Multifocal structured illumination microscopy (MSIM) can achieve optically sectioned images with twice the diffraction limited resolution at 1 Hz imaging speed and at imaging depths up to 50 ?m. Compared with the traditional wide-field SIM, MSIM has greater imaging depth and optical sectionning ability, it is more suitable for long-term 3D super-resolution imaging of living thick samples. However, MSIM has some problems, such as slow imaging speed and complex image post-processing process. In this paper, a fast super-resolution imaging method and system based on flat-field multiplexed MSIM (FM-MSIM) is proposed. By inserting a beam shaping device into the illumination light path, the Gaussian beam is reshaped into a uniform flat-top profile, so as to improve the intensity uniformity of excitation multi-spot focal array and expand the field of view; By elongating each diffraction limited excitation focal point four times along the Y direction to form a new multiplexed multifocal array pattern, reduce the number of scanning steps, improve the energy utilization, and then improve the imaging speed and signal-to-noise ratio. Combined with the sparse Bayesian learning image reconstruction algorithm based on multiple measurement vector model, the image reconstruction steps are simplified, At least 4 times of the imaging speed can be improved while ensuring the spatial resolution of MSIM. On this basis, the established FM-MSIM system is used to carry out the super-resolution imaging experiments of BSC cell microtubule samples and mouse kidney slices. The experimental results prove the fast three-dimensional super-resolution imaging ability of the system, which is of great significance for the development of fast MSIM.