bappJ_epdf 12/18/2018 21:3:4 Page 1

Appendix J

Derivation of the Time Domain Solution

of State Equations

J.1 Derivation
Rather than using the Laplace transformation, we can solve the equations directly in the time
domain using a method closely allied to the classical solution of differential equations. We
will find that the final solution consists of two parts that are different from the forced and
natural responses
First, assume a homogeneous state equation of the form
xt Ax t J1
Since we want to solve for x, we assume a series solution, just as we did in elementary scalar
differential equations. Thus,
xt b0 b1 t b2 t 2 ∙∙∙ bk t k bk 1 t k 1
∙∙∙ J2

Substituting Eq. (J.2) into (J.1) we get

b1 2b2 t ∙∙∙ kbk t k 1 k 1 bk 1 t k ∙∙∙
J3
A b0 b1 t b2 t 2 ∙ ∙ ∙ bk t k bk 1 t k 1
∙∙∙

Equating like coefficients yields

b1 Ab0 J 4a
1 1 2
b2 Ab1 A b0 J 4b
2 2

1 k
bk A b0 J 4c
k
1
bk 1 Ak 1 b0 J 4d
k 1

Substituting these values into Eq. (J.2) yields

1 2 2 1 k k 1
xt b0 Ab0 t A b0 t ∙∙∙ A b0 t Ak 1 b0 t k 1
∙∙∙
2 k k 1
J5
1 22 1 kk 1
I At A t ∙∙∙ At Ak 1 t k 1
∙∙∙ b0
2 k k 1

A-161
bappJ_epdf 12/18/2018 21:3:4 Page 2

A-162 Appendix J Derivation of the Time Domain Solution of State Equations

But, from Eq. (J.2),

1
x0 b0 J6

Therefore,
1 22 1 kk 1
xt I At A t ∙∙∙ At Ak 1 t k 1
∙∙∙ x0 J7
2 k k 1

Let
1 22 1 kk 1
eAt I At A t ∙∙∙ At Ak 1 t k 1
∙∙∙ J8
2 k k 1

where eAt is simply a notation for the matrix formed by the right-hand side of Eq. (J.8). We
use this definition because the right-hand side of Eq. (J.8) resembles a power series
expansion of eat, or
1 22 1 kk 1
eat 1 at a t ∙∙∙ at ak 1 t k 1
∙∙∙ J9
2 k k 1

Using Eq. (J.7), we have

xt eAt x 0 J 10

We give a special name to eAt: it is called the state-transition matrix2, since it performs a
transformation on x(0), taking x from the initial state, x(0), to the state x(t) at any time, t. The
symbol, Φ(t), is used to denote eAt. Thus,

Φt eAt J 11

and
xt Φt x0 J 12

There are some properties of Φ(t) that we will use later when we solve for x(t) in the
text. From Eq. (J.12),
x0 Φ0x0 J 13

Hence, the first property of Φ(t) is

Φ0 I J 14

where I is the identity matrix. Also, differentiating Eq. (J.12) and setting this equal to
Eq. (J.1) yields

xt Φt x0 Ax t J 15

which, at t = 0, yields
Φ0x0 Ax 0 J 16

1
In this development we consider the initial time, t0, to be 0. More generally, t0 ≠ 0. After completing this
development, the interested reader should consult Appendix K on www.wiley.com/college/nise for the more
general solution in terms of initial time t0 ≠ 0.
2
The state-transition matrix here is for the initial time t0 = 0. The derivation in Appendix K on at www.wiley.com/
go/Nise/ControlSystemsEngineering8e for t0 ≠ 0 yields x t eA t t0 x t0 .
bappJ_epdf 12/18/2018 21:3:4 Page 3

Bibliography A-163

Thus, the second property of Φ(t) follows from Eq. (J.16):

Φ0 A J 17
In summary, the solution to the homogeneous, or unforced, system is
xt Φt x0 J 18
where
Φ0 I J 19
and
Φ0 A J 20
Let us now solve the forced, or nonhomogeneous, problem. Given the forced state
equation
x t Ax t Bu t J 21
At
rearrange and multiply both sides by e :
At At
e xt Ax t e Bu t J 22
At
Realizing that the left-hand side is equal to the derivative of the product e x(t), we obtain
d At At
e xt e Bu t J 23
dt
Integrating both sides yields
t
At t At Aτ
e xt 0
e xt x0 e Bu τ dτ J 24
0

since e At evaluated at t = 0 is the identity matrix (from Eq. (J.8)). Solving for x(t) in
Eq. (J.24) we obtain
t
At At τ
xt e x0 e Bu τ dτ
0
t J 25
Φt x0 Φt τ Bu τ dτ
0

where Φ(t) = eAt by definition.

Bibliography
Timothy, L. K., and Bona, B. E. State Space Analysis: An Introduction. McGraw-Hill, New York,
1968.