Abstract: High-temperature alloy casting is critical in aerospace and energy fields, but it faces challenges such as complex processes, multi-physics coupling, and difficult defect control such as shrinkage porosity and hot cracks. Traditional methods rely on empirical trial-and-error, resulting in low efficiency and high costs, while existing digital technologies suffer from data heterogeneity and poor model generality. This study innovatively introduces domain ontology knowledge, constructing a metal manufacturing domain ontology model based on the IOF industrial ontology framework to standardize semantic expression of process knowledge. A digital model of the investment casting process is developed, integrating full-chain Internet-of-Things data for real-time simulation and optimization. Validated with a superalloy blade casting case, the model identifies root causes like pouring temperature fluctuations (±15°C) and slurry viscosity (>500 cP). After parameter adjustments, the product qualification rate increases from 40% to 66%, with slag inclusion and delamination defects reduced by 58% and 37%, respectively. This work offers an effective approach for intelligent transformation in high-temperature alloy casting.