160 kW induction-machine FOC F1 simulation

Purpose

This test adds a 160 kW induction machine to the hybrid field-oriented control (FOC) simulations. It validates bipolar speed tracking, opposing load-torque steps, constant-flux operation, the transition to field weakening, voltage and current limits, and optional load-torque feedforward.

The implementation consists of:

• src/simulators/hybrid_foc_speed_f1_simulator.jl: the general hybrid F1 speed-control simulator and shared profile support;
• src/simulators/hybrid_foc_speed_f1_160kw_simulator.jl: the 160 kW machine parameters, bipolar test profiles, and specialized simulation entry point;
• src/controls/FOC/outer_speed_flux_f1_discrete.jl: the outer speed loop, constant-flux F1 reference, and field-weakening law;
• scripts/run_foc_speed_f1_160kw.jl: the executable test, metrics, plots, and figure export.

The specialized simulator and its two profile functions are included and exported by src/InductionMachineDrives.jl:

simulate_foc_speed_f1_im_160kw
im_160kw_speed_reference
im_160kw_load_torque_profile

Simulation structure

The simulation combines discrete control and estimation with a continuous machine plant:

speed reference
    -> F1 outer speed/flux controller
    -> dq current references
    -> discrete dq current controller
    -> inverse Park transformation
    -> stationary alpha-beta induction-machine plant (RK4)
    -> rotor-flux and torque observer
    -> measured speed and observed dq quantities

An optional Kalman estimator supplies a load-torque estimate to the outer-loop feedforward path. The available modes are:

• :none: no load estimate;
• :actual: exact simulated load torque, useful for initial validation;
• :kalman: estimated load torque.

The plant, observer, controllers, and load estimator have independent parameter scale factors. This allows robustness tests such as changing the actual rotor resistance without changing the resistance assumed by the observer.

160 kW machine parameters

The default electrical parameters are derived from the 50 Hz equivalent-circuit data used by the original initialization model. Reactances are converted to inductances using

\[L = \frac{X}{2\pi f_n}.\]

Important defaults include:

The simulator returns both the parameter objects used by each subsystem and a nominal tuple containing the derived inductances, leakage/transient inductance, coupling factor, and rotor time constant.

Constant flux and field weakening

F1 operation uses a constant nominal d-axis current below the field-weakening base speed:

\[i_{sd}^{*} = i_{sd,nom}.\]

When field weakening is enabled and the magnitude of the filtered or commanded speed exceeds wm_base_fw, the desired d-axis current is reduced according to

\[i_{sd}^{*} = \operatorname{clamp}\left( i_{sd,nom}\frac{\omega_{base}}{|\omega|}, i_{sd,min}, i_{sd,nom} \right).\]

The d-axis reference is rate-limited by id_dot_max to avoid an abrupt flux change. The controller exposes field_weakening_active so the transition can be checked directly.

The default base speed is

wm_base_fw = 992.0 * 2pi / 60.0  # approximately 103.9 rad/s

The standard test commands ±50 rad/s. It therefore validates constant-flux F1 operation while leaving field weakening enabled and ready for higher speeds. To exercise field weakening explicitly, use a speed above the base speed, for example:

res_fw = simulate_foc_speed_f1_im_160kw(
    wm_ref_abs = 120.0,
    use_field_weakening = true,
)

maximum(res_fw.field_weakening_active) == 1.0

Speed-reference profile

The default 13 s profile tests both rotation directions:

The profile magnitude and all transition times are keyword arguments of simulate_foc_speed_f1_im_160kw.

Load-torque profile and sign convention

The mechanical equation is

\[J\frac{d\omega_m}{dt} = T_e + T_{load} - B\omega_m.\]

Consequently, the load profile uses negative torque during positive rotation and positive torque during negative rotation. The default steps are:

Thus, the disturbance opposes the commanded motion in both halves of the test.

Running the test

From the project root, start Julia:

jl

At the Julia prompt, run:

include("scripts/run_foc_speed_f1_160kw.jl")

The script uses the exact simulated load torque for feedforward by default:

load_estimator = :actual
use_load_feedforward = true

Change load_estimator to :none to test feedback alone or to :kalman to test the estimator-based feedforward path.

Results and plots

The returned named tuple contains speed, torque, dq current and voltage, observer, flux, three-phase, power, saturation, and field-weakening signals. The run script reports:

• speed RMSE and peak absolute speed error;
• maximum reference and measured current magnitude;
• maximum applied and unsaturated voltage magnitude;
• current-, torque-, and q-current-saturation fractions;
• field-weakening active fraction;
• nominal rotor time constant and transient inductance.

MakieControlPlots generates three figures:

1. speed and torque;
2. dq currents, voltage utilization, and rotor flux;
3. current limits, power, and saturation flags.

They are saved under:

results/im_160kw_f1/

with the filenames speed_and_torque.png, currents_voltage_flux.png, and current_power_saturation.png.

Suggested validation sequence

1. Run the default ±50 rad/s case and confirm that field_weakening_active remains zero.
2. Confirm that the measured speed follows both positive and negative references and that the applied load always opposes motion.
3. Compare :actual, :none, and :kalman load-estimator modes.
4. Raise wm_ref_abs above wm_base_fw and confirm that the d-axis current and flux references decrease while field_weakening_active is one.
5. Inspect vs_mod, vs_mod_unsat, and ctrl_p.Vs_max to assess voltage utilization in the field-weakening region.
6. Apply plant/observer scale-factor mismatches to evaluate robustness.