在目前临床脑电医生常用的脑癫痫信号识别方法基础上，提出了一种利用计算机自动识别脑癫痫信号的方法及实现过程，该方法通过对脑电信号的形态特征的提取与在时空域进行综合分析的方法，模拟脑电专家的逻辑推理过程，将脑癫痫信号分为５类：尖波、棘波、尖慢波、棘慢波和多棘慢波来识别。经３０例临床脑电信号检验，本系统检出率达９０．３８％，证明了系统的有效性。 Based on the current methods for identifying epilepsy signals in clinical EEG, a method of automatic recognition of epilepsy signals by computer is proposed. The method is based on the extraction of the morphological features of EEG signal and the implementation in the time-space domain A comprehensive analysis of the method to simulate the EEG expert logic process, the epilepsy signals are divided into five categories: spikes, spikes, sharp slow waves, Spike slow wave and spike wave to identify. After 30 cases of clinical EEG signal detection, the detection rate of this system is 90.38%, which proves the effectiveness of the system.