数据驱动的船舶智能故障诊断方法  

Data-driven Vessel Smart Fault Diagnosis method

作　　者：贾宝柱 贾志涛[2] 余培文 安连彤[1] JIA Bao-zhu;JIA Zhi-tao;YU Pei-wen;AN Lian-tong(Maritime college, Guangdong Ocean University, Zhanjiang 524088, China;Marine Engineering college, Dalian Maritime University, Dalian 116026, China)

机构地区：[1]广东海洋大学海运学院,广东湛江524088 [2]大连海事大学轮机工程学院,辽宁大连116026

出　　处：《控制工程》2019年第10期1892-1898,共7页Control Engineering of China

基　　金：国家自然科学基金资助(51479017);广东省“冲一流”省财政专项资金建设项目(231419010)

摘　　要：船舶机舱综合监控系统中存在的大量预示系统健康及故障状态的数据信息,针对数据中的属性冗余问题,定义了基于粗糙集的等价属性概念,并提出了一种基于等价属性的快速约简方法。以基本可信度及证据决策系数为依据衡量条件属性对决策属性的贡献度,据此进行证据推理最后得到最简属性决策表。对中央冷却系统测试数据的算例表明,所提出的方法对故障隐患状态的识别是有效的。There are huge of data in the engineroom integrated monitoring and control system, which prognosis the health and fault status of system. Aiming at the redundant condition attributes in raw datum, the equivalence attribute is defined based on rough set, and an effective attribute reduction method is proposed depend on equivalence attribute. The basic probability and evidential decision coefficient are used to measure the contribution of condition attribution to decision attribution. According to this, the minimization attribute decision table is derived after evidential reasoning. The calculating example with sample datum of central cooling water system shows that the proposed method is effective for system hidden fault diagnosis.

关 键 词：粗糙集理论 D-S证据理论 智能故障诊断 中央冷却系统 

分 类 号：U676.4[交通运输工程—船舶及航道工程] U672.7[交通运输工程—船舶与海洋工程]