中文摘要：

为分析冲击地压发生前的电磁辐射前兆规律，以鹤岗南山矿237工作面“1．12冲击”电磁辐射前兆序列为实例，运用SAS统计分析软件检验序列的随机性和平稳性，建立了适用于该序列的ARMA拟合模型，并预测了序列未来12期的监测值．结果表明，该序列是一个平稳非白噪声序列，可以用AR（3）模型来拟合和预测，模型预测值和真实值的相似程度在66％左右，一致性较高．ARMA模型能够提取电磁辐射序列中的前兆信息，定量描述序列的变化趋势．

英文摘要：

SAS statistical analysis software was used to test the randomness of electromagnetic radiation （EMR） observed during the ＂1.12 rock burst＂ of the number 237 working face in the Nanshar/coal mine. An auto-regressive-moving-average （ARMA） model was fitted to the EMR data and used to forecast twelve observations into the future. The results show that the rock burst EMR data are non-white noise, stationary and can be fitted with an AR（3） model. Comparing the model EMR values to the real data, the similarity degree is about 66%. An ARMA model can use data preceding an event to describe changes in the EMR trends quantitatively.