Abstract: Wet coal preparation is the main coal preparation method in China, which results in the mixing of water and coal slurry to form a large amount of coal slurry water during the coal preparation process. The dosing treatment of smile water plays a very important role in coal preparation plants. The accuracy of dosage directly affects the economic benefit of coal preparation plants. In this paper, model predictive control (MPC) algorithm is designed for flocculant and coagulant dosing process to achieve the setpoint tracking. On this basis, the problem that model parameters will change due to equipment aging is considered, and the recursive least squares (RLS) algorithm is introduced to identify the system model online. So as to improve the control performance of the system. Experiments have been carried out on a coal smile water dosing process with actual data, which shows the effectiveness of the proposed method.