Abstract: To address the challenges of delayed quality monitoring and inefficient manual judgment in continuous casting production, an online quality monitoring and judgment system for casting slabs was developed based on multi-source data fusion. A three-tier architecture of "data acquisition, rule engine, and intelligent judgment" was designed, incorporating a real-time data acquisition module covering 707 process parameters across critical processes such as mold vibration and secondary cooling water distribution. A dynamically configurable three-tier metallurgical quality rule base was innovatively proposed, establishing a core rule engine for slab judgment to support real-time process parameter monitoring and anomaly warnings based on statistical process control(SPC), while achieving multi-level collaborative responses through sound-light alarms and WeChat notifications. To resolve defect traceability challenges, a three-dimensional coupling judgment model was established for mold level fluctuations, casting speed variations, and stopper rod movements, with formula-based characterization of the mapping relationship between process parameter fluctuations and quality defects. Application at a steel plant in Tangshan demonstrated significant improvements: through dynamic optimization of the metallurgical rule base and closed-loop feedback mechanisms for process parameters, the internal defect judgment accuracy increased from 25% to 85.6%, comprehensive judgment accuracy reached 98.5%, and manual re-inspection workload was reduced by over 75%. This research provides a full-process solution of "rule configuration, process monitoring, and quality traceability" for digital control of continuous casting, effectively addressing industry pain points such as outdated knowledge updates in traditional expert systems and insufficient interpretability of data-driven models.