Flow Lab

Syllabus / Module 13: Applications: Control Systems

Lesson 13.6

State-Space Control & Pole Placement

State-space models, controllability, full-state feedback u = −Kx, pole placement, and observability.

Written byClaude Sonnet

🌊
Why this matters for simulation: Modern flight control is designed in state space.

By the end of this lesson you can…

  • Write the cruise plant, a double integrator and a short-period model in the form $\dot{\mathbf x} = A\mathbf x + B u$, $y = C\mathbf x$, and say what each matrix means.
  • Show that full-state feedback $u = -K\mathbf x + Nr$ gives the closed-loop matrix $A - BK$, so that the closed-loop poles are the eigenvalues of $A - BK$.
  • Test controllability with the matrix $[B, AB, \dots]$ and explain, with a left eigenvector, why an uncontrollable mode cannot be moved.
  • Place the poles of a two-state system by matching characteristic polynomials, and use Ackermann's formula for $n = 2$ and $n = 3$.
  • Get zero steady-state error with a feedforward gain $N$ and with integral augmentation, and say which one survives a modelling error.
  • Explain what pole placement leaves out: control effort that grows like $\omega_n^2$, sensor noise gain, unmodelled dynamics and the zeros that feedback cannot move.
On this page
·
  1. State-space form
  2. Full-state feedback moves the eigenvalues
  3. Controllability: when placement is possible
  4. Pole placement by Ackermann’s formula
  5. Tracking a reference: feedforward and integral action
  6. What pole placement does not tell you
  7. A Module 12 example: placing the Dutch roll
  8. Placement against classical design
  9. Concept checks
  10. Playground and exercise
  11. Problems
  12. What comes next
  13. Key Takeaways

13.5 Frequency Response & Bode Plots described a loop by how it treats a sinusoid, for one input, one output and a controller whose shape was fixed in advance. It never said how to choose the dynamics, and the yaw damper of 12.6 Lateral-Directional Modes fed back exactly one state of an aircraft that has many. Here we feed back all of them, u=−Kxu = -K\mathbf x, and ask what that buys: the eigenvalues of 7.2 Linear Systems & the Eigenvalue Method become design knobs. The price, as always, is paid in control effort, and the lab prints it.

