Abstract: In the foundry workshop of a home furnishing enterprise, sand core manufacturing and casting are two key operations with strong interdependence and precedence constraints: sand core manufacturing must be completed before casting, while casting depends on the progress of upstream sand core production. In actual production, poor coordination between these two production schedules often causes delays in one stage to propagate to the other, resulting in overall order tardiness. In addition, mold constraints in the production process further increase scheduling complexity. To address these issues, this paper constructs a joint scheduling optimization model with the objective of minimizing total order tardiness. An improved grey wolf optimization algorithm is proposed to solve the job scheduling problem in casting workshops, and a knowledge-based evolutionary search operator is introduced to enhance the global search capability and improve solution accuracy and efficiency. Multiple benchmark instances generated from practical orders and process structures are used to optimize the parameters through orthogonal experiments. Experimental results show that the proposed algorithm is superior to traditional scheduling algorithms in both solution efficiency and optimization quality. The improved algorithm significantly outperforms the benchmark algorithm in reducing order tardiness and performs well in practical production, effectively reducing operational uncertainty and improving the accuracy and flexibility of production scheduling.