In the realm of renewable energy, biomass plays a crucial role. A key component of power plants, the biomass boiler unit, is responsible for steam production. This unit operates as a nonlinear, highly coupled multivariable process. Traditional controllers used in the industry are ineffective for such systems. To address this, this paper presents a novel approach: a model predictive controller designed for biomass boiler plants. Fuzzy modelling, employed to approximate nonlinear functions to linear ones, is used for system identification. The methodology is implemented using MATLAB/Simulink and the Fuzzy modelling and identification (FMID) toolbox, utilizing input-output data from the Wenji-Shoa sugar factory for fuzzy model identification. The proposed controller demonstrates significant improvements, achieving settling times of 7.5, 13, and 7 seconds, with acceptable overshoots of 0.5%, 0.39%, and 0.46% for pressure, temperature, and level, respectively, for MISO systems. In contrast, the MPC shows improved performance in MIMO systems compared to MISO systems, with settling times of 5, 4, and 7 seconds, while the overshoot is reduced only for the pressure output, with 0.214%.