▶ Video The cruise plant as a two-state model, feedback on both states moving the eigenvalues, the controllability matrix, a feedforward gain and an integrator for tracking, and the effort that fast poles costcaptions only
Transcript
  1. Every linear system in the course fits one template: x dot equals A x plus B u, and y equals C x. The state is the smallest set of numbers that fixes the future. The cruise plant is two states, speed y and engine a, with eigenvalues minus 0.1 and minus 1. The double integrator, a position and a velocity driven by an acceleration, has both poles at the origin. The short period is the w q block of its matrix, eigenvalues minus 2.6649 plus or minus 3.6921 i. The lesson prints no elevator column for it, so the lab uses a stand-in, B equal 0 and minus 4: illustrative, not the aircraft's own.
  2. Feed back every state: u equals minus K x plus N r. Substitute, and the closed loop is x dot equals A minus B K times x plus B N r. Its eigenvalues are the poles, so placing poles means choosing K. For the cruise plant the characteristic polynomial is s squared plus 1.1 plus k 2, times s, plus 0.1 times 1 plus k 1 plus k 2. Match it to a damped pair at zeta 0.7: omega n equal 1 gives K equal 8.7 and 0.3, poles minus 0.7 plus or minus 0.7141 i. Omega n 0.5 gives k 2 equal minus 0.4: slower than the plant, so placement removes damping if you ask.
  3. Can every pole be placed? Only if the controllability matrix, B, A B and so on, has rank n. The cruise plant gives determinant minus 0.1 and the double integrator minus 1: both controllable. If a left eigenvector w satisfies w B equal 0, then w times A minus B K equals lambda w for every K: that pole is stuck. The lab's diagonal minus 1, minus 2 with B equal 1, 0 has rank 1 of 2, and the pole at minus 2 cannot move. Otherwise Ackermann's formula, from memory, gives K in one line: on the cruise plant the last row of the inverse is 10, 0, and K is 8.7, 0.3 again.
  4. Placement is exact, and the design can still be bad. After a unit step in r the first command is N r, which for the cruise plant is 10 omega n squared. The peak commands at zeta 0.7 are 3.0642 at omega n 0.5, then 10, 40 and 160 at 1, 2 and 4, while the settling time falls from 5.985 seconds at 1 to 3.000 and 1.500. Doubling the speed quadruples the command and only halves the settling time. The double integrator is the same: 4 at 2, 16 at 4. And the sensor noise gain on y rises from 8.7 to 154.5, a factor of 17.8.
  5. Now tracking. The feedforward gain N is minus 1 over C times A minus B K inverse times B, and for the cruise plant it is 10 omega n squared. At omega n 1 the step settles at 1 with 4.60 percent overshoot in 5.985 seconds. But N comes from the model. Give the car 20 percent less authority and the output ends at 0.97561. At omega n 0.5 the miss is 0.90909, and at omega n 4 only 0.99844: more loop gain, smaller miss, but never zero.
  6. Integral augmentation adds a state z with z dot equal r minus y and places one more pole. On the cruise plant with a third pole at minus 3: K equal 47.7 and 3.3, k I equal minus 30. Overshoot 4.27 percent, settling 6.33 seconds. With the same 20 percent error the output still ends at exactly 1.0000, because the command ends at 1.25 to make up the lost authority. And this is PID: proportional 51, derivative 33, integral 30, with characteristic polynomial 10 s cubed plus 44 s squared plus 52 s plus 30: the three poles we asked for. That is 13.3 again.
  7. Saturation changes the loop. At omega n 2 the feedforward design asks for 40, a limit of 10 gives it 10, and it still settles in 3.54 seconds with 3.32 percent overshoot: there is nothing to wind up. The integral design, K equal 118.3 and 4.7, k I minus 120, has 3.12 percent overshoot and settles in 3.285 seconds unlimited. With the limit of 10 it is 16.75 percent and 3.975 seconds: windup, with no anti-windup in the lab. And feedback does not move zeros. The short period stand-in overshoots 58.6 percent at omega n 6, and 128.0 at omega n 10.
  8. A Module 12 example: the lateral model of 12.6, four states, rudder input, controllability rank 4. The yaw damper moved the Dutch roll to minus 1.6647 plus or minus 1.5465 i but dragged the spiral to minus 0.1731. Ask instead for a Dutch roll at omega n 2.5, zeta 0.7 and keep roll and spiral: Ackermann gives a yaw-rate gain of 0.49943, close to 12.6's 0.5, and the eigenvalues come out minus 9.3790, minus 1.7500 plus or minus 1.7854 i, and minus 0.0068. But the spiral stays slow: only 0.0950 of the reference in 15 seconds. Move roll and spiral too and the rudder command is 12.15 radians per unit of yaw rate: about 35 degrees at a reference of 0.05, with 572 percent overshoot.
  9. Pole placement answers where the poles should be, and leaves the harder question open. Omega n equal 1 was a guess, and the obvious lateral placement asked for 12.15 radians of rudder. Lesson 13.7, on LQR and state estimation, is not yet written. It is meant to choose K by optimisation, paying for state error and control effort, and to add an observer, a model run alongside the plant that estimates the states you cannot measure.

State-space form

Every linear time-invariant system the course has met fits one template:

x˙=Ax+Bu,y=Cx.\dot{\mathbf x} = A\mathbf x + B u, \qquad y = C\mathbf x .

The state x\mathbf x is the smallest set of numbers that, together with the future input, fixes the future. AA is the matrix of 7.1 Systems & State Space, BB says where the input pushes, and CC says what the sensor sees. The transfer function is G(s)=C(sI−A)−1BG(s) = C(sI - A)^{-1}B, and its poles are eigenvalues of AA.

The cruise plant, two states. The car of 13.1 Feedback & Block Diagrams with the engine lag of 13.2 has 10y˙=a−y10\dot y = a - y and a˙=u−a\dot a = u - a. Take x=(y,a)\mathbf x = (y, a):

A=[−0.10.10−1],B=[01],C=[10].A = \begin{bmatrix} -0.1 & 0.1 \\ 0 & -1 \end{bmatrix}, \quad B = \begin{bmatrix} 0 \\ 1 \end{bmatrix}, \quad C = \begin{bmatrix} 1 & 0 \end{bmatrix}.

