收集整理了我国大陆东部地区44 组水库地震序列，其中震群型32 组，主余型和孤立型12 组。分别计算了44 组序列的h 值、b 值、归一化熵值K、能量均匀度U、地震发生方式参数ρ 等5 个序列参数。在95% 的置信水平下，对每一参数分震群型、主余型加孤立型两大类进行差异性检验。结果表明，h、U、ρ 值对水库地震序列类型的分类能力不强;b 值平均值虽存在一定差异，但数值分布范围有部分重叠;K 值差异性显著，K = 0. 35 可作为区分震群型与主余型加孤立型两类水库地震序列的判别指标。在此基础上，建立以上述5 个参数为自变量的Fisher 判别函数，用于序列类型的综合判定。结果显示，所有数据全部参与建立判别函数并进行回溯性内符判别检验，识别正确率为97. 6% ;利用32 组序列参数参与判别函数建立，另外10 组进行外推检验，识别正确率为100% ，表明上述综合判别方法具有较高的水库地震序列分类能力。

44 reservoir earthquake sequences in the east of Chinese mainland have been collected． Among them，32 are multiple mainshock type and 12 belong to mainshock-aftershock sequences or isolated earthquakes． 5 parameters，h value，b value，energy entropy K，energy uniformity U，earthquake mode ρ，of these sequences have been calculated． With the 95% confidence level，average value of each parameter and its confidence limitation have been calculated for above two types． Statistical results show that h，U，ρ express low capability for classification of reservoir earthquake sequences． Though the mean values of b between above two types are different，data ranges get a certain overlap． The value K of above two types has a significant difference，and K = 0. 35 can be the criterion index to distinguish the multiple mainshock type sequences and ( mainshock-aftershock sequences + isolated earthquakes) ． In addition，with five parameters of different sequences as the independent variables，Fisher discriminant functions have been established for the synthetical judgment of sequence types． The results show that the classification accuracy of data is 97. 6% when all the data involved in establishing discriminant function; the classification accuracy of data is 100% when 32 groups of data involved in establishing discriminant function and 10 groups of data as extrapolation inspection． It indicates that the Fisher discriminant method has high ability on classification of the reservoir earthquake sequences．

国家“十二·五”科技支撑计划项目专题(2012BAK19B02-05)资助