The invention relates to a personalized MI-EEG training and collecting method based on mirror image virtualization and Skinner reinforced learning. According to the method, a mirror image therapy in the rehabilitation therapy field and a Skinner reinforced learning theory are integrated into the MI-EEG training and collecting process; personalized audio-visual combined stimulation is designed in a training mode, an unaffected hand wears a data glove and performs opening and holding actions in a more relaxed state, data of the actions of the unaffected hand is read and decoded in real time through a computer serial interface, and then the actions of a mirror image virtualization affected hand model in a collecting interface is muzzled. According to the method, a subject stimulates and induces movement of the mirror image virtualization hand through the actions of the unaffected hand of the subject, the electrical activity of mirror image neurons in the movement function area of the brain is enhanced, and a closed-loop electroencephalogram collecting and controlling mode is designed on the basis of the Skinner reinforced learning theory to test the training effect; meanwhile, internal motivation is further enhanced through Skinner online reward feedback, and the MI-EEG quality is improved. The method is easy to operate and is expected to achieve better actual application.