The existing risk assessment methods for transmission tower slopes mainly focus on static geological characteristics and environmental factors,overlooking the coupling effect between the tower foundation and the slope. These methods also lack effective response and early warning mechanisms under extreme weather conditions,making it challenging to comprehensively evaluate slope stability.To address this issue,this study integrates slope risk and health factors-such as slope height,slope angle,distance between the tower foundation and the slope,and base conditions- and employs an enhanced Bayesian optimization algorithm to optimize a residual fully connected neural network.A Bayesian-optimized RFCN-based risk assessment model for transmission tower slopes was developed.Comparative experiments were conducted using BP neural networks,deep fully connected neural networks,and unoptimized RFCN as baseline models.The results demonstrated that the proposed model outperformed the others,achieving MAE of approximately 0.0102 and 0.0081,RMSE of0.0573 and 0.0551,and MAPE as low as 1.475%and1.451%for risk and health assessments,respectively.The model provides effective graded early warnings under routine inspections and rainfall,enhancing the accuracy and early warning capability of transmission tower slope risk assessment.