Page 261 - Modern Control of DC-Based Power Systems

Simulation 225

Figure 6.8 Current—PID CPL—Step 10.3 - 17.8MW.

It is perceived that the voltage overshoot is kept small by all the con-
trol strategies, and the settling time, defined as the time until the step
response stays within a range of 2.5% of the final, is rather fast.
The transients of current and voltage for the virtual disturbance
approaches display nearly overlapping transients, the differences in mag-
nitude of voltage and current are negligible in the averaged model. The
time evolution of both virtual disturbance control strategies is identical,
this resides in the fact that both control strategies use a steady-state
Kalman Filter. Consequently, the input in the control system is a
brutally-linearized software measurement. The Kalman Filter reflects
itself in a remaining steady-state error which is caused by an estimation
deviation of 0.5%.
In contrast to the virtual disturbance-based controls the Adaptive
Backstepping is able to achieve a zero steady-state error under all CPL
implementations; furthermore, one can observe a different behavior,
which lies on the fact that the Adaptive Backstepping uses a nonlinear
model for estimation combined with a nonlinear control law, while the
Backstepping with virtual disturbance estimation uses a linear plant
model. This strengthens the assumption that the difference in control per-
formance may originate from the disturbance modeling and its
estimation.
The comparison between different CPL implementations (ideal, PI,
PID) reveals the interdependency between control bandwidths and their
interactions. Selecting a PID control with a more realistic model of the
CPL causes higher disturbances compared to the ideal CPL. On the other