This study addresses high fault uncertainty, time-varying dynamics and non-reversible reasoning in natural gas station regulators by integrating fuzzy fault tree analysis with Bayesian networks. The approach combines component-level reliability models to tackle complex structural uncertainties and dynamic failure scenarios more accurately than traditional methods like binary-state event relationships or T-S fuzzy gates. By leveraging explicit causal links through fuzzy logic while enabling probabilistic predictions using Bayesian inference over time-varying dependencies, the method provides robust probabilistic assessments. The subsequent calculation of posterior probability and sensitivity identifies weak links influencing the system’s reliability. Experimental validation demonstrates its capability to identify critical weak points affecting system performance, affirming applicability and potential impact on design optimization and maintenance strategies for real-world installations.