值计算，通过计算结果确定航迹点优劣；

5）根据步骤4的结果，确定下一步的最佳个体；

6）随机选择个体实现异变与交叉，选取最好的个体输入到下一代种群之中；

7）比较对数，如果对比结果中，大于最大值就可以结束求解，输出最优解，如果小于最大值就重新返回步骤4。

 ${f_s} = \frac{1}{{\displaystyle\sum\limits_{i = 1}^N {\left( {{\omega _1}{l_i} + {\omega _2}{f_{TAi}}} \right)} *\sum\limits_{i = 1}^N {d\left( {{p_i},{p_{i + 1}}} \right)} }} \text{，}$ (10)

 ${E_s} = \frac{{\displaystyle\sum\limits_{i = 1}^N {\left( {{\omega _1}{l_i} - {\omega _2}{f_{TAi}}} \right)} *\sum\limits_{i = 1}^N {d\left( {{p_i},{p_{i + 1}}} \right)} }}{{{q_c}{q_m}\sqrt {{f_s}} }} \text{。}$ (11)

2 实验结果与分析

2.1 海洋环境数据栅格模型构建

 图 1 海洋环境栅格化模型 Fig. 1 Rasterization model of Marine environment
2.2 船舶航迹点生成

 图 2 本文算法航迹点生成结果 Fig. 2 Track point generation results of the algorithm in this paper

 图 3 对比算法航迹点生成 Fig. 3 Track point generation of comparison algorithm
2.3 外部环境干扰下航迹点生成情况

 图 4 不同海洋环境下航迹点生成偏差 Fig. 4 Deviation of track point generation in different Marine environments
2.4 算法性能验证

 图 5 优化鱼群算法收敛过程 Fig. 5 Convergence process of optimized fish swarm algorithm
3 结　语

 [1] 郑振涛, 赵卓峰, 王桂玲, 等. 面向港口停留区域识别的船舶停留轨迹提取方法[J]. 计算机应用, 2019, 39(1): 113-117. DOI:10.11772/j.issn.1001-9081.2018071625 [2] 刘娇, 史国友, 杨学钱, 等. 基于DE-SVM的船舶航迹预测模型[J]. 上海海事大学学报, 2020, 41(1): 34-39+115. [3] 王森杰, 何正伟. 基于生成对抗网络的船舶航迹预测模型[J]. 中国航海, 2021, 44(2): 72-77. DOI:10.3969/j.issn.1000-4653.2021.02.012 [4] 王维刚, 初秀民, 蒋仲廉, 等. 基于加权朴素贝叶斯的船舶轨迹分类算法[J]. 中国航海, 2020, 43(4): 20-26. DOI:10.3969/j.issn.1000-4653.2020.04.004 [5] 姜佰辰, 关键, 周伟, 等. 基于多项式卡尔曼滤波的船舶轨迹预测算法[J]. 信号处理, 2019, 35(5): 741-746. [6] 曹伟, 刘亚帅, 管志强. 采用编码降维及DTW算法改进的船舶航迹聚类[J]. 现代防御技术, 2019, 47(5): 151-156. [7] 黄亮, 张治豪, 文元桥, 等. 基于轨迹特征的船舶停留行为识别与分类[J]. 交通运输工程学报, 2021, 21(5): 189-198. [8] 祝亢, 黄珍, 王绪明. 基于深度强化学习的智能船舶航迹跟踪控制[J]. 中国舰船研究, 2021, 16(1): 105-113. [9] 沈智鹏, 邹天宇, 王茹. 基于扩张观测器的欠驱动船舶轨迹跟踪低频学习自适应动态面输出反馈控制[J]. 控制理论与应用, 2019, 36(06): 867-876. [10] 陈天元, 袁伟, 俞孟蕻. 基于显式模型预测控制的无人船航迹控制方法[J]. 船舶工程, 2020, 42(9): 122-127. [11] 张黎翔, 朱怡安, 陆伟, 等. 基于AIS数据的船舶轨迹修复方法研究[J]. 西北工业大学学报, 2021, 39(1): 119-125. DOI:10.3969/j.issn.1000-2758.2021.01.015