Abstract: In order to ensure the safety of high temperature molten metal operation in metallurgy industry， the concept of safety integrity of safety-related systems in IEC 61508 was extended to high temperature molten metal operation system， and a safety integrity evaluation method for high temperature molten metal operation based on Bayesian networks Noisy-or model was proposed. A safety integrity evaluation model for high temperature molten metal operations was established， which includes human factors， material factors， environmental factors and management factors. The prior probability of each root node of the Bayesian network was determined by the fuzzy number of fuzzy theory. In order to reduce the amount of calculation and uncertainty， the Noisy-or model was used to determine the conditional probability， and the synergistic effect of each intermediate node on the safety and integrity of the operation was considered. The results show that the evaluation model causal reasoning can determine the safety integrity level of high temperature molten metal operations， and diagnostic reasoning can also determine its sensitive factors， providing guidance for the formulation of safety measures.

Key words: high temperature molten metal operation; Bayesian network; fuzzy theory; Noisy-or model; safety integrity