一、齐次矩阵微分方程的解
齐次矩阵微分方程
x
˙
=
A
x
\color{red}\pmb{\dot{x}=Ax}
x˙=Ax,初始状态给定为
x
(
0
)
=
x
0
\pmb {x(0)=x_0}
x(0)=x0,则微分方程有唯一解。
x
(
t
)
=
e
A
t
x
0
,
t
≥
t
0
\pmb{x(t)}=e^{\pmb{A}t}\pmb{x_0},\quad t\geq t_0
x(t)=eAtx0,t≥t0
证明:假设解 x ( t ) \pmb x(t) x(t) 为 t t t 的矢量幂级数形式
x ( t ) = b 0 + b 1 t + b 2 t 2 + ⋯ + b k t k + ⋯ \color{red}\pmb{x(t)}=\pmb{b_0}+\pmb{b_1}t+\pmb{b_2}t^2+\cdots+\pmb{b_k}t^k+\cdots x(t)=b0+b1t+b2t2+⋯+bktk+⋯ 将其代入微分方程得
b 1 + 2 b 2 t + 3 b 3 t 2 + ⋯ + k b k t k − 1 + ⋯ = A ( b 0 + b 1 t + b 2 t 2 + ⋯ + b k t k + ⋯ ) \pmb{b_1}+2\pmb{b_2}t+3\pmb{b_3}t^2+\cdots+k\pmb{b_k}t^{k-1}+\cdots=\pmb{A}(\pmb{b_0}+\pmb{b_1}t+\pmb{b_2}t^2+\cdots+\pmb{b_k}t^k+\cdots) b1+2b2t+3b3t2+⋯+kbktk−1+⋯=A(b0+b1t+b2t2+⋯+bktk+⋯) 可得 b 0 = x 0 , b 1 = A b 0 , b 2 = 1 2 A b 1 = 1 2 ! A 2 b 0 , ⋯ , b k = 1 k ! A k b 0 \pmb{b_0=x_0},\pmb{b_1=Ab_0},\pmb{b_2}=\dfrac{1}{2}\pmb{Ab_1}=\dfrac{1}{2!}\pmb{A^2b_0},\cdots,\pmb{b_k}=\dfrac{1}{k!}\pmb{A^kb_0} b0=x0,b1=Ab0,b2=21Ab1=2!1A2b0,⋯,bk=k!1Akb0
将其代入幂级数表达式,得到x ( t ) = ( I + A t + 1 2 ! A 2 t 2 + ⋯ + 1 k ! A k t k + ⋯ ) x 0 \pmb{x(t)}=(\pmb{I}+\pmb{A}t+\dfrac{1}{2!}\pmb{A^2}t^2+\cdots+\dfrac{1}{k!}\pmb{A^k}t^k+\cdots)\pmb{x_0} x(t)=(I+At+2!1A2t2+⋯+k!1Aktk+⋯)x0 即
x ( t ) = e A t x 0 \color{red}\pmb{x(t)}=e^{\pmb{A}t}\pmb{x_0} x(t)=eAtx0
二、非齐次矩阵微分方程的解
1、若 A = Λ = [ λ 1 ⋯ 0 ⋮ ⋱ ⋮ 0 ⋯ λ n ] \pmb{A}=\pmb{\Lambda}=\left[\begin{matrix} \lambda_1&\cdots&0\\ \vdots&\ddots&\vdots\\ 0&\cdots&\lambda_n\\\end{matrix}\right] A=Λ= λ1⋮0⋯⋱⋯0⋮λn ,则 e A t = Φ ( t ) = [ e λ 1 t ⋯ 0 ⋮ ⋱ ⋮ 0 ⋯ e λ n t ] e^{\pmb{A}t}=\Phi(t)=\left[\begin{matrix} e^{\lambda_1t}&\cdots&0\\ \vdots&\ddots&\vdots\\ 0&\cdots&e^{\lambda_nt}\\\end{matrix}\right] eAt=Φ(t)= eλ1t⋮0⋯⋱⋯0⋮eλnt
2、若 A \pmb A A 可对角化,即 P − 1 A P = Λ \pmb{P^{-1}AP=\Lambda} P−1AP=Λ,则 e A t = Φ ( t ) = P [ e λ 1 t ⋯ 0 ⋮ ⋱ ⋮ 0 ⋯ e λ n t ] P − 1 e^{\pmb{A}t}=\Phi(t)=\pmb{P}\left[\begin{matrix} e^{\lambda_1t}&\cdots&0\\ \vdots&\ddots&\vdots\\ 0&\cdots&e^{\lambda_nt}\\\end{matrix}\right]\pmb{P^{-1}} eAt=Φ(t)=P eλ1t⋮0⋯⋱⋯0⋮eλnt P−1
