0% found this document useful (0 votes)
13 views5 pages

EE580 Fall 2021 Homework 5 Solutions

Uploaded by

Fiaz Hossain
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
13 views5 pages

EE580 Fall 2021 Homework 5 Solutions

Uploaded by

Fiaz Hossain
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

EE580, Fall 2021

Homework five
Due on December 1, 2023, 11:00PM. Upload your solution to Canvas.
Please write down name, student ID and course code on the first page of the solution.

Problem 1: (15 points) Consider the circuit in Figure 1, where Vs is the input and V1 , V2 are the states.
(P1) Develop the state-space equation for the system.
(P2) Show that the system is not controllable.
(P3) Explain why this system is not controllable in terms of the currents and voltages of the circuit.
(P4) Draw an equivalent circuit of the system that is controllable.

Fig. 1. Figure of Problem 1

Solution: (P1) (5 points) I = Vs −VR1 −V2 = C dV dV2


dt = C dt . So the state-space equation is given by
1

   1 1    1 
V̇1 (t) − RC − RC V1 (t)
= 1
1
1
1
+ RC 1
1
Vs (t)
V̇2 (t) − RC 2
− RC 2
V 2 (t) RC 2

1 1
 
(P2) (5 points) [B AB] = 1 −( RC 1
+ RC 2
) B and its rank is 1. So the system is not controllable.
(P3) (3 points) We can only control Vs , which in turn controls the amount of current flowing through the circuit.
Since this current is equal through both capacitors and current directly affects the voltage across a capacitor, there
is no way to individually control the voltages across the capacitors.
C1 C2
(P4) (2 points) We can control V3 in this circuit where C3 = .
C1 + C2

Fig. 2. Figure of Problem 7


Problem 2: (15 points) Is the following system controllable? Observable? Justify your results.
   
0 0 1
ẋ(t) = x(t) + u(t), y(t) = [0 e−t ]x(t).
0 −1 e−t
 
0 0
Solution: (8 points) Consider ẋ(t) = x(t). Since x1 (t) = x1 (t0 ) and x2 (t) = e−(t−t0 ) x2 (t0 ), the state
 0 −1

1 0
transition matrix is Φ(t, t0 ) = . We have
0 e−(t−t0 )
Z t1
e−t1 (t1 − t0 )
 
T T t1 − t0
Wc (t0 , t1 ) = Φ(t1 , τ )B(τ )B(τ ) Φ(t1 , τ ) dτ = .
t0 e−t1 (t1 − t0 ) e−2t1 (t1 − t0 )
The determinant of Wc (t0 , t1 ) is zero for all t0 and all t1 ≥ t0 . So the system is uncontrollable at any t.
(7 points) We use Theorem 6.O12 to verify observability. Let
N0 (t) = C(t) = [0 e−t ]
d
N1 (t) = N0 (t)A(t) + N0 (t)
 dt 
0 0
= [0 e−t ] + [0 − e−t ] = [0 − 2e−t ].
0 −1
 
N0 (t)
The determinant of is 0 for any finite t. Thus the system is unobservable at any t.
N1 (t)

Problem 3: (15 points) Reduce the system


   
−1 4 1
ẋ(t) = x(t) + u(t), y(t) = [1 1]x(t)
4 −1 1
into a controllable one. Is the reduced system
 observable?
  
1 3 −1 1 0
Solution: The controllability matrix C = has rank 1. So we choose P = and thus P =
  1 3 1 1
1 0
. Hence, we have
−1 1
   
−1 3 4 1
Ā = P AP = , B̄ = P B = , C̄ = CP −1 = [2 1].
0 −5 0
Thus, x̄ = P x transforms the system into
   
˙ 3 4 1
x̄(t) = x̄(t) + u(t), y(t) = [2 1]x̄(t),
0 −5 0
and the following subsystem is controllable:
x̄˙ 1 = 3x̄1 + u, y = 2x̄1 .
This subsystem is also observable since its observability matrix is 2 6= 0.

