Digital Twin for Electric Bus Fleet Utilization

The patent implements dynamic prediction models that continuously adapt to changing environmental conditions (temperature, humidity, traffic patterns) rather than using static simplified estimates. The system updates battery performance predictions in real-time based on actual operating conditions, resolving the contradiction by making the system complex enough to capture dynamics but structured enough to remain manageable.

The patent introduces digital twin technology as an intermediary between physical battery systems and prediction algorithms. These virtual models serve as mediators that simulate battery behavior under various conditions, providing accurate predictions without requiring direct complex measurements from physical systems, thus maintaining prediction accuracy while managing system complexity.

The patent performs preliminary analysis of environmental factors by pre-processing and categorizing environmental data before main prediction computations. The system pre-identifies relevant environmental conditions (temperature ranges, traffic patterns) and prepares prediction models in advance, reducing the computational energy required during actual prediction operations while maintaining reliability through comprehensive environmental consideration.

The patent dynamically adjusts prediction model parameters based on environmental conditions, activating only the necessary computational models for current conditions. For example, thermal models are activated only when temperature variations are significant, reducing unnecessary computational energy consumption while maintaining prediction reliability when environmental factors are relevant.

The patent segments the fleet optimization problem into modular components: vehicle state monitoring, environmental condition assessment, prediction model selection, and assignment optimization. Each module handles a specific aspect independently, allowing real-time optimization without requiring a monolithic complex system. This modular approach improves operational efficiency while managing system complexity through divide-and-conquer strategy.

The patent implements feedback loops where actual operational data from fleet vehicles continuously updates prediction models and optimization algorithms. The system monitors real-time vehicle states, compares predicted vs. actual performance, and adjusts assignments dynamically. This feedback mechanism enables sustained operational efficiency improvements without permanently increasing system complexity, as the system learns and adapts over time.