LUT-based Modelling of IPMSM for Electric Vehicle Applications

Accurately simulating Permanent Magnet Synchronous Motor (PMSM) drives requires motor models that capture magnetic nonlinearities — something classical linear dq-axis models with constant parameters simply cannot do. This thesis presents a complete workflow for bringing Finite Element Method (FEM) motor-characterization data into a Simulink-based Field-Oriented Control (FOC) environment. The methodology relies on three-dimensional FEM lookup tables (LUTs) defined over d-axis current, q-axis current, and rotor angle. To use these data with the Simscape Electrical FEM-Parameterized PMSM block, they require several processing steps. Every stage is implemented in MATLAB, forming a reproducible FEM-to-Simulink pipeline. The processed LUTs then feed the derivation of all control-oriented quantities needed to implement the drive model — local inductances, torque constant, effective flux-linkage parameter, and proportional-integral (PI) gains for the current and speed controllers. These quantities come together in a complete, closed-loop Simulink model comprising the LUT-based motor representation, inverter, measurement blocks, cascaded control loops, and test-signal generation. The final validation is focused on speed tracking. The reported results include a speed step test and three WLTP-based operating segments corresponding to the Medium-, High-, and Extra-High-speed phases. The WLTP profile was examined by separate simulations because a full continuous run exceeded the available computational and memory limits. The results show that the LUT-based model can reproduce stable closed-loop behavior over progressively more demanding reference-speed conditions, while highlighting the effect of control and operating constraints in the upper speed region. Overall, the work shows that FEM-based non-linear motor models can be integrated inside a structured control-oriented simulation framework, providing higher electromagnetic fidelity than simplified linear analytical models.