Aiming at the problem of collaborative modeling between short-term mechanical responses and long-term climate dispatch strategies encountered in implementing power forecasting functions for intelligent operation and maintenance systems of hydropower units, A hybrid forecasting method integrating Temporal Convolutional Network (TCN) and Transformer is proposed in this paper. The methodology involves three key phases: initial correlation analysis of critical operational parameters such as water head, discharge, and guide vane opening with power generation, followed by construction of a TCN-Transformer hybrid architecture where TCN extracts short-term local features while Transformer captures long-term temporal dependencies, ultimately enhanced through multi-scale feature fusion optimization to refine prediction accuracy. Experimental results demonstrate that the proposed model achieves approximately 40 and 65% reductions in Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) respectively compared with standalone TCN and Transformer architectures, with Mean Absolute Percentage Error (MAPE) consistently below 0.4%. Moreover, it effectively characterizes nonlinear dynamics of hydropower units under complex operational conditions. This method enhances power prediction accuracy and demonstrates extended applicability in equipment trend analysis, condition assessment, and fault prewarning scenarios, delivering robust technical support for intelligent operation and maintenance of hydropower plants.