Problem 4: (15 points) Consider the process depicted schematically in Figure 3. A reaction tank of volume
V = 5, 000 gallons accepts a feed of reactant that contains a substance A in concentration CA,0 . The feed enters
at a rate of F gallons per hour and at a temperature T0 . In the tank, some of reactant A is returned into the
desired product B . The output product is removed from the tank at the same rate that the feed enters the tank.
The maximum in the tank has a uniform concentration of A, CA and a uniform temperature T . The temperature
of the water in the jacket is assumed uniform at TJ . The temperature of the water flowing into the jacket is TJ,0 .
The system is controlled by measuring the temperature T in the tank and controlling the flow of the water in the
jacket FJ by activating a valve.
The equation describing the evolution of CA , T and TJ are given by:
F F k2
ĊA = CA,0 − T − k1 CA e− T
V V
F F k2
Ṫ = T0 − T − k3 k1 CA e− T − k4 (T − TJ )
V V
FJ FJ
ṪJ = TJ,0 − TJ + k5 (T − TJ ),
VJ VJ
where CA,0 = 0.5, T0 = 70F , TJ,0 = 70F and k1 to k5 are appropriate constants.
A linearized state-space model ẋ = Ax + Bu and y = Cx around the equilibrium point C̄A = 0.245, T̄ = 140
and T̄J = 93.3 is given by:
−1.7 −2.13 ∗ 10−4
   
0 0
A =  696 2.9 2.4  , B =  0  , C = [0 1 0],
0 6.5 −19.5 −0.16
where x1 = δCA , x2 = δT , x3 = δTJ and u = δFJ . Is the system controllable? Is the system observable?

Fig. 3. Figure of Problem 4

Solution: Notice that the controllability matrix of the system is


 
0 0 0.0001
C = B AB A2 B = 
 
0 −0.3840 6.3744 
−0.1600 3.1200 −63.3360
and the observability matrix of the system is
   
C 0 1 0
O =  CA  =  696 2.9 2.4  .
CA 2 835.2 23.8618 −39.84
Both C and O have full rank 3. Thus, the system is both controllable and observable.

Problem 5: (20 points) Consider the LTI system ẋ(t) = Ax(t) + Bu(t). The reachable on [0, tf ] subspace R[0, tf ]
includes all the states xf for which there exists an input which transfers the state from x(0) = 0 to x(tf ) = xf . The
controllable on [0, tf ] subspace C[0, tf ] includes all the states x0 for which there exists an input which transfers
the state from x(0) = x0 to x(tf ) = 0.
The state-space model of the electrical network in Figure 4 is given by

− R11C1
   1 
0
ẋ(t) = x(t) + R11C1 u(t)
0 − R21C2 R2 C2
.
(P1) Consider the subsystem ẋ1 (t) = − R11C1 x1 (t) + R11C1 u(t). Is the subsystem controllable?
(P2) Consider the subsystem ẋ2 (t) = − R21C2 x2 (t) + R21C2 u(t). Is the subsystem controllable?
(P3) Let R1 = R2 = R and C1 = C2 = C . Is the above second-order system controllable.
(P4)  R2 = R and C1 = C2 = C . Show that, for any t1 > 0, R[0, t1 ] of the above system is given by
LetR1 =
1
{α , α ∈ R}.
1
 R2 = R and C1 = C2 = C . Show that, for any t1 > 0, C[0, t1 ] of the above system is given by
(P5) LetR1 =
1
{α , α ∈ R}.
1

Fig. 4. Figure of Problem 5

Solution: (P1) The controllability matrix is R11C1 6= 0 and has full rank. Thus, the subsystem is controllable.
(P2) The controllability matrix is R21C2 6= 0 and has
 full rank. Thus, the subsystem is controllable.
1 1
− 2 2
(P3) The controllability matrix is RC 1
R C
1 and has rank 1. The system is not controllable.
RC − R2 C 2
1
(P4) Let w = RC . The solution to the system is given by:
Z t Z t
x1 (t) = e−wt x1 (0) + we−w(t−τ ) u(τ )dτ, x2 (t) = e−wt x2 (0) + we−w(t−τ ) u(τ )dτ.
0 0
Since x(0) = 0, we have    
−wt 1 1
x(t) = e x(0) + α(t) = α(t)
1 1
R t −w(t−τ )
 α(t)
where  =w 0e u(τ )dτ . We cannot transfer the system from 0 to state x̄ with x̄1 6= x̄2 . Thus, R[0, t1 ] =
1
{α , α ∈ R}.
1
(P5) Now we want to transfer the system from x(0) to the origin. Then we need
 
