Sie sind auf Seite 1von 4

1) Obtain a realization in the observer canonical form.

The transfer function is:

10𝑧 2 + 5𝑧
𝐺(𝑧) = 2
𝑧 + 5𝑧 − 10

Divide always by the highest exponential value and it does not matter if it is
denominator or the numerator, which is this case is 𝑧 2

10 + 5𝑧 −1 𝑌(𝑧)
𝐺(𝑧) = ; 𝐺(𝑧) ≜ ;
1 + 5𝑧 −1 − 10𝑧 −2 𝑈(𝑧)

𝑌(𝑧) 10 + 5𝑧 −1
= → (𝑌(𝑧))(1 + 5𝑧 −1 − 10𝑧 −2 ) = (𝑈(𝑧))(10 + 5𝑧 −1 );
𝑈(𝑧) 1 + 5𝑧 −1 − 10𝑧 −2

𝑌(𝑧) 𝑈(𝑧)
= = 𝑀(𝑧);
10 + 5𝑧 −1 1 + 5𝑧 −1 − 10𝑧 −2

𝑈(𝑧)
−1 −2
= 𝑀(𝑧) → 𝑈(𝑧) = 𝑀(𝑧) + 5𝑧 −1 𝑀(𝑧) − 10𝑧 −2 𝑀(𝑧)
1 + 5𝑧 − 10𝑧

𝑌(𝑧)
= 𝑀(𝑧) → 𝑌(𝑧) = 10𝑀(𝑧) + 5𝑧 −1 𝑀(𝑧);
10 + 5𝑧 −1

Let define: 𝑋1 (𝑧) = 𝑧 −2 𝑀(𝑧); 𝑋2 (𝑧) = 𝑧 −1 𝑀(𝑧);

This leads to the next two equations 𝑋2 (𝑧) = 𝑧 −1 𝑀(𝑧) → 𝑧𝑋2 (𝑧) = 𝑀(𝑧);

𝑋1 (𝑧) = 𝑧 −2 𝑀(𝑧) → 𝑧 2 𝑋1 (𝑧) = 𝑀(𝑧); We know that 𝑧𝑋2 (𝑧) = 𝑀(𝑧); therefore we
set 𝑀(𝑧) equal to each other to get the final part 𝑧 2 𝑋1 (𝑧) = 𝑧𝑋2 (𝑧) → 𝑧𝑋1 (𝑧) =
𝑋2 (𝑧); Therefore all the equation that are need is:

𝑋1 (𝑧) = 𝑧 −2 𝑀(𝑧); 𝑋2 (𝑧) = 𝑧 −1 𝑀(𝑧); 𝑧𝑋2 (𝑧) = 𝑀(𝑧); 𝑧𝑋1 (𝑧) = 𝑋2 (𝑧);

This now means we can start plugging in with the knowledge that
𝑧 𝑛 𝑋𝑚 (𝑧) ⟺ 𝑥𝑚 (𝑘 + 𝑛):

𝑧𝑋1 (𝑧) = 𝑋2 (𝑧) ⟺ 𝑥1 (𝑘 + 1) = 𝑥2 (𝑘);

𝑈(𝑧) = 𝑧𝑋2 (𝑧) + 5𝑋2 (𝑧) − 10𝑋1 (𝑧) ⟺ 𝑢(𝑘) = 𝑥2 (𝑘 + 1) + 5𝑥2 (𝑘) − 10𝑥1 (𝑘);
𝑌(𝑧) = 10𝑧𝑋2 (𝑧) + 5𝑋2 (𝑧) ⟺ 𝑦(𝑘) = 10𝑥2 (𝑘 + 1) + 5𝑥2 (𝑘);

For 𝑦(𝑘), the function never have 𝑥𝑚 (𝑘 + 𝑛), but that is not a problem, as
𝑢(𝑘) = 𝑥2 (𝑘 + 1) + 5𝑥2 (𝑘) − 10𝑥1 (𝑘) →
𝑥2 (𝑘 + 1) = 10𝑥1 (𝑘) − 5𝑥2 (𝑘) + 𝑢(𝑘), therefore

𝑦(𝑘) = 10(10𝑥1 (𝑘) − 5𝑥2 (𝑘) + 𝑢(𝑘)) + 5𝑥2 (𝑘);


𝑦(𝑘) = 100𝑥1 (𝑘) − 50𝑥2 (𝑘) + 10𝑢(𝑘) + 5𝑥2 (𝑘);
𝑦(𝑘) = 100𝑥1 (𝑘) − 45𝑥2 (𝑘) + 10𝑢(𝑘);

Final equations:

𝑥1 (𝑘 + 1) = 0𝑥1 (𝑘) + 𝑥2 (𝑘) + 0𝑢(𝑘);


𝑥2 (𝑘 + 1) = 10𝑥1 (𝑘) − 5𝑥2 (𝑘) + 𝑢(𝑘);
𝑦(𝑘) = −100𝑥1 (𝑘) − 45𝑥2 (𝑘) + 10𝑢(𝑘);

𝑥 (𝑘 + 1) 0 1 𝑥1 (𝑘) 0
[ 1 ]=[ ][ ] + [ ] 𝑢(𝑘);
𝑥2 (𝑘 + 1) 10 −5 𝑥2 (𝑘) 1

𝑥1 (𝑘)
[𝑦(𝑘)] = [100 −45] [ ] + [10]𝑢(𝑘);
𝑥2 (𝑘)

Last step is to find the Canonical Form as it was typed in Phase Variable Form:

