# NATIONAL CHENG KUNG UNIVERSITY

Department of Mechanical Engineering

LINEAR SYSTEM HOMEWORK 4

Instructor: Student: Student ID: Department: Class:

Prof. Szu – Chi Tien Nguyen Van Thanh P96007019 Inst. of Manufacturing & Information Systems 1001- N154000 – Linear System

November 8, 2011

.10 Linear System Theory Page 1 .........................................................................................Contents Problem 1 .................................................................................................................................... 2 Problem 2 ...............

5. 1.Problem 1 Given a linear time-invariant system for the three-mass-spring system described in Example 3 of Section 5.3. Is the system controllable? Explain. Consider the controllability matrix 𝐶𝑀 = [𝐵 CM = 𝐴𝐵 𝐴2 𝐵 𝐴3 𝐵 𝐴4 𝐵 𝐴5 𝐵] Linear System Theory Page 2 . The measured output y(t)is the position of mass m2. Note that in this problem we have equal forces u(t) applied simultaneously to masses m1 and m3 but in the opposite directions.

5000 ⎦ ⎣ 0.0000 ⎦ 0.2041𝑖 ⎥ ⎢ 0.0000 ⎢ 0. 𝐵]) = 6. this mode is controllable. this mode is uncontrollable.7071 ⎥ ⎢ 0.000 ⎥ ⎢ 0.4082𝑖 ⎥ ⎥ ⎢ ⎥ ⎢ ⎥ 𝑉1 = . Called V1.2041𝑖 −0.5000 ⎤ ⎢ 0. V5. V3. 𝑉2 = . 2. 𝐵]) = 5. V2.Rank (CM) = 2 < 6.000 0.5000𝑖 −0.2041𝑖 ⎥ ⎢ −0.3536 ⎦ ⎣ −0.5774 ⎥ ⎣ 0.7071 ⎥ ⎢ ⎥ ⎢ 0. we calculate the eigenvalues of matrix A and then use PHB rank Test. 𝑉3 = ⎢ 0.5000𝑖 ⎡0.4082𝑖 ⎥ ⎢ 0. 𝐵]) = 5.0000 ⎦ PHB Test 𝑟𝑎𝑛𝑘([𝜆2 𝐼 − 𝐴 𝑟𝑎𝑛𝑘([𝜆4 𝐼 − 𝐴 𝑟𝑎𝑛𝑘([𝜆3 𝐼 − 𝐴 𝑟𝑎𝑛𝑘([𝜆1 𝐼 − 𝐴 𝐵]) = 5.5000𝑖 ⎥ ⎢ 0.0000 ⎤ ⎡ −0. V6 are corresponding to these six eigenvalues.5774 0.3536 ⎦ ⎣ 0. Linear System Theory Page 3 . hence the system is not controllable.0000 ⎤ ⎡ −0.000 ⎥ ⎥ ⎢ ⎥ ⎢ ⎥ .5774 ⎥ ⎣ −0. this mode is uncontrollable. Provide a physical meaning to each of the controllable modes (if any).2041𝑖 ⎡0.3536 ⎤ ⎡−0. 𝑉6 = ⎢ 𝑉4 = 0.000 0. V4.5000 ⎤ ⎡−0.3536 ⎤ ⎢ 0.5000𝑖 ⎥ ⎢ 0. this mode is uncontrollable.5774 0.5774 ⎥ ⎢− 0.5000 ⎦ ⎣ 0. Firstly. 𝑉5 = .5774 ⎥ ⎢ 0. 0.0000 ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢− 0. Eigenvalues of matrix A. So the controllability matrix is not full rank. In order to identify which modes are controllable or not. Identify the modes that are controllable and uncontrollable.

