A Data-Driven Framework for Off-Road Drivetrain Modeling and Optimization

Abstract

This paper investigates how recorded drive-cycle data can support the design and dimensioning of electric drivetrains for off-road applications. A measured wheel-loader duty cycle is analyzed to extract torque–speed distributions, transient operating patterns, efficiency behavior, loss profiles, and thermal behavior. The results show wide operating variations, including rapid direction changes, short high-torque intervals, and transitions between current-limited and voltage-limited regions, all critical for defining drivetrain requirements. Placing the measured operating points on the motor efficiency map and applying the resulting losses to a first-order thermal estimate shows that the examined machine is substantially overdimensioned for this duty cycle, demonstrating how the method exposes such mismatches. Building on this, a system-level modeling framework is presented, consisting of a permanent-magnet synchronous motor loss model with field-oriented control, an inverter loss model with modulation-dependent switching and conduction losses, and thermal models for the motor and inverter, as a basis for evaluating electrical and thermal constraints, loss segregation, and overload capability. This addresses a practical need existing approaches do not fully cover. Standardized cycles such as WLTP focus on light-duty road vehicles and do not represent reversal-heavy profiles of off-road machinery. Road-vehicle optimization frameworks typically target energy and cost without modeling motor and inverter thermal behavior under real loading and rely on standardized cycles rather than recorded off-road data even when addressing system-level trade-offs. The proposed methodology structures how recorded duty-cycle data should guide laboratory testing and model validation, laying the groundwork for a complete simulation environment. As the framework still requires calibration, its current output is best interpreted as indicative trends rather than final dimensioning guidance.