In order to improve the classification accuracy of EEG signals, a classification method based on an artificial bee colony algorithm and back propagation (BP) neural network is implemented. In order to improve the poor global search abilities and sensitivity to initial weights of BP neural networks, the global search factor is used to enhance an artificial bee colony algorithm search formula, which is proficient in exploration but required further development. A crossover operation is used to improve the global search capacity of the artificial bee colony algorithm. This enhanced algorithm is further used to optimize the sensitivity of the BP neural network to initial weights, enabling classification of EEG signals. The experiment results show that the proposed algorithm produces a highly accurate EEG signal classification of 91.5% with an accelerated convergence speed.