模式识别法最近已用来进行地震预测.我们用其中一种无监督分类方法——集群或聚类分析法,研究了华北和宁夏两地区,其结果表明:有序聚类比系统聚类结果好,此结果与交替的地震活动高潮与低潮相对应,从数学的角度有效地模拟了地震成组活动的现象;误识率分析及预测检验表明误识率较低且结果较稳定. Pattern recognition has recently been used for earthquake prediction.We used one of the unsupervised classification methods, cluster or cluster analysis, to study the North China and Ningxia regions, and the results show that the ordered clustering results are better than the systematic clustering results Well, this result corresponds to the climax and alternation of alternating seismic activities, and effectively simulates the phenomenon of grouping of earthquakes from the perspective of mathematics. The analysis of misidentification rate and the test of prediction show that the error rate is low and the result is stable.