this mode is controllable. the displacement and the velocity of the mass 1 and the mass 3 are symmetric about the mass 2 and the mass 1 and the mass 3 are going to be far away from the mass 1. Two modes controllable. this mode is uncontrollable.𝑟𝑎𝑛𝑘([𝜆5 𝐼 − 𝐴 𝑟𝑎𝑛𝑘([𝜆6 𝐼 − 𝐴 𝐵]) = 6. Linear System Theory Page 4 . the displacement and the velocity of the mass 1 and the mass 3 are symmetric about the mass 2 and the mass 1 and the mass 3 are going to close to the mass 1. for example. we cannot find any input that brings system to these states. Can you find a control input u(t) that bring the system initial states to the following final states x(tf)? We pick up two independent column vectors of the controllability matrix. the eigenvectors are V4 and V5. For V4. hence dim(𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑎𝑙𝑏𝑒 𝑠𝑢𝑏𝑠𝑝𝑎𝑐𝑒) = 2. For V5. For the others. 𝐵]) = 5. we pick up the first tow column vectors of the controllability matrix The controllability matrix CM has rank 2. 3.

25*sin(101/5)^2 + 10176).(101*i)/10)*i)/(2*exp(conj(t)*i (101*i)/10)*(conj(t) . (or in the controllable subspace) hence we can find a control input u(t) that bring the system from initial state to this final states.1 A control input signal Linear System Theory Page 5 . that means 𝑥 (a) We can easily see that 𝑥 ⃗ �𝑡𝑓 � = 𝑃 ⃗�𝑡𝑓 � is in the range space of 𝑃1→2 .((i/(2*exp(conj(t)*i .25*sin(101/5)^2 + 10176) . Figure 1.(exp(conj(t)*i (101*i)/10)*i)/2)*(100*sin(101/5) + 2020))/(50*cos(101/5) 25*cos(101/5)^2 .(101*i)/10)*i)/(2*(conj(t) 101/10))))/(50*cos(101/5) . 1 0 ⎡1 0⎤ ⎢0 0⎥ ⎢ ⎥ = 0⎥ ⎢0 ⎢ 0 −1⎥ ⎣−1 0 ⎦ ut = .(101*i)/10)) .((100*cos(101/5) 100)*(((conj(t)*i .101/10)) + (exp(conj(t)*i (101*i)/10)*(conj(t)*i .25*cos(101/5)^2 . A control input is given below.𝑃1→2 �⃗1 . Assume that 𝑥 ⃗ (𝑡0 = 0) = 0.

3 Sate of mass 1 Linear System Theory Page 6 .Figure 1. Figure 1.2 Displacement of system from t0 to tf.

mass2 doesn’t move. Fig.5. we can see that.1. mass 1 and mass 3 are symmetric about mass 1.4 Sate of mass 2 Figure 1.5 Sate of mass 3 Through fours figures. Linear System Theory Page 7 .2 to Fig.1.Figure 1.

hence we cannot find a control input u(t) that bring the system from initial state to this final states. 6. (c) Check 𝑟𝑎𝑛𝑘([𝑃1→2 𝑥 ⃗ (𝑡𝑓 )]) = 3 > 2. No. hence the system is unobservable.(b) Check 𝑟𝑎𝑛𝑘([𝑃1→2 𝑥 ⃗ (𝑡𝑓 )]) = 3 > 2. No. because the system is not controllable. 7. 4. that means 𝑥 ⃗�𝑡𝑓 � is not in the controllable subspace. hence we cannot find a control input u(t) that bring the system from initial state to this final states. 5. Can you find a similarity transformation that places the system into the controllable form? If no. Linear System Theory Page 8 . Can you find a similarity transformation that places the system into the controller form? If no. explain why not. Rank (O) = 4 < 6. Is the system observable? Explain. explain why not. because the system is not controllable. that means 𝑥 ⃗�𝑡𝑓 � is not in the controllable subspace. Identify the modes that are observable and unobservable. Provide a physical meaning to the unobservable modes (if any).

