This session is chaired by:

Low-voltage motors are increasingly used in steering, cooling, pumping, and cabin-control subsystems of newenergy vehicles, and excessive temperature can impair their reliability. This paper develops a compact Modelica thermal-network model for an external-rotor low-voltage permanent-magnet synchronous motor. The model represents the principal heat sources, lumped heat capacities, and conductive, convective, and radiative heat-transfer paths. For the tested mass-produced motor, an end-to-end simulation of 30 min of physical operation completed in approximately 2s. Comparison with measured housing temperature for one operating case yielded a maximum relative discrepancy of 5.9%, occurring at the housing-temperature measurement point. These results indicate that the model is a computationally efficient reduced-order tool for the motor and operating range studied. Because the experimental comparison is limited to one motor type and one measured temperature location, accuracy and parameter transferability must be reassessed before the model is applied to other motor topologies or cooling arrangements.

With the continuous growth in global sales and market penetration, electric vehicles are increasingly required to operate under extreme temperatures, high altitudes, and humid or saline conditions, posing substantial challenges to thermal management. To address this challenge, this paper proposes a 1D model-based framework for integrated thermal management system (ITMS), enabling the unified design of modeling, simulation, and control. A vehicle-level thermal model is developed in Modelica to capture the coupled dynamics of the refrigeration cycle, cabin, battery, and motor, and is driven by driving cycles to simulate diverse operating conditions. The functional mock-up interface (FMI) is adopted to bridge the physical simulation model and the co-simulation environment, establishing a control interface that supports multivariable coordination and facilitates the integration of data-driven methodologies, such as reinforcement learning (RL). Within this framework, a RL-based control model is implemented to coordinate the system’s actuators. Under dynamic driving cycles, we evaluate the learned RL policy against a fixed-parameter baseline and a rule-based controller. The results validate the feasibility of the proposed integrated architecture in coupling physicsbased level modeling with adaptive data-driven control, providing a systematic approach for thermal management design in electric vehicles.