Note
Go to the end to download the full example code.
2.2-kW IM, discrete-time current control#
This example compares the continuous-time and direct discrete-time designs of the current controller in sensorless current-vector control of a 2.2-kW induction machine (IM). The sampling frequency is 5 kHz and the current-control bandwidth is 2π·700 rad/s, which is high compared with the sampling frequency.
Compute base values based on the nominal values.
nom = utils.NominalValues(U=400, I=5, f=50, P=2.2e3, tau=14.6)
base = utils.BaseValues.from_nominal(nom, n_p=2)
Configure the machine parameters.
par = model.InductionMachineInvGammaPars(
n_p=2, R_s=3.7, R_R=2.1, L_sgm=0.021, L_M=0.224
)
est_par = par # Assume accurate model parameter estimates
Simulate the drive in torque-control mode at the constant speed of 0.5 p.u. The discrete-time design includes the delays, so the control system does not advance the angle of its voltage reference (k_comp = 0).
def simulate(discrete: bool):
mdl = model.Drive(
model.InductionMachine(par),
model.ExternalRotorSpeed(),
model.VoltageSourceConverter(u_dc=540, modulator=model.CarrierComparison()),
)
mdl.mechanics.set_external_rotor_speed(lambda t: 0.5 * base.w_M)
cfg = control.CurrentVectorControllerCfg(
psi_s_nom=base.psi,
i_s_max=2 * base.i,
alpha_c=2 * pi * 700,
T_s=200e-6,
discrete=discrete,
)
vector_ctrl = control.CurrentVectorController(est_par, cfg)
vector_ctrl.observer.w_M = 0.5 * base.w_M # Initial speed estimate
ctrl = control.VectorControlSystem(vector_ctrl)
ctrl.set_torque_ref(
lambda t: nom.tau * (0.3 * (t > 0.3) + 0.2 * (t > 0.32) - 0.8 * (t > 0.34))
)
return model.Simulation(mdl, ctrl).simulate(t_stop=0.37)
res_cont, res_disc = simulate(discrete=False), simulate(discrete=True)
Plot the current responses in per-unit values.
set_style()
_, (ax1, ax2) = plt.subplots(2, 1, figsize=(8, 5), sharex=True)
for res, ls, label in [(res_cont, "--", "continuous"), (res_disc, "-", "discrete")]:
t, i_s = res.ctrl.t, res.ctrl.fbk.i_s / base.i
ax1.plot(t, i_s.real, ls, ds="steps-post", label=label)
ax2.plot(t, i_s.imag, ls, ds="steps-post", label=label)
t, i_s_ref = res_disc.ctrl.t, res_disc.ctrl.ref.i_s / base.i
ax1.plot(t, i_s_ref.real, "k:", ds="steps-post", label="reference")
ax2.plot(t, i_s_ref.imag, "k:", ds="steps-post", label="reference")
ax1.set_ylabel(r"$i_\mathrm{d}$ (p.u.)")
ax1.set_ylim(0.5, 0.7)
ax2.set_ylabel(r"$i_\mathrm{q}$ (p.u.)")
ax2.set_xlabel("Time (s)")
ax2.set_xlim(0.29, 0.37)
ax2.legend(loc="lower left")
plt.show()

At this bandwidth, the continuous-time design results in poorly damped oscillations, while the discrete-time design is well damped.
Total running time of the script: (0 minutes 3.071 seconds)