−wt 1
0=e x(0) + α(t) .
1
   
1 1
This is only possible only when x(0) is aligned with . Thus, C[0, t1 ] = {α , α ∈ R}.
1 1

Problem 6: (20 points) Consider a LTV system Σ: ẋ(t) = A(t)x(t) + B(t)u(t). Let z(t) = P (t)x(t) where P (t)
is a continuously differentiable and nonsingular matrix and denote z -system by Σ0 . Show that Σ is controllable at
t0 if and only if Σ0 is controllable at t0 .
Solution: Note that
d
ż(t) = (P (t)x(t)) = Ṗ (t)x(t) + P (t)ẋ(t)
dt
= A0 (t)z(t) + B 0 (t)u(t),
where A0 (t) , [Ṗ (t) + P (t)A(t)]P (t)−1 and B 0 (t) , P (t)B(t).
Let XΣ (t) be a fundamental matrix of system Σ. Then XΣ (t0 ) is non-singular and ẊΣ (t) = A(t)XΣ (t). We
now proceed to show XΣ0 (t) = P (t)XΣ (t) is a fundamental matrix of system Σ0 . Since XΣ (t0 ) and P (t0 ) are
non-singular, so is XΣ0 (t0 ). In addition, we have
d
ẊΣ0 (t) = (P (t)XΣ (t)) = Ṗ (t)XΣ (t) + P (t)ẊΣ (t)
dt
= Ṗ (t)XΣ (t) + P (t)A(t)XΣ (t) = A0 (t)XΣ0 (t).
So, XΣ0 (t) = P (t)XΣ (t) is a fundamental matrix of system Σ0 .
Notice that
ΦΣ0 (t, t0 ) = XΣ0 (t)XΣ0 (t0 )−1 = (P (t)XΣ (t))(P (t0 )XΣ (t0 ))−1
= P (t)XΣ (t)XΣ (t0 )−1 P (t0 )−1 = P (t)ΦΣ (t, t0 )P (t0 )−1 .
Now let WΣ (t0 , t1 ) and WΣ0 (t0 , t1 ) be the controllability grammians for systems Σ and Σ0 . Then we have
Z t1
WΣ0 (t0 , t1 ) = ΦΣ0 (t1 , τ )B 0 (τ )B 0 (τ )T ΦΣ0 (t1 , τ )T dτ
t
Z 0t1
= (P (t1 )ΦΣ (t1 , τ )P (τ )−1 )P (τ )B(τ )B(τ )T P (τ )T (P (t1 )ΦΣ (t1 , τ )P (τ )−1 )T dτ
t
Z 0t1
= P (t1 )ΦΣ (t1 , τ )B(τ )B(τ )T ΦΣ (t1 , τ )T P (t1 )T dτ = P (t1 )WΣ (t0 , t1 )P (t1 )T .
t0

Since P (t1 ) is non-singular, WΣ0 (t0 , t1 ) is non-singular if and only if WΣ (t0 , t1 ) is non-singular. By the control-
lability theorem, system Σ is controllable at t0 if and only if system Σ0 is controllable at t0 .

Common questions

Powered by AI

For a second-order system with specified conditions such as R1 = R2 = R and C1 = C2 = C, the reachability subspace R[0,t1] is identified by analyzing the dynamic response of the system. Under these conditions, the system's states evolve such that both states move proportionally as α (a constant scalar), due to identical impedances and capacitances affecting both states equally. Thus, R[0,t1] can only span the direction along [1 1], indicating the system's limited ability to reach distinct states independently .

