The multiobjective long-term concentrate ingredient planning (MLCIP) plays an important role in modern smelting production measurement systems. MLCIP is a constrained multiobjective optimization problem that comprises a series of continuous ingredient stages and numerous decision variables under dynamic environments. The existing on-site ingredient planning methods primarily focus on satisfying short-term production constraints, while neglecting the resulting global impact on long-term objectives. Consequently, the overall duration of ingredient lists and the interest arising from concentrate hoarding cannot be estimated manually. Thus, we formulated MLCIP into a multistage-constrained large-scale multiobjective optimization problem, where the multiple objectives and constraints are formulated by the safety production rules from the smelting system. Furthermore, an adaptive weighted optimization framework (AWOF) is constructed to reduce the dimensions of searching space in the grouping strategy while adaptively determining the termination of weight optimization process. Specifically, a two-stage stochastic coding simulation method is proposed for the handling of dynamic constraints of the MLCIP. Extensive experiments on various benchmark problems and the MLCIP of actual inventory concentration data acquired from a copper industry were comprehensively used to validate the effectiveness of the designed AWOF.