The eigenvalues are −0.1-0.1 and −1-1, the two poles of 1/((10s+1)(s+1))1/((10s+1)(s+1)). A PID controller of 13.3 sees only yy (and its derivative). State feedback also sees the engine state aa.

Two more plants appear in the lab. The double integrator (A=[0100]A = \begin{bmatrix}0&1\\0&0\end{bmatrix}, B=(0,1)B = (0,1), a position and a velocity driven by an acceleration) is the cleanest test plant: both open-loop poles sit at the origin. The short period of 12.5 Trim, Linearization & Longitudinal Modes is the (w,q)(w, q) block of its matrix, rows and columns two and three:

A=[−2.4666449.9461−0.27371−2.86316],A = \begin{bmatrix} -2.46664 & 49.9461 \\ -0.27371 & -2.86316 \end{bmatrix},

with eigenvalues −2.6649±3.6921i-2.6649 \pm 3.6921i (the lab’s readout; the 4×44\times4 matrix has −2.6647±3.6964i-2.6647 \pm 3.6964i because the speed and pitch states couple in slightly). That lesson prints no elevator column for this block, so the lab uses a stand-in, B=(0,−4)B = (0, -4) per radian, of plausible sign and size: the elevator numbers below are illustrative, not the 12.4 aircraft’s.

Full-state feedback moves the eigenvalues

Choose u=−Kx+Nru = -K\mathbf x + Nr, where KK is a row of gains and rr the reference. Substitute:

x˙=(A−BK)x+BN r.\dot{\mathbf x} = (A - BK)\mathbf x + BN\,r .

The input has disappeared and a new matrix has taken the place of AA. Its eigenvalues are the closed-loop poles, so placing the poles means choosing KK so that A−BKA - BK has the eigenvalues we want.

Worked example 1: the cruise plant by matching coefficients. With K=(k1,k2)K = (k_1, k_2),

A−BK=[−0.10.1−k1−1−k2],det⁡(sI−A+BK)=s2+(1.1+k2)s+0.1(1+k1+k2).A - BK = \begin{bmatrix} -0.1 & 0.1 \\ -k_1 & -1-k_2 \end{bmatrix}, \qquad \det(sI - A + BK) = s^2 + (1.1 + k_2)s + 0.1(1 + k_1 + k_2).

We want the polynomial of a damped pair, s2+2ζωns+ωn2s^2 + 2\zeta\omega_n s + \omega_n^2. Matching, k2=2ζωn−1.1k_2 = 2\zeta\omega_n - 1.1 and k1=10ωn2−1−k2k_1 = 10\omega_n^2 - 1 - k_2. For ωn=1\omega_n = 1, ζ=0.7\zeta = 0.7: k2=0.3k_2 = 0.3 and k1=8.7k_1 = 8.7. The lab reports K=[8.7000, 0.3000]K = [8.7000,\ 0.3000] with closed-loop eigenvalues −0.7000±0.7141i-0.7000 \pm 0.7141i.

Two gains, two poles anywhere; proportional control on yy alone (13.2) had one gain and a locus. With ωn=0.5\omega_n = 0.5 the formula gives k2=−0.4k_2 = -0.4: a negative gain, because the target poles are slower than the plant’s own. Placement will happily remove damping if you ask.

◆ Interactive Cruise plant with full-state feedback: choose wn and zeta, read K, the closed-loop eigenvalues and the control effort; the poles of A - BK sit exactly where you put them

Loading simulation…

Controllability: when placement is possible

For one input, the controllability matrix is Wc=[ B, AB, …, An−1B ]W_c = [\,B,\ AB,\ \dots,\ A^{n-1}B\,]. The input pushes the state along BB; the dynamics rotate that push to ABAB, then A2BA^2B, and so on. If these nn directions span the whole state space, the input can reach every state. The rule (a theorem, stated here with the necessity argument only): every closed-loop pole can be placed anywhere (conjugate pairs together) if and only if WcW_c has rank nn.