𝑥 (𝑘 + 1) −5 1 𝑥1 (𝑘) −45
[ 1 ]=[ ][ ]+[ ] 𝑢(𝑘);
(𝑘
𝑥2 + 1) 10 0 𝑥2 (𝑘) 100

𝑥 (𝑘)
[𝑦(𝑘)] = [1 0] [ 1 ] + [10]𝑢(𝑘);
𝑥2 (𝑘)

To check if the answer is correct use the formula 𝐺(𝑧) = det ((𝐶⃗)(𝑧𝐼⃗ − 𝐴⃗)(𝐵⃗⃗ ) + 𝐷
⃗⃗ ),
remember order matters in matrix algebra and it is always from left to right:

Here is WolframAlpha agreeing to check if the answer is correct

http://www.wolframalpha.com/input/?i=det(%7B%7B1,0%7D%7D+%7B%7B5+%2B+z,+-
1%7D,+%7B-10,+z%7D%7D%5E-1+%7B%7B-45%7D,%7B100%7D%7D+%2B+%7B10%7D)
2) Consider the state equation:

0 1 0
𝑥⃗(𝑘 + 1) = [ ] 𝑥⃗(𝑘) + [ ] 𝑢(𝑘);
−2 −3 1

[𝑦⃗(𝑘)] = [3 1]𝑥⃗(𝑘);

Obtain the transfer function with the formula 𝐺(𝑧) = det ((𝐶⃗)(𝑧𝐼⃗ − 𝐴⃗)(𝐵
⃗⃗ ) + 𝐷
⃗⃗) ;

1 0 0 1 𝑧 0 0 1 𝑧 −1
𝑧𝐼⃗ − 𝐴⃗ = (𝑧) [ ]−[ ]=[ ]−[ ]=[ ];
0 1 −2 −3 0 𝑧 −2 −3 2 𝑧+3

Now we need to find the inverse, with the equation


𝑇
(𝑐𝑜𝑓𝑎𝑐𝑡𝑜𝑟(𝐽⃗))
𝐽⃗−1 = ;
det(𝐽⃗)

Remember: 𝑎𝑚,𝑛 where 𝑚 is the one that goes up and down (vertically) and 𝑛 is the
one that goes side to side (horizontally); Let’s find the cofactor first, which is:

𝑎1,1 = (−1)1+1 (𝑧 + 3) = 𝑧 + 3; 𝑎1,2 = (−1)1+2 (2) = −2;


𝑎2,1 = (−1)2+1 (−1) = 1; 𝑎2,2 = (−1)2+2 (𝑧) = 𝑧 ∴

𝑧 + 3 −2
𝑐𝑜𝑓𝑎𝑐𝑡𝑜𝑟(𝐽⃗) = [ ];
1 𝑧

Now find the transpose of the cofactor of 𝐽⃗, which leads to:
𝑇
𝑧+3 1
(𝑐𝑜𝑓𝑎𝑐𝑡𝑜𝑟(𝐽⃗)) = [ ];
−2 𝑧

Last thing to do is find det(𝐽⃗) = (𝑧)(𝑧 + 3) + 2 = 𝑧 2 + 3𝑧 + 2 = (𝑧 + 2)(𝑧 + 1) ∴

𝑧+3 1
1 𝑧+3 1 (𝑧 + 2)(𝑧 + 1) (𝑧 + 2)(𝑧 + 1)
𝐽⃗−1 = [ ]= ;
(𝑧 + 2)(𝑧 + 1) −2 𝑧 −2 𝑧
[(𝑧 + 2)(𝑧 + 1) (𝑧 + 2)(𝑧 + 1)]

The next step is (𝐶⃗)(𝑧𝐼⃗ − 𝐴⃗):


𝑧+3 1
(𝑧 + 2)(𝑧 + 1) (𝑧 + 2)(𝑧 + 1)
(𝐶⃗)(𝑧𝐼⃗ − 𝐴⃗) = [3 1] =
−2 𝑧
[(𝑧 + 2)(𝑧 + 1) (𝑧 + 2)(𝑧 + 1)]
3𝑧 + 7 𝑧+3
[ ]
(𝑧 + 1)(𝑧 + 2) (𝑧 + 1)(𝑧 + 2)

Remember the way you multiply matrix is horizontal of the right sided matrix to the
vertical side of the matrix, now finally is doing (𝐶⃗)(𝑧𝐼⃗ − 𝐴⃗)(𝐵
⃗⃗ ) as 𝐷
⃗⃗ doesn’t exist for
the particular problem:

3𝑧 + 7 𝑧+3 0 𝑧+3
(𝐶⃗)(𝑧𝐼⃗ − 𝐴⃗)(𝐵
⃗⃗ ) = [ ][ ] = [ ];
(𝑧 + 1)(𝑧 + 2) (𝑧 + 1)(𝑧 + 2) 1 (𝑧 + 1)(𝑧 + 2)

The final answer for the is problem is:

𝑧+3 𝑧+3
𝐺(𝑧) = 𝑑𝑒𝑡 [ ] → 𝐺(𝑧) = ;
(𝑧 + 1)(𝑧 + 2) (𝑧 + 1)(𝑧 + 2)

Here is WolframAlpha agreeing to check if the answer is correct

http://www.wolframalpha.com/input/?i=det(%7B%7B3,1%7D%7D+%7B%7Bz,+-
1%7D,+%7B2,+z+%2B+3%7D%7D%5E(-1)+%7B%7B0%7D,%7B1%7D%7D)