Predictive Control with Constraints
ERRORS DISCOVERED SO FAR

Date: 25 September 2001

Page xiii

`Downloadable solution's manual' should be `Downloadable solutions manual'.

Page 12

There is an error in equation (1.23) in the book. Since the free response $\hat{y}_f(k+P_i\vert k)$ is defined to be the response when the input remains at its last value, namely $u(k-1)$, each term in (1.23) involving an input value $\hat{u}(k+j\vert k)$ should in fact involve the difference $\hat{u}(k+j\vert k)-u(k-1)$. Thus the correct expression for (1.23) is:

$$\begin{multline*} \hat{y}(k+P_i\vert k) = \hat{y}_f(k+P_i\vert k)+ H(P_i)[\hat... ...\vert k)-u(k-1)]+ S(P_i-H_u+1)[\hat{u}(k+H_u-1\vert k)-u(k-1)] \end{multline*}$$

Subsequent expressions are correct.

Page 47

Last line of Example 2.4: The dimensions given of matrices $E$ and $F$ are wrong. They should be $16\times 9$ in each case. The dimensions given for the matrix $G$ are correct.

Page 70

Exercise 2.3: The formula for the gradient is not (2.19), but the formula for $\nabla V$ which appears at the end of Mini-Tutorial 1.

Page 70

Exercise 2.4: The question should refer to Example 2.3, not Example 2.4. (But the reader could answer the question for Example 2.4 too; it is a bit more complicated.)

Page 72

Exercise 2.12: Not really an error, but the solution to this exercise will probably be clearer after reading sections 3.1 and 3.2 in the next chapter.

Page 73

Exercise 2.15(b): Actually it is much easier to design the state observer by hand than by using the Model Predictive Control Toolbox in this case. But you need to use the Toolbox representation of the model in order to perform the simulation in (c) using the Toolbox.

Pages 79 and 81

Figures 3.1 and 3.2: The signal labelled $\Delta(k)$ should be labelled $\Delta u(k)$ (as in Figure 3.3).

Page 105

Exercise 3.4: The reference to equation (2.23) is incorrect. It should be to equation (2.66). (I suppose this should have been a Chapter 2 exercise really.)

Page 106

Exercise 3.10: The reference to equation (3.4) is spurious. The question should read: `Check that the optimization problems (3.88) and (3.90)-(3.91) are standard QP problems.'

Page 215

Exercise 7.5(c): The parameter value $H_p=30$ is too low; this value gives instability. Use $H_p=60$ instead.

Page 226

After `2. Control:' $K_{LQ}$ should be $K_\infty$ (twice) to make it consistent with Figure 8.3.

Page 319

Table C.1: The Model Predictive Control Toolbox parameters ywt and uwt correspond to $Q(i)^{1/2}$ and $R(i)^{1/2}$, respectively (that is, to $S_{\cal Q}$ and $S_{\cal R}$ in section 3.1.2).