For the cruise plant Wc=[00.11−1]W_c = \begin{bmatrix} 0 & 0.1 \\ 1 & -1\end{bmatrix} has determinant −0.1-0.1: controllable. For the double integrator Wc=[0110]W_c = \begin{bmatrix}0&1\\1&0\end{bmatrix} has determinant −1-1.

Why failure means a stuck pole. Suppose a row vector w\mathbf w satisfies wA=λw\mathbf wA = \lambda\mathbf w and wB=0\mathbf wB = 0: a left eigenvector the input is blind to. Then for any KK,

w(A−BK)=λw−(wB)K=λw,\mathbf w(A - BK) = \lambda\mathbf w - (\mathbf wB)K = \lambda\mathbf w ,

so λ\lambda is still an eigenvalue of A−BKA - BK. The lab’s last plant shows it: A=diag⁡(−1,−2)A = \operatorname{diag}(-1, -2) and B=(1,0)B = (1, 0). The controllability matrix has rank 11 of 22, the lab refuses to place poles, and the pole at −2-2 cannot move.

Pole placement by Ackermann’s formula

Matching coefficients is fine for n=2n = 2 and gets painful beyond that. Ackermann’s formula (from memory, not derived here) does it for one input in one line. If ϕ(s)=sn+c1sn−1+⋯+cn\phi(s) = s^n + c_1s^{n-1} + \dots + c_n is the desired characteristic polynomial,

K=[0⋯01]Wc−1 ϕ(A),ϕ(A)=An+c1An−1+⋯+cnI.K = \begin{bmatrix} 0 & \cdots & 0 & 1 \end{bmatrix} W_c^{-1}\,\phi(A), \qquad \phi(A) = A^n + c_1A^{n-1} + \dots + c_nI .

Wc−1W_c^{-1} exists exactly when the system is controllable. Here ϕ(A)\phi(A) is the desired polynomial evaluated at the plant matrix.

Worked example 2: Ackermann on the cruise plant. Wc−1=[101100]W_c^{-1} = \begin{bmatrix}10 & 1\\ 10 & 0\end{bmatrix}, so its last row is (10,0)(10, 0). For ωn=1\omega_n = 1, ζ=0.7\zeta = 0.7: ϕ(s)=s2+1.4s+1\phi(s) = s^2 + 1.4s + 1, and

ϕ(A)=A2+1.4A+I=[0.870.0300.6],K=(10,0) ϕ(A)=(8.7, 0.3).\phi(A) = A^2 + 1.4A + I = \begin{bmatrix} 0.87 & 0.03 \\ 0 & 0.6 \end{bmatrix}, \qquad K = (10, 0)\,\phi(A) = (8.7,\ 0.3).

The same answer, as it must be. The double integrator is faster: matching s2+k2s+k1s^2 + k_2 s + k_1 with s2+2ζωns+ωn2s^2 + 2\zeta\omega_n s + \omega_n^2 gives K=[ωn2, 2ζωn]K = [\omega_n^2,\ 2\zeta\omega_n], which the lab confirms: K=[4,2.8]K = [4, 2.8] at ωn=2\omega_n = 2 and [16,5.6][16, 5.6] at ωn=4\omega_n = 4 (ζ=0.7\zeta = 0.7).

n=3n = 3. The formula is the same with 3×33\times3 matrices. For a chain of three integrators (B=(0,0,1)B = (0,0,1)) the closed loop is s3+k3s2+k2s+k1s^3 + k_3s^2 + k_2s + k_1, so the gains are the coefficients of ϕ\phi in reverse: poles at −1,−2,−3-1, -2, -3 give K=[6,11,6]K = [6, 11, 6]. The exercise checks this case.

Tracking a reference: feedforward and integral action

Regulating x\mathbf x to zero is not the job; making yy follow rr is. Two ways.

Feedforward gain NN. At steady state 0=(A−BK)x+BNr0 = (A - BK)\mathbf x + BNr and we want y=Cx=ry = C\mathbf x = r, so

N=−1C(A−BK)−1B.N = \frac{-1}{C(A - BK)^{-1}B} .

