Thermal Management System (TMS) in Electric Vehicles (EVs) is tasked with providing optimal thermal conditions for the EV components while keeping the passengers comfortable. An accurate TMS model prevents overengineered components during the early design phase, but high-fidelity models like CFD or FEM become computationally infeasible when simulating the whole system. Neural Networks (NNs) provide accuracy without heavy computational loads, however, their extrapolation capabilities can be limited when predicting coolant temperatures for EVs in the design phase. To solve this, the authors introduce an NN-based TMS simulation approach using analytical equations and dedicated look-up tables. The results show that the proposed approach outperforms the baseline approach only utilizing neural networks up to 11.5% during dynamic driving.
«
Thermal Management System (TMS) in Electric Vehicles (EVs) is tasked with providing optimal thermal conditions for the EV components while keeping the passengers comfortable. An accurate TMS model prevents overengineered components during the early design phase, but high-fidelity models like CFD or FEM become computationally infeasible when simulating the whole system. Neural Networks (NNs) provide accuracy without heavy computational loads, however, their extrapolation capabilities can be limit...
»