Aiming at the problems that traditional reliability analysis models are based on the assumption of independent factors, making it difficult to quantify multi-factor coupling effects and nonlinear superposition of fault probability, a mechanism-data fusion model for system fault analysis under multi-factor coupling is proposed. Based on failure physics, influential factors such as temperature and current density, as well as four core fault mechanisms including electromigration and corrosion, are defined. Coupling terms and coupling coefficients between fault mechanisms are constructed and quantified, and the total failure rate model is derived. System fault probability is calculated through numerical integration, while the analytic derivative method is used to quantify the sensitivity of fault probability to each factor, establishing a two-dimensional analysis method of probability importance and critical importance. Taking electrical components as an example, the analysis shows that the model can describe the fault evolution law under continuous changes of environmental factors, identify key factors for reliability control such as temperature and current density, and adapt to scenarios with low data volume and limited computing resources. The results are consistent with failure physics laws and physical constraints of coupling coefficients.