GA-optimized fractional-order PID control with data-driven thermal forecasts for PMSM drives in electric vehicles

Electric vehicle (EV) traction drives must deliver fast and precise speed control while keeping permanent-magnet synchronous motors (PMSMs) within safe thermal limits. Conventional PID and even fixed fractional-order PID (FO-PID) controllers typically ignore the coupled electrical–thermal dynamics and rely on conservative derating rules, which can degrade efficiency and performance. This paper proposes a thermal-aware GA-optimized FO-PID speed controller in which short-horizon data-driven temperature forecasts are explicitly integrated into the control loop. Sequence models based on Long Short-Term Memory (LSTM) and Transformer architectures are trained on the public Paderborn PMSM dataset to jointly predict rotor and stator temperatures over a 5 s horizon under realistic load profiles. The multi-step forecasts are compressed into a scalar thermal-risk index that schedules both the gains and fractional orders of the FO-PID controller. A Genetic Algorithm (GA) is then used to jointly optimize the nominal FO-PID parameters and the scheduling gains under a composite objective that penalizes tracking error, control effort and thermal-limit violations. Simulation studies aligned with EV drive-cycle statistics compare the proposed controller against conventional PID and static FO-PID baselines using integral error indices, overshoot, settling time, and thermal margin. The results show that the thermal-aware GA-optimized FO-PID improves speed-tracking performance. while reducing time above the specified temperature limit to near-zero and preserving a comfortable thermal margin across the tested scenarios Execution-time measurements further confirm that temperature prediction, risk computation and FO-PID updates fit within a 1 kHz control cycle on a standard CPU, without requiring GPUs or specialized hardware. These findings indicate that embedding short-horizon thermal forecasts into FO-PID control is a practical and effective strategy for enhancing both dynamic performance and thermal safety in PMSM drives for EV applications.