基于遗传算法优化的虚拟机舱协作智能评估  

Intelligent assessment of virtual engine room collaboration based on genetic algorithm optimization

作　　者：段尊雷[1,2] 任光[1] 曹辉[1] 姚文龙[3] 

机构地区：[1]大连海事大学轮机工程学院,辽宁大连116026 [2]辽宁海事局船员考试中心,辽宁大连116001 [3]青岛科技大学自动化学院,山东青岛266041

出　　处：《哈尔滨工程大学学报》2017年第4期514-520,共7页Journal of Harbin Engineering University

基　　金：国家自然科学基金项目(51479017);中央高校基本科研业务费专项资金项目(3132016316)

摘　　要：为满足机舱资源管理的特殊要求,针对传统的轮机模拟器存在的问题,提出了任务型协作训练模式和基于遗传算法优化的智能评估方法。该方法包括建立驾机联动模拟的评估知识库;构造评估指标隶属度函数库和不同需求下的优化目标函数;采用熵权法和历史评估数据动态调整评估指标的权重并利用遗传算法优化;根据实时的系统参数检测结果和隶属度函数得出模糊关系矩阵;经多重模糊综合评判得出评估结果。在实例中对遗传算法优化的效果进行了对比分析。结果表明:评估结果比较客观,误差最小;提出的训练模式和评估方法科学合理,尤其是对于缺少实船服务经历的学员具有一定的实用价值。To meet the special requirements of engine room resource management and address the deficiencies of the traditional engine room simulator,we propose a mission-based collaborative training mode and an intelligent assess-ment method based on genetic algorithm. The proposed method involves building the assessment knowledge base of linkage simulation between the ship bridge and the engine room,constructing an evaluation index membership func-tion library and the optimization target functions for different requirements,and dynamically adjusting weights using the entropy weight method and the historical assessment data. The weights are then optimized with genetic algorithm. In addition,this method obtains the fuzzy relationship matrix based on the checking results of the system parameters in real time,as well as the membership functions,and obtains assessment results by multiple fuzzy comprehensive evaluations. We comparatively analyzed the effect of this genetic algorithm optimization in an example and confirmed the assessment results to be objective and having the least error. The proposed training mode and assessment method are scientific and reasonable,and they have practical value for marine college students, especially for those who lack seagoing service experience.

关 键 词：遗传算法 机舱 资源管理 协作 智能评估 隶属度函数 机舱资源 

分 类 号：U676.2[交通运输工程—船舶及航道工程]