In order to optimize the preparation of sintered tumbling blocks and improve the quality of the blocks, on the basis of the analysis of the sintering process characteristics of tumbling mill blocks and the application of non-minimum state space model prediction (NMSSMPC) in ethylene cracking furnace, a Kalman filter-based non-minimum state space model prediction control (KF-NMSSMPC) method was propased and applied to the tumbling mill sintering furnace. The control algorithm eliminates the effects of measurement noise and channel noise during sintering by adding filters to the Model Predictive Control (MPC). Numerous simulation results show that different from NMSSMPC system, KF-NMSSMPC can effectively suppress disturbances, quickly track sintering trajectories, reduce errors, and improve sintering control process accuracy under the imitation of actual industrial conditions. KF-NMSSMPC can improve the sintering temperature control accuracy of the rolling sintering furnace, which in turn improves the sintering quality of the rolling grinding block.