Professional Documents
Culture Documents
Dynamic Modeling
Presentation #1
2017
c
2 Linearization
3 Dynamics of PV link
When switching-mode power converters are used for PV power interfaces, the
state space averaging technique is required to derive mathematical models.
The concept was first introduced at Power Electronics Specialists Conference
(PESC) in 1976.
After then, the method had been widely utilized for various converter
topologies in order to derive mathematical models for control analysis.
The averaging approach is based on the condition that the switching
frequency is much higher than the system critical dynamics, which are formed
by energy storage components, such as inductor and capacitor.
Under the condition, the nonlinear switching dynamics can be neglected for
dynamic analysis and controller synthesis.
Based on one switching cycle, the system state space model can be derived
from the on state and off state of the power switch.
Through averaging, the averaged value of continuous signals can be used to
form a dynamic model without the representation of switching ripples.
If it falls into nonlinearity, the linearization process is required to find a
mathematical model, which can be utilized through the linear control theorem.
Photovoltaic Power System: Modelling, Design, and Control 3 / 21
Approximation based on Taylor Series expansion
where n! denotes the factorial of n and f (n) (x0 ) denotes the nth derivative
evaluated at the point x0 .
Based on the Taylor series, a small-signal models can be derived from the 1st
order approximation by neglecting high order terms, as expressed in
df (x)
f (x) ≈ f (x0 ) + (x − x0 )
dx x=x0
| {z }
C1
Applying the linearization process, the linearized dynamics of the two state
variables can be expressed as
d x̃1 ∂f (x1 , x2 , u) ∂f (x1 , x2 , u) ∂f (x1 , x2 , u)
= x̃1 + x̃2 + ũ
dt ∂x1
X2 ,U ∂x2
X1 ,U ∂u
X1 ,X2
| {z } | {z } | {z }
a11 a12 b1
(2a)
d x̃2 ∂g(x1 , x2 , u) ∂g(x1 , x2 , u) ∂g(x1 , x2 , u)
= x̃1 + x̃2 + ũ
dt ∂x1
X2 ,U ∂x2
X1 ,U ∂u
X1 ,X2
| {z } | {z } | {z }
a21 a22 b2
(2b)
0
8
6
GPV (s)
iPV (A)
4 −0.5
I−V GPV
2
MPP(37V,7.79A) at MPP(37V,−0.21S)
0 −1
0 10 20 30 40 0 10 20 30 40
300 0
RPV
RPV (Ω)
pPV (W)
200 at MPP(37V,−4.77Ω)
−50
100 P−V
MPP(37V,288.3W)
0 −100
0 10 20 30 40 0 10 20 30 40
vPV (V) vPV (V)
The absolute value of the dynamic resistance is used for dynamic modelling,
but changes with the operating point at the I-V curve.
The absolute value becomes higher when the operating point deviates from
the MPP and approaches to the left hand side, which can be called as the
current source region.
A circuit of buck converter can be used for the PVSC, which has been
discussed in Chapter 5.
The converter is controlled by the pulse width modulation (PWM), of which the
switching duty cycle is the control input variable.
When the PV link voltage is considered as the control variable, the output
voltage, Vo , is assumed to be constant for dynamic modeling.
Q on-state dynamics: Q off-state dynamics:
diL diL
L = vpv − V0 (5a) L = −Vo (6a)
dt dt
dvpv dvpv
Cin = ipv − iL (5b) Cin = ipv (6b)
dt dt
Averaging:
diL 1
= [dvpv − Vo ] (7a)
dt |L {z }
f (vpv ,d,iL )
dvpv 1
= [ipv − diL ] (7b)
dt Cin
| {z }
g(vpv ,d,iL )
D Vpv
d ĩL
0
dt
= L ĩL + L d̃
d ṽpv
D 1 ṽpv IL
− −
dt Cin RPV Cin Cin
where the symbols of D, VPV , and IL represent the switching duty cycle, the
PV terminal voltage, the inductor current, which are considered to be constant
in steady state.
The signals of ĩL and ṽpv are the state variables; d̃ represents the control
variable in the small signal model.
When the system enter steady state, a small perturbation can be applied
periodically to the duty cycle in order to evaluate the step response of vpv .
For comparison, the small signal model works in parallel with the simulation
model
The output of the small signal model does not show any switching ripples, but
captures the critical dynamics during the transient period.
37.5 vpv−small
since it matches very
37
well at the nominal
operating condition. 36.5
200 210 220 230 240 250 260 270 280 290 300
Time (ms)
When the PV link voltage is considered as the control variable, the output
voltage, Vo , is assumed to be constant for dynamic modeling.
Either pair of switches is on-state: All switches are off-state:
diL diL
L = Nvpv − V0 (9a) L = −Vo (10a)
dt dt
dvpv dvpv
Cin = ipv − NiL (9b) Cin = ipv (10b)
dt dt
Averaging is applied:
diL 1
= [2dNvpv − Vo ] (11a)
dt |L {z }
f (vpv ,d,iL )
dvpv 1
= [ipv − 2dNiL ] (11b)
dt Cin
| {z }
g(vpv ,d,iL )
2ND 2NVPV
d ĩL
0
dt
= L ĩL + L d̃
d ṽpv
2ND 1 ṽpv 2NIL
− −
dt Cin RPV Cin Cin
where the symbols of D, VPV , and IL represent the switching duty cycle, the
PV terminal voltage, the inductor current , which are considered to be
constant at steady state.
The signals of ĩL and ṽpv are the state variables that represent any small
variation; d̃ represents the control variable in the small signal model.
Even though the circuit is more complex than the buck topology, the dynamic
modeling approach can follow the same except for including the winding turn
ratio.
The dynamic model can be standardized the format:
K0 (βs + 1)
G0 (s) =
s2 + 2ξωn s + ωn2
where the undamped natural frequency and damping factor are expressed as
ωn and ξ, respectively.
This is a 2nd order system with two poles and one minimal-phase zero.
K0 is negative that represents the change of PV terminal voltage follows the
opposite direction of the duty cycle.
When the absolute value of Rpv is higher, the damping becomes lighter, which
causes more oscillation.
The damping becomes critical when the operating point enters the current
source region.
Photovoltaic Power System: Modelling, Design, and Control 20 / 21
Modelling verification for full bridge transformer isolated DC/DC converter as PVSC
When the system enter steady state, a small perturbation can be applied
periodically to the duty cycle in order to evaluate the step response of vpv .
For comparison, the small signal model works in parallel with the simulation
model
The output of the small signal model does not show any switching ripples, but
captures the critical dynamics during the transient period.
32.6
The negative gain is
37.5 vpv−small
since it matches very
37
well at the nominal
operating condition. 36.5
200 210 220 230 240 250 260 270 280 290 300
Time (ms)