基于双重相似度检索的舰船装备维修成本案例推理预测

Case-based reasoning of forecasting ship equipment maintenance cost based on double similarity retrieval

由于舰船装备维修成本属性间关系复杂、案例数量不断增加等特点,案例推理方法在案例检索过程存在信息单一、检索效率不高等问题。为解决该问题,首先根据舰船特征属性的数据特点,将条件属性分类,分别构建了相应的相似度函数;其次,考虑属性间关系,分别采用改进模糊粗糙集和熵权法确定条件属性权重并构建组合权重;然后,提出了两级分类与双重相似度匹配的案例检索策略,逐级缩小了检索范围,提高了检索效率;最后,基于两种相似度进行案例检索,使检索结果包含了两种相似度度量下的信息,并定义案例密度辅助检索过程的推进,增强了结果的客观性。实例分析表明:该方法能够提高舰船装备维修成本的预测准确度。

Due to the complex relationship between ship equipment maintenance cost attributes and the increasing number of cases,CBR has some problems in case retrieval process,such as singleness of information and low retrieval efficiency.According to the data characteristics of ship feature attri-butes,the conditional attributes were classified and the corresponding similarity functions were constructed respectively.Considering the relationship among attributes,the improved fuzzy rough set and entropy weight method were used to determine the weight of conditional attributes and construct the combined weight.And then,a two-stage classification and dual similarity matching case retrieval strategy was proposed so as to reduce the retrieval scope step by step,and thus to improve the retrie-val efficiency.The retrieval result contains information under two kinds of similarity measure based on two kinds of similarity case retrieval,and the case density was defined to promote the retrieval process,which can enhance the objectivity of the results.An example shows that this method can improve the forecasting accuracy of ship equipment maintenance cost.