𝐶 𝜆 𝐼 − 𝐴 �� = 6. this mode is observable.5000𝑖 ⎥ ⎣ 0. this mode is observable.000 ⎥ ⎢ ⎥ ⎥ . 𝐶 unobservable modes associated with eigenvectors 0. Linear System Theory Page 9 . 𝐶 𝜆 𝐼 − 𝐴 𝑟𝑎𝑛𝑘 �� 5 �� = 5 < 6.5000 ⎦ ⎣ 0. hence the sensor ysen(t) placed on mass m2 perceives no output.5000𝑖 −0. PHB Test 𝜆 𝐼 − 𝐴 𝑟𝑎𝑛𝑘 �� 1 �� = 6. 𝐶 𝜆 𝐼 − 𝐴 𝑟𝑎𝑛𝑘 �� 6 �� = 6. the masses m1 and m3 oscillate symmetrically about the mass m2 which stays stationary.000 ⎥ ⎢ 0.5000 ⎤ ⎡−0.5000𝑖 ⎥ ⎢ 0.000 ⎥ 𝑉4 = ⎢ ⎢ 0. 𝑉5 = ⎢ 0. this mode is unobservable. this mode is unobservable.5000 ⎤ ⎢ 0.5000 ⎦ Through the PHB rank test.5000𝑖 ⎡ −0.000 ⎥ ⎢− 0. this mode is observable. this mode is observable. 𝑟𝑎𝑛𝑘 �� 2 𝐶 𝜆 𝐼 − 𝐴 𝑟𝑎𝑛𝑘 �� 3 �� = 6. ysen(t)=0. 𝐶 𝜆 𝐼 − 𝐴 𝑟𝑎𝑛𝑘 �� 4 �� = 5 < 6.Use PHB rank test Eigenvalues of matrix A. we find that the modes λ4= i and λ5= -i are the In these modes.

because the system is unobservable. 𝐶 �? 1. No. explain why not. explain why not. Problem 2 Recall that all the units in Problem 1 are respectively m and m/sec for the mass positions and velocities. 𝐵 ̃ . u(t)in units of lbs. No. 𝑦 � = 𝑃𝑦. your high-level manager wants you to present the state-space model of the three-mass-spring system given in Problem 1 to the company executives in the following units for the states x(t). x2(t)in units of inch/sec. x4(t)in units of ft/min.8. 𝑢 � = 𝐿𝑢. What are your new state model matrices? 𝐴 Assume that. 9. 𝑥 � = 𝑇𝑥 . Can you find a similarity transformation that places the system into the observer form? If no. • y(t)in units of feet. input u(t) and output y(t): • • • x1(t)in units of inches. x6(t)in units of mile/hr. Can you find a similarity transformation that places the system into the observable form? If no. because the system is unobservable. The transformation: 𝑥̇ = 𝐴𝑥 + 𝐵𝑢 �̇̇ = 𝐴𝑇 −1 𝑥 � + 𝐵𝐿−1 𝑢 � � + 𝑇𝐵𝐿−1 𝑢 � 𝑇 −1 𝑥 𝑥 �̇̇ = 𝑇𝐴𝑇 −1 𝑥 → � −1 → � � −1 −1 −1 −1 𝑦 = 𝐶𝑥 + 𝐷𝑢 � = 𝐶𝑇 𝑥 � + 𝐷𝐿 𝑢 � � + 𝑃𝐷𝐿 𝑢 � 𝑃 𝑦 𝑦 � = 𝑃𝐶𝑇 𝑥 Page 10 Linear System Theory . x3(t)in units of feet. and the applied force u(t) in N. 𝐷 � . For some odd reasons. ̃ . x5(t)in units of miles.