For the cruise plant the closed-loop transfer function is 0.1N/(s2+2ζωns+ωn2)0.1N/(s^2 + 2\zeta\omega_n s + \omega_n^2), since the plant’s numerator 0.10.1 is untouched by KK, so N=10ωn2N = 10\omega_n^2. At ωn=1\omega_n = 1, N=10N = 10 and yy settles at 11 with 4.60%4.60\% overshoot (ζ=0.7\zeta = 0.7 gives e−πζ/1−ζ2=4.60%e^{-\pi\zeta/\sqrt{1-\zeta^2}} = 4.60\%) and 2%2\% settling in 5.9855.985 s. The weakness: NN comes from the model. Give the car 20%20\% less authority (the lab’s model-error slider, real input gain 1+e1 + e with e=−0.2e = -0.2) and the feedforward design ends at y=0.97561y = 0.97561 instead of 11. At ωn=0.5\omega_n = 0.5 the miss is 0.909090.90909 and at ωn=4\omega_n = 4 it is 0.998440.99844: more loop gain, smaller miss, the sensitivity story of 13.1 Feedback & Block Diagrams again.

Integral augmentation. Add a state zz with z˙=r−y\dot z = r - y and feed it back: u=−Kx−kIzu = -K\mathbf x - k_Iz. The augmented plant has n+1n + 1 states,

[x˙z˙]=[A0−C0][xz]+[B0]u+[01]r,\begin{bmatrix}\dot{\mathbf x}\\ \dot z\end{bmatrix} = \begin{bmatrix} A & 0 \\ -C & 0\end{bmatrix}\begin{bmatrix}\mathbf x\\ z\end{bmatrix} + \begin{bmatrix} B \\ 0\end{bmatrix}u + \begin{bmatrix}0\\1\end{bmatrix}r ,

and we place n+1n + 1 poles with the same Ackermann formula. At steady state z˙=0\dot z = 0 forces y=ry = r whatever the plant gain is. Cruise plant, ωn=1\omega_n = 1, ζ=0.7\zeta = 0.7 and a third pole at −3-3 (the lab’s third-pole slider): K=[47.7, 3.3]K = [47.7,\ 3.3], kI=−30k_I = -30 (negative because zz integrates the error). The step has 4.27%4.27\% overshoot, settles in 6.336.33 s and peaks at a command of 7.20467.2046. With the same 20%20\% model error the output still ends at exactly 1.00001.0000, the command ends at 1.25=1/0.81.25 = 1/0.8 because it must make up the lost authority, the overshoot rises to 5.57%5.57\% and the settling to 6.486.48 s.

This is PID. On this plant a=10y˙+ya = 10\dot y + y, so −k1y−k2a=−(k1+k2)y−10k2y˙-k_1y - k_2a = -(k_1 + k_2)y - 10k_2\dot y: a proportional term 5151, a derivative term 3333 and the integral term 3030. Substituting Kp=51K_p = 51, Ki=30K_i = 30, Kd=33K_d = 33 into 13.3’s characteristic polynomial 10s3+(11+Kd)s2+(1+Kp)s+Ki10s^3 + (11 + K_d)s^2 + (1 + K_p)s + K_i gives 10s3+44s2+52s+3010s^3 + 44s^2 + 52s + 30, which is 10(s2+1.4s+1)(s+3)10(s^2 + 1.4s + 1)(s + 3): the three poles we asked for. (Here the reference enters only through the integrator.) So PID is state feedback for this plant, three gains placing three poles. It stops being one when the plant has more states than a PID has gains. The short period has two states, the lateral model four, and a full-state design uses all of them.

What pole placement does not tell you

The method is exact and the design can still be bad. Each warning has a lab number.

