Abstract—Electroencephalography (EEG) is an electrophysiological monitoring method to record electrical activity of the brain. EEG-based control is increasingly being discovered by many researchers aims to support disabled people. In order to build control command set, the system will classify the EEG signal received of user looking at the different types of images. The good results of EEG signal classification will help make control more effectively. In this paper, a novel approach proposes the classification of EEG signals based on Wavelet transform, K-means clustering algorithm and Multi-Layer Neural Network. The system architecture was designed and evaluated with the dataset of 21,000 samples. The best accuracy rate can obtain 93.57 %.