0000 0.0000 0.0000 0.344⎦ 0.0254 inches/s.0000 0.0000 0.0167 0.0003 ⎥ 0.0000 0.2248 lbs.0000 0.0000 12.0000 3.0000 ⎦ Page 11 𝑢 � = 0.00000 0.0254 1 1 𝑥 . 1 m = 1/1609 miles.0000 0.0000 0.0000 0.0000 ⎥ ⎢ 0.0000 ⎥ 0.2808 ft.344 0.0000 0.0000 0.0000 0.0000 0.0000 ⎥ ⎢ 0. 1 ⎡ 0.00 0. 0. Hence.2808 ft.0000 0.00000 ⎣ 1609.0000 ⎤ ⎢0.2808 ∗ 60 1 0.2808 0.0254 ⎢ ⎥ → 𝑇 = ⎢ 0.0000 ⎢ 0.0000 0.0000 0.0000 1.344 2 𝑥 �2 = 1 m = 3.00000 0.0000 0.0000 0. 𝑥 �1 = 1 𝑥 . 1 m = 3.0000 0.0000 ⎤ ⎥ 0. 1 m/s = 3600/1609 miles/hr 𝑥 �4 = 3.2808𝑦 3600 𝑥 1609. 𝑃 = 3.2808𝑥3 𝑥 �6 = T is diagonal matrix.2808 0.2808 * 60 ft/min.0254 inches.000 0.0000 0.0000 0.000 0.0000 ⎢ 0.6818 0.00000 0. 1 N = 0.0000 0.2248.0000 3. 0.0000 0.0000 ⎥ 0. 𝑥 �3 = 3.0000 ⎥ 0.2808 ∗ 60𝑥4 . 𝑦 � = 3.0000 ⎡−1.0254 2 1 𝑥 �5 = 𝑥 .0000 0.0000 −120.0000 0.344 6 ̃ = 𝑇𝐴𝑇 −1 → 𝐴 Linear System Theory .0000 0.0000 ⎣ 0.̃ 𝑥 � 𝑢 𝑥 �̇̇ = 𝐴 � + 𝐵 � � + 𝑇𝐵𝐿−1 𝑢 � 𝑥 �̇̇ = 𝑇𝐴𝑇 −1 𝑥 → � � −1 −1 ̃ � � + 𝑃𝐷𝐿 𝑢 � 𝑦 � = 𝑃𝐶𝑇 𝑥 𝑦 � = 𝐶 𝑥 � + 𝐷𝑢 � ̃ = 𝑇𝐴𝑇 −1 𝐴 −1 � � 𝐵 = 𝑇𝐵𝐿−1 ̃ = 𝑃𝐶𝑇 𝐶 � = 𝑃𝐷𝐿−1 𝐷 Unit conversion: 1 m = 1/0.0000 316800 0.0000 ⎥ 0. 1 m/s = 1/0.2248𝑢. 1 m/min = 3. 1609.0000 0.9999 0.0000 0.0254 ⎥ 1 ⎢0.0000 0.00000 −3600 𝐿 = 0.0000 3600 ⎥ ⎢ 1609.0000 =⎢ ⎢ 4.0000 0.

No. Since rank of the new observability matrix has rank 4.0000 ⎥ ⎢ 0.8504 ⎤ ⎢ 0. explain.� = 𝑇𝐵𝐿−1 𝐵 ̃ = 𝑃𝐶𝑇 −1 = [0 𝐶 2. Yes.5029⎦ 0 1 0 0 0] Linear System Theory Page 12 . What are the system eigenvalues of your new state model? Are they the same as the original model? ̃: The eigenvalues of matrix 𝐴 They are the same to the eigenvalues of matrix A. Is any of the system controllability properties changed? If yes. explain. Is any of the system observability properties changed? If yes.0000 ⎥ ⎥ =⎢ ⎢ 0.0000 ⎡ 8. 3. � = 𝑃𝐷𝐿−1 = 0 𝐷 0. Since rank of the new controllability matrix has rank 3 more than the original controllability matrix.0000 ⎥ ⎣−0. 4.

Sign up to vote on this title