In gas-phase polyolefin processes, it is important to evaluate the melting temperature of particles since exceeding this temperature may cause particle sticking and aggregation. In this work, a model combining thermodynamic aspects and representing the evolution of the physical properties of the polymer is used to predict the melting onset temperature of polyethylene particles in a fluidized bed reactor. In this way, the model accounts for the effects of the polymer density and particle swelling by penetrants on the melting temperature. This model is then used within an optimization strategy to control the transition between different polymer grades, while avoiding particle sticking. The controlled properties are the polymer density and melt index, and the manipulated variables are the flow rates of hydrogen and comonomer and the bed temperature. Constraints are considered on the upper and lower limits of the flow rates. The bed temperature was constrained to remain lower than the polymer onset melting temperature, with a safety margin. It is shown that controlling the properties while respecting the constraints is feasible over a specific range.