The invention discloses a quick start method of an MEMS (microelectromechanical systems) gyroscope based on parallel estimation composite learning and is intended to solve the technical problem that existing MEMS gyroscope modal control methods are poor in practicality. According to the technical scheme, a neural network prediction error is constructed according to a parallel estimation model anda kinetic model; the neural network prediction error is combined with a tracking error to design composite adaptive law of neural network weight, a weight coefficient of the neural network is corrected, and effective dynamic estimation of unknown kinetics is achieved; a sliding mode controller is designed according to a sliding mode hypersurface and double-exponential reaching law, feedforward compensation of unknown kinetics is achieved, a vibration error of a detection mass block is quickly converged, and the need for quick start of the gyroscope is met. BY designing the composite adaptive law of neural network weight, correcting the weight coefficient of the neural network and introducing the sliding mode controller designed based on the sliding mode hypersurface and the double-exponential reaching law, the vibrational error of the detection mass block is quickly converged, and the need for quick start of the gyroscope is met; the method of the invention has good practicality.