Effort grows like ωn2\omega_n^2. After a step in rr the first command is u(0)=Nru(0) = Nr. For the cruise plant N=10ωn2N = 10\omega_n^2, and the lab’s peak commands at ζ=0.7\zeta = 0.7 are 2.52.5 (ωn=0.5\omega_n = 0.5; the peak is 3.06423.0642 because the negative k2k_2 lets the command overshoot), 1010, 4040 and 160160 at ωn=1\omega_n = 1, 22 and 44. Doubling the speed quadruples the command, and settling time falls only from 5.9855.985 s to 3.0003.000 s and 1.5001.500 s. For the double integrator it is the same: 44 at ωn=2\omega_n = 2 and 1616 at ωn=4\omega_n = 4. A throttle saturates long before.

Saturation changes the loop. With the command limited to 1010 at ωn=2\omega_n = 2, the feedforward design requests 4040, receives 1010, and still settles (3.543.54 s, 3.32%3.32\% overshoot): there is no integrator to wind up. The integral design at ωn=2\omega_n = 2 (third pole at −3-3, K=[118.3,4.7]K = [118.3, 4.7], kI=−120k_I = -120) has 3.12%3.12\% overshoot and 3.2853.285 s settling unlimited, and with the same limit of 1010 the lab shows 16.75%16.75\% and 3.9753.975 s: windup, the lesson of 13.3, with no anti-windup in the lab.

Gain on noise. A sensor error in x\mathbf x reaches the actuator multiplied by KK. The cruise gain on yy is 8.78.7 at ωn=1\omega_n = 1 and 154.5154.5 at ωn=4\omega_n = 4, a factor of 17.817.8. Faster poles are noisier poles.

Unmodelled dynamics. The cruise model has no sensor lag, rate limit or delay; poles far to the left make the loop fast compared with exactly those neglected effects, and the delay margin of 13.5 Frequency Response & Bode Plots shrinks. Placement guarantees the poles of the model only.

Zeros do not move. Feedback changes the poles and leaves the zeros alone. The short-period output qq has a zero at s=−2.46664s = -2.46664 whatever KK is. Place the pair at ωn=6\omega_n = 6, ζ=0.7\zeta = 0.7 and the step overshoots by 58.6%58.6\% (128.0%128.0\% at ωn=10\omega_n = 10), far from the 4.6%4.6\% of a plain second-order pair. The peak elevator also scales as ωn2\omega_n^2 (3.64873.6487 and 10.135210.1352 rad per unit of qq reference: a ratio of 2.78=(10/6)22.78 = (10/6)^2, the stand-in BB making the units illustrative). Poles set the modes; zeros decide how much of each a reference excites.

A Module 12 example: placing the Dutch roll

The lateral model of 12.6 is a 4×44\times4 system in (v,p,r,ϕ)(v, p, r, \phi) with the rudder column (−4.2952,0.3568,5.0643,0)(-4.2952, 0.3568, 5.0643, 0) per radian. The lab uses those printed five-digit entries, so its eigenvalues are 12.6’s: roll −9.3790-9.3790, Dutch roll −0.4832±2.2898i-0.4832 \pm 2.2898i, spiral −0.00678-0.00678. The controllability matrix has rank 44: the rudder can move all four poles in principle.

The yaw damper of 12.6, δr=−Kr\delta_r = -Kr with K=0.5K = 0.5 s, moved the Dutch roll to −1.6647±1.5465i-1.6647 \pm 1.5465i but dragged the spiral to −0.1731-0.1731. Full-state feedback can move one mode and leave the others. Ask for a Dutch roll at ωn=2.5\omega_n = 2.5, ζ=0.7\zeta = 0.7, i.e. −1.7500±1.7854i-1.7500 \pm 1.7854i, and keep roll and spiral where they are (the lab’s checkbox). Ackermann on the 4×44\times4 gives

K=[−0.00315, −0.02571, 0.49943, −0.09087]K = [-0.00315,\ -0.02571,\ 0.49943,\ -0.09087]

in rad of rudder per (m/s, rad/s, rad/s, rad)(\text{m/s},\ \text{rad/s},\ \text{rad/s},\ \text{rad}), and the closed-loop eigenvalues come out −9.3790-9.3790, −1.7500±1.7854i-1.7500 \pm 1.7854i and −0.0068-0.0068. The yaw-rate gain 0.499430.49943 is very close to 12.6’s 0.50.5 s, and the other gains are small. The unit-reference command is only 0.02470.0247 rad. But the output (yaw rate) is slow: with the spiral kept at −0.0068-0.0068, the step reaches only 0.09500.0950 of the reference in 1515 s. A mode you leave alone stays as slow as you left it.

