<p>Industrial production processes typically constitute complex systems featuring operating conditions and hierarchical structures. These complexities, along with the inherent nonlinearity, uncertainties, and constraints involved, pose significant challenges to product quality and economic performance. To address these issues, a double-layer adaptive predictive optimization control strategy is proposed. At the upper layer, an improved adaptive genetic algorithm (IAGA) is employed to solve an optimization problem with a nonlinear economic objective function, providing optimal setpoints online for various production scenarios. These setpoints are then passed to the lower layer, where an adaptive model predictive control (MPC) approach is used to accurately track the decomposed optimal setpoints under system uncertainties and constraints. Moreover, the recursive feasibility and stability of the adaptive MPC can be theoretically demonstrated. Simulation studies demonstrate the effectiveness of the proposed strategy.</p>