Objective To address the issues of mutual exclusivity among primal cuts, overlapping cutting regions, and the limitations of existing methods in balancing technological rules, market characteristics and multi-objective production decision-making in pig carcass cutting, this paper develops a multi-objective optimization method for pig carcass cutting incorporating multi-attribute constraints.

Method The selection of primal cuts and the allocation ratio of overlapping regions are defined as decision variables. Three optimization objectives are set: maximizing total revenue, minimizing preservation loss, and minimizing risk of unsold products. Corresponding constraints are constructed, including mutual exclusivity of primal cuts, overlap allocation, complete carcass cutting, sales probability, and shelf life. An improved NSGA-III is proposed with hybrid coding, Logistic chaotic initialization, adaptive hybrid crossover and mutation, and a constrained dominance rule. Furthermore, the Stochastic Multi-Attribute Acceptability Analysis (SMAA) is employed to comprehensively rank the Pareto solutions. A case study is conducted using the production and sales data from a pig slaughtering and processing plant of Wens Food Group Co., Ltd.

Result Experimental results demonstrate that the improved NSGA-III significantly outperforms both the standard NSGA-III and MOEA/D-DE in terms of the HV and IGD+ metrics, exhibiting superior Pareto front approximation capability, objective space coverage, and solution stability. SMAA can identify recommended solutions from feasible Pareto solutions that achieve a balance among revenue, preservation loss, and unsalable risk. Sensitivity analysis shows that the recommended solution maintains good stability under common variations in carcass quality, and can be adjusted accordingly in response to market price changes of key profitable primal cuts.

Conclusion The proposed method enables synergistic optimization of revenue, preservation loss, and risk of unsold products in pig carcass cutting, and accomplishes solution ranking and recommendation under weight uncertainty. It can provide methodological support for refined cutting and production decision-making in pig slaughtering and processing enterprises.