In determining the controllability and observability of a linearized state-space model around an equilibrium point, evaluate the ranks of the controllability and observability matrices formed by the model's A, B, and C matrices. Each matrix (controllability, observability) must have full rank equal to the number of state variables. This involves examining system behavior through equilibrium linearizations to see if all internal states can be influenced by inputs or measured by outputs. For instance, in the document example, the full rank of both controllability and observability matrices confirmed the system's properties .

To confirm a subsystem's controllability and observability, apply a transformation matrix P such that the controllability matrix of the transformed system attains full rank, and similarly for the observability matrix. For instance, in transforming  into , the transformed matrices ÿıA' = PAP^{-1}, B' = PB, and C' = CP^{-1} must ensure full rank for both controllability and observability matrices. The choice of P must strategically align with system dynamics to meet these conditions, as was done in the document example through specific matrix selections .

To transform a system into a controllable and observable one, apply a state transformation that alters the system matrices A, B, and C to new forms that fulfill the controllability and observability conditions. The transformation is achieved by selecting a matrix P such that the transformed system matrices (A', B', C') meet the desired properties. For instance, by choosing P^{-1} = [1,0;1,1], the example system's state matrices are transformed such that A' = PAP^{-1}, B' = PB, and C' = CP^{-1} become [3 4; 0 -5], [1; 0], and [2 1] respectively, making the subsystem controllable and observable .

To verify the observability of a system using Theorem 6.O12, assess the observability matrix by constructing it using the system matrices C and A and their higher derivatives if necessary. Calculate N0(t) = C(t) and N1(t) = N0(t)A(t) + dN0/dt(t). Then, the system is observable if the matrix composed of these N0, N1, ... matrices has full rank. In the document, N0(t) = [0 e^(-t)] and N1(t) = [0 -2e^(-t)] have a determinant of 0, thus indicating the system is not observable as their combined matrix does not have full rank .

Understanding that in an uncontrollable circuit system, manipulating the input voltage source Vs only affects the overall current passing through both capacitors without the ability to independently adjust the voltage across each capacitor is crucial. Since controlling current is the same for both, individual voltages cannot be independently controlled, making the system uncontrollable. This lack of individualized control over state variables through direct inputs defines the uncontrollability in terms of circuit dynamics .

To demonstrate that an electrical system is not controllable, you need to examine the controllability matrix, which is formed by the matrices B and AB, and their subsequent powers if required. Check the rank of this matrix; if it is less than the number of state variables, the system is not controllable. For instance, in the provided document, the matrix [B AB] = [1, −(1/RC1 + 1/RC2)] has a rank of 1, indicating the system is not controllable because the rank does not equal the number of states, which is 2 .

Designing a controllable equivalent circuit helps resolve non-controllability issues by restructuring circuit elements to enable independent control of desired states. This involves modifying the topology or parameters, allowing independent adjustment of state variables like voltage across capacitors. In the provided example, transforming the circuit to include additional components or altering configurations achieved control over previously uncontrollable voltages by rebalancing load distributions, providing separate control channels, thus creating an architectural intervention aligning with the desired controllability metrics .

The control mechanism in a reaction tank involving temperature ensures the detuning and observability of the reaction process through careful regulation of the jacket's water flow governed by temperature differentials. The use of state-space representation helps model the change in concentrations and temperatures, factoring in reaction kinetics and thermal dynamics. The documented system employs temperature feedback to adjust water flow, which in turn impacts reactant concentration stability and comprehension. Here, observability is supported by the clear direct measurement of state changes, ensuring robust closed-loop control behavior .

A linear time-varying (LTV) system Σ is simultaneously controllable with its transformed system Σ' if and only if the controllability grammian for one is invertible, implying that of the other is too. This condition is met since the transformation matrix P(t) connecting the two systems is non-singular across the relevant time interval. Therefore, the controllability of Σ can be translated to Σ' through this relationship, as both systems share a fundamental matrix relationship preserving this property .

You might also like