基于事例推理系统中的模糊检索

Fuzzy Retrieval in Case-Based Reasoning Systems

针对现有建筑结构初步设计专家系统的检索方法存在的不足,介绍了一种基于事例推理的新检索方法.该方法基于模糊相似优先的实例检索思路,采用海明距离的模型描述属性间距离,更好地反映了不同事例属性值的差异,并赋予属性权重系数更加直观的意义,使之更容易选取,适用于普通数据库(非模糊数据库)中的数量型属性的检索.结合一个高层建筑结构初步设计结构选型的工程案例,进行事例检索,说明了新方法的可行性和合理性.

A new method on building of expert systems in structural preliminary design was introduced, which is so-called case-based reasoning. Some disadvantages on retrieval strategies for this method were analyzed. According to the case indexing model, which is based on fuzzy analogy, the definition of the retrieval model and that of attribute distances between different cases are modified and the Haming distance is adopted, which represents the difference between the value of two case attributes better and can be used in the retrieval of quantity attributes. In this new method, the signification of the weight of a feature is more intuitionistic and can be defined more easily. A case retrieval example of structural scheme in structure preliminary design of high-rise buildings was given, which shows the feasibility and rationality of the new method.