3、非齐次矩阵微分方程
x
˙
=
A
x
+
B
u
\color{red}\pmb{\dot{x}=Ax+Bu}
x˙=Ax+Bu,初始状态给定为
x
(
0
)
=
x
0
\pmb {x(0)=x_0}
x(0)=x0,则微分方程的解为
x
(
t
)
=
Φ
(
t
)
x
(
t
)
+
∫
0
t
Φ
(
t
−
τ
)
B
u
(
τ
)
d
τ
,式中
Φ
(
t
)
=
e
A
t
\color{red}\pmb{x}(t)=\pmb{\Phi}(t)\pmb{x}(t)+\int_0^t\pmb{\Phi}(t-\tau)\pmb{Bu}(\tau)d\tau,式中\pmb{\Phi}(t)=e^{\pmb{A}t}
x(t)=Φ(t)x(t)+∫0tΦ(t−τ)Bu(τ)dτ,式中Φ(t)=eAt
解由两部分组成 { ①表示由初始状态引起的自由运动 ②表示由控制激励作用引起的强制运动 \color{blue}\quad解由两部分组成\begin{cases} \color{black}①表示由初始状态引起的自由运动\\ \color{black}②表示由控制激励作用引起的强制运动 \end{cases} 解由两部分组成{①表示由初始状态引起的自由运动②表示由控制激励作用引起的强制运动
证明: x ˙ − A x = B u \color{red}\pmb{\dot{x}-Ax=Bu} x˙−Ax=Bu 两边同时左乘 e − A t e^{-\pmb{A}t} e−At
即e − A t [ x ˙ ( t ) − A x ( t ) ] = e − A t B u ( t ) e^{-\pmb{A}t}\pmb{[\dot{x}}(t)-\pmb{Ax}(t)\pmb{]}=e^{-\pmb{A}t}\pmb{Bu}(t) e−At[x˙(t)−Ax(t)]=e−AtBu(t)
即d d t [ e − A t x ( t ) ] = e − A t B u ( t ) \dfrac{d}{dt}\pmb{\big[}e^{-\pmb{A}t}\pmb{x}(t)\pmb{\big]}=e^{-\pmb{A}t}\pmb{Bu}(t) dtd[e−Atx(t)]=e−AtBu(t)
在 0 ∼ t 0\sim t 0∼t 间积分,e − A t x ( t ) − x 0 = ∫ 0 t e − A t B u ( τ ) d τ e^{-\pmb{A}t}\pmb{x}(t)-\pmb{x_0}=\int_0^te^{-\pmb{A}t}\pmb{Bu}(\tau)d\tau e−Atx(t)−x0=∫0te−AtBu(τ)dτ
即e − A t x ( t ) = x 0 + ∫ 0 t e − A τ B u ( τ ) d τ e^{-\pmb{A}t}\pmb{x}(t)=\pmb{x_0}+\int_0^te^{-\pmb{A}\tau}\pmb{Bu}(\tau)d\tau e−Atx(t)=x0+∫0te−AτBu(τ)dτ
两边同时左乘 e A t e^{\pmb{A}t} eAt,则x ( t ) = e A t x 0 + ∫ 0 t e A ( t − τ ) B u ( τ ) d τ x ( t ) = Φ ( t ) x 0 + ∫ 0 t Φ ( t − τ ) B u ( τ ) d τ \begin{aligned}\pmb{x}(t)&=e^{\pmb{A}t}\pmb{x_0}+\int_0^te^{\pmb{A}(t-\tau)}\pmb{Bu}(\tau)d\tau\\ \color{red}\pmb{x}(t)&\color{red}=\pmb{\Phi}(t)\pmb{x_0}+\int_0^t\pmb{\Phi}(t-\tau)\pmb{Bu}(\tau)d\tau\end{aligned} x(t)x(t)=eAtx0+∫0teA(t−τ)Bu(τ)dτ=Φ(t)x0+∫0tΦ(t−τ)Bu(τ)dτ