介绍了模糊联想存储器ＦＡＭ（ＦｕｚｚｙＡｓｓｏｃｉａｔｉｖｅＭｅｍｏｒｙ）的基本结构、ＦＡＭ的自适应学习算法以及ＦＡＭ推理机的原理，并成功地将其用于地震预报专家系统中．其成果表现为：在知识获取方面，用积空间聚类的方法，自适应地产生地震预报规则；在推理方面，用ＦＡＭ推理机实现了对结构性语言经验的综合推理． This paper introduces the basic structure of fuzzy associative memory (FAM), the adaptive learning algorithm of FAM and the principle of FAM inference engine, and successfully uses it in the earthquake prediction expert system. The results show as follows: In terms of knowledge acquisition, the method of product space clustering is used to generate the earthquake prediction rules adaptively. In the aspect of reasoning, FAM reasoning engine is used to synthesize the structural linguistic experience.