Move the roll and the spiral too, to −6.25-6.25 and −5-5 with the same Dutch roll, and the lab’s gains become K=[−0.32551, −8.23599, 1.17259, 2.47815]K = [-0.32551,\ -8.23599,\ 1.17259,\ 2.47815] with a rudder command of 12.1512.15 rad per unit of yaw-rate reference. At a reference of 0.050.05 rad/s that is 0.610.61 rad, about 35∘35^\circ of rudder, and the response overshoots by 572%572\%. Placement did what it was told; the aircraft cannot do it. These two experiments are why the next lesson exists.

◆ Interactive Lateral block of 12.6 with the rudder: place the Dutch roll at wn = 2.5, zeta = 0.7 and keep roll and spiral (checkbox), then untick it and watch the rudder demand explode

Loading simulation…

◆ Interactive Cruise plant with integral augmentation, input limit 10 and wn = 2: no anti-windup, so the overshoot grows; then switch to feedforward and add a model error

Loading simulation…

Placement against classical design

A root locus moves poles along curves with one gain; placement picks the destination and finds the gains. PID has three gains in a fixed structure; state feedback has one gain per state but needs the states. The margins of 13.5 remain the test of the finished loop.

Concept checks

✅ Concept check

With $u = -K\mathbf x + Nr$ applied to $\dot{\mathbf x} = A\mathbf x + Bu$, the closed-loop poles are the eigenvalues of:

✅ Concept check

For the double integrator, $\ddot x = u$ with $u = -k_1 x - k_2\dot x$, what is $k_1$ for a critically damped pair ($\zeta = 1$) at $\omega_n = 3$?

✅ Concept check

For the cruise plant, the command at the instant a unit reference step arrives is $N = 10\omega_n^2$. What is it for $\omega_n = 3$?

✅ Concept checkselect all that apply

Select every statement that is true.

Playground and exercise

The first cell applies Ackermann’s formula at four speeds, the second simulates the step and measures the peak command (with SciPy’s place_poles as a cross-check), and the exercise has you write Ackermann’s formula yourself.

🐍 PythonPlayground: Ackermann's formula on the cruise plant at four speeds
⌘/Ctrl + Enter
🐍 PythonPlayground: simulate the step, measure the peak command, cross-check with SciPy
⌘/Ctrl + Enter
🧪 Coding exerciseExercise: controllability and Ackermann's formula
⌘/Ctrl + Enter

Problems

✏️ Pen & paperPlace a double integrator●○○

For x¨=u\ddot x = u with u=−k1x−k2x˙u = -k_1x - k_2\dot x, find KK that puts the poles at −2-2 and −4-4, and say what ζ\zeta and ωn\omega_n that is.

Show worked solution

The closed loop is s2+k2s+k1s^2 + k_2s + k_1, to match (s+2)(s+4)=s2+6s+8(s+2)(s+4) = s^2 + 6s + 8. So K=[8, 6]K = [8,\ 6]. The product of the roots is ωn2=8\omega_n^2 = 8, ωn=2.828\omega_n = 2.828, and 2ζωn=62\zeta\omega_n = 6 gives ζ=1.061\zeta = 1.061: overdamped, with no overshoot.

✏️ Pen & paperReal poles for the cruise plant●●○

For the cruise plant, find KK and NN that put the closed-loop poles at −1-1 and −2-2.

Show worked solution

From worked example 1 the closed-loop polynomial is s2+(1.1+k2)s+0.1(1+k1+k2)s^2 + (1.1 + k_2)s + 0.1(1 + k_1 + k_2). Match s2+3s+2s^2 + 3s + 2. Then 1.1+k2=31.1 + k_2 = 3 gives k2=1.9k_2 = 1.9, and 0.1(1+k1+1.9)=20.1(1 + k_1 + 1.9) = 2 gives k1=17.1k_1 = 17.1. The closed-loop transfer function is 0.1N/(s2+3s+2)0.1N/(s^2 + 3s + 2) with DC gain 0.05N0.05N, so N=20N = 20. Check: A−BK=[−0.10.1−17.1−2.9]A - BK = \begin{bmatrix}-0.1 & 0.1\\ -17.1 & -2.9\end{bmatrix} has trace −3-3 and determinant 0.29+1.71=20.29 + 1.71 = 2.

