The energy crisis and environmental pollution have constrained the development of China’s automotive industry, making new energy-saving technologies a key focus. With the advancement of intelligent transportation and vehicle intelligence, the application of intelligent traffic information to achieve energy-efficient driving has attracted significant attention. This paper proposes an energy-saving technology for electric buses equipped with an instantaneous energy consumption prediction system. Based on the Hybridformer model, this system predicts energy consumption by integrating short-term and long-term prediction modules through dynamic sequence decomposition methods. The AO algorithm is employed to train the model, optimizing hyperparameters with adaptive mutation rates and non-elite elimination strategies. A predictive energy-optimized control (PEOC) controller is designed based on the prediction results to dynamically adjust speed and power, thereby reducing energy consumption while ensuring comfort and punctuality. Simulations on the Carsim-Simulink platform across three driving conditions demonstrate that the AO-optimized Hybridformer model achieves high prediction accuracy and that the PEOC controller significantly enhances the energy-saving performance of electric buses.