摘要
在船舶风险管理中,为从已有的风险监测实例中挖掘出潜在的风险诊断知识,将人工智能的思想应用于风险状态识别,利用基于信息熵的决策树的知识获取方法从监测实例中推理风险与相应特征的对应关系。采用主元分析方法约简条件属性,采用C4.5算法度量风险监测实例表中各条件属性对状态识别的重要性,建立基于状态监测实例库的风险诊断推理模型。通过运用此模型对船体腐蚀程度进行风险监测管理,验证了诊断推理的效果,相应的信息处理流程为船舶的持续风险监控提供了手段支持。
The artificial intelligence method of knowledge acquisition for ship risk management to extract hidden diagnostic knowledge from the risk monitoring instances is introduced. The knowledge about condition-risk relations are acquired from monitor instances through building decision trees based on information entropy. The attributes of observations are reduced by principal component analysis and the importance of each attribute for the condition identification is measured by the C4. 5 algorithm afterwards. With the results of the processing, the inference model for ship risk diagnosis based on monitor instances is established. The method has been used to predict ship hull corrosion and proved to be effective. The process is suitable for continuous risk monitoring and management.
出处
《中国航海》
CSCD
北大核心
2014年第4期84-87,共4页
Navigation of China
基金
浙江省重大科技专项(2013C03033)
舟山市科技计划项目(2013C11015)
浙江省自然科学基金(LQ14E090001)
浙江海洋学院科研启动经费资助(201185011513)
关键词
水路运输
船舶风险诊断
推理系统
知识获取
决策树
waterway transportation
ship risk diagnosis
reasoning system
knowledge acquisition
decision tree