✏️ Pen & paperWhich input can reach everything?●●○

For A=[−110−2]A = \begin{bmatrix} -1 & 1 \\ 0 & -2\end{bmatrix} decide whether B=(1,0)B = (1, 0) and B=(0,1)B = (0, 1) give a controllable pair. For the one that fails, find the pole that cannot move.

Show worked solution

B=(1,0)B = (1, 0): AB=(−1,0)AB = (-1, 0) is parallel to BB, so [B,AB][B, AB] has rank 11. A left eigenvector with wB=0\mathbf wB = 0 must be w=(0,1)\mathbf w = (0, 1), and wA=(0,−2)=−2w\mathbf wA = (0, -2) = -2\mathbf w: the pole at −2-2 cannot move (the second state is untouched by the input and obeys x˙2=−2x2\dot x_2 = -2x_2). B=(0,1)B = (0, 1): AB=(1,−2)AB = (1, -2) and det⁡[011−2]=−1\det\begin{bmatrix}0&1\\1&-2\end{bmatrix} = -1, controllable. The input enters the second state, which feeds the first through the coupling.

What comes next

Pole placement answers “where should the poles be?” and leaves the harder question open. For the cruise plant, ωn=1\omega_n = 1 was a guess; for the lateral aircraft the “obvious” placement asked for 12.1512.15 rad of rudder. The next lesson, 13.7 on LQR and state estimation (13.7 LQR & State Estimation), chooses KK by optimisation instead: pay a cost for state error and a cost for control effort, and let the compromise decide where the poles land. It also answers where the states come from: an observer is a model run alongside the plant that estimates the states you cannot measure, so the full-state feedback of this lesson becomes usable.

Key Takeaways

  • State-space form x˙=Ax+Bu\dot{\mathbf x} = A\mathbf x + Bu, y=Cxy = C\mathbf x; the cruise plant is two states (y,a)(y, a) with eigenvalues −0.1-0.1 and −1-1.
  • With u=−Kx+Nru = -K\mathbf x + Nr the closed loop is A−BKA - BK, so the poles are its eigenvalues. On the cruise plant, K=[8.7,0.3]K = [8.7, 0.3] places −0.7±0.7141i-0.7 \pm 0.7141i (ωn=1\omega_n = 1, ζ=0.7\zeta = 0.7).
  • Placement needs [B,AB,… ][B, AB, \dots] to have rank nn; a left eigenvector blind to the input fixes its eigenvalue for every KK.
  • Ackermann’s formula K=[0⋯0 1]Wc−1ϕ(A)K = [0\cdots0\ 1]W_c^{-1}\phi(A) (from memory) places poles for one input; for n=2n = 2 matching coefficients is just as good.
  • N=−1/(C(A−BK)−1B)N = -1/(C(A-BK)^{-1}B) tracks only if the model is exact (0.975610.97561 with a 20%20\% gain error); integral augmentation tracks regardless (1.00001.0000), and on the cruise plant it is a PID with Kp=51K_p = 51, Ki=30K_i = 30, Kd=33K_d = 33.
  • Effort and noise gain grow fast: the cruise step command is 10ωn210\omega_n^2 (1010, 4040, 160160 at ωn=1,2,4\omega_n = 1, 2, 4), and zeros do not move (short period: 58.6%58.6\% overshoot at ωn=6\omega_n = 6).
  • On the lateral aircraft, moving only the Dutch roll to −1.75±1.7854i-1.75 \pm 1.7854i needs Kr=0.49943K_r = 0.49943, close to 12.6’s yaw damper; dragging every pole costs 12.1512.15 rad of rudder per unit of yaw rate.
Finished this lesson?

Mark it complete to track your progress (saved in this browser). Reading every section does this for you.

Where this leads

These later lessons build directly on this one: