Particles in a gas-solid separation fluidized bed have fluid-like properties in a fluidized state, and particle flow has a great influence on bed uniformity and stability. To study the complex spatial characteristics of particle flow, the macro flow characteristics of particles and the mass flow rate model for solid particle flow were studied with theoretical calculations and machine learning. The results showed that particle flow has obvious nonuniform characteristics in axial space. The theoretical solid mass flow rate model was then modified by introducing a bed expansion ratio. The theoretical calculation error was controlled to within 20%, and the machine learning error was within 15%. Based on these results, the machine learning method was found to be more accurate and comprehensive than the theoretical method. An artificial network model was also proposed to predict the spatial uniformity of a gas-solid fluidization system. The results showed that the accuracy of the training set was 94.9%, and that of the test set was 95.6%. Thus, the ANN model makes it possible to predict the spatial uniformity of a gas-solid fluidization system.