The invention discloses a method of a global sliding mode control of a neural network of a micro-gyroscope, which comprises the following steps of establishing a global sliding mode control system of the neural network, designing a control law and taking as a control input of the micro-gyroscope, designing a self-adaptive law based on an Lyapunov function theory, and verifying the stability of a closed-loop system. The global sliding mode control is realized by designing a dynamic nonlinear sliding mode surface, the defect of no robustness in a reach movement stage of the sliding mode control is eliminated, the system has robustness in the whole process of response, and the respective defects of the global sliding mode control and the neural network are reduced by utilizing the intelligent control function of global sliding mode and the neural network. According to the method, the selection of sliding mode coefficients is simplified, the transient performance and the robustness of the sliding mode control system are improved, so that the closed-loop control system has global robustness, buffeting in the sliding mode control is eliminated, and thus a powerful foundation is provided for expansion of an application range of the micro-gyroscope.