Model-Based Optimal Anti-jerk Control for Electric Vehicle Powertrain with Backlash

Abstract This paper presents a practical and observer-free backlash control strategy for electric vehicles. Unlike conventional feedback methods that rely on sensor-based state estimation and often trade off backlash duration to reduce jerk, the proposed approach achieves both goals simultaneously. By modeling the driveline and backlash dynamics, the optimal motor torque is analytically derived using Pontryagin’s maximum principle. The control operates in an open-loop manner using only in-vehicle sensors, and a lightweight algorithm detects the backlash entry point in real time without full-state estimation. The practical viability was further validated by demonstrating robustness to sensor noise and stability against drivetrain parameter uncertainties. Simulation and testbench experiments show that the proposed method reduces backlash duration by over 20% while maintaining the same jerk level as conventional controllers. The results confirm the method’s theoretical soundness and practical applicability to real EV powertrains.