訊號與系統(18)-連續時間系統的複頻域分析:拉普拉斯反變換之分部分式展開法

Zhongzheng Wang發表於2020-12-01

拉普拉斯的正變換相對容易,但是反變換中,其積分表現為一個在 s s s平面的二維複函式積分,其積分相對複雜,如果使用單純的公式計算求解會有很多麻煩。這樣的積分的求法通常有兩種常用方法:分部分式展開法以及留數法。這一部分內容將對分部分式展開法進行介紹。

1. 分部分式展開法

分部分式展開法,是根據拉普拉斯變換的線性特性,將複雜的拉普拉斯變換展開為多個簡單的拉普拉斯變換的組合,並通過對這些簡單拉普拉斯變換求解他們的反變換,依據拉普拉斯變換的齊次性和疊加性,進而得到原拉普拉斯變換的拉普拉斯反變換。

F ( s ) F(s) F(s)為某訊號的拉普拉斯變換,如果F(s)可以展開為一個有理函式形式,即:
F ( s ) = N ( s ) D ( s ) = b m s m + b m − 1 s m − 1 + b m − 2 s m − 2 + ⋯ + b 1 s + b 0 a n s n + a n − 1 s m − 1 + a n − 2 s m − 2 + ⋯ + a 1 s + a 0 F(s)=\frac{N(s)}{D(s)}=\frac{b_ms^m+b_{m-1}s^{m-1}+b_{m-2}s^{m-2}+\cdots +b_1s+b_0}{a_ns^n+a_{n-1}s^{m-1}+a_{n-2}s^{m-2}+\cdots +a_{1}s+a_0} F(s)=D(s)N(s)=ansn+an1sm1+an2sm2++a1s+a0bmsm+bm1sm1+bm2sm2++b1s+b0
F ( s ) F(s) F(s)可以進行分部分式展開,表示為多個簡單的有理分式之和。

在進行分部分式展開式,根據 m m m n n n的關係,以及 D ( s ) = 0 D(s)=0 D(s)=0有無重根的情況,需要對展開方式及其相應的反變換的方法進行分類討論。

1.1 m < n m<n m<n,且 D ( s ) = 0 D(s)=0 D(s)=0沒有重根

1.1.1 沒有重根,且根均為實數時

D ( s ) = 0 D(s)=0 D(s)=0的根為 s 1 s_1 s1 s 2 s_2 s2,……, s n s_n sn,則可以將 F ( s ) F(s) F(s)表示為:
F ( s ) = K 1 s − s 1 + K 2 s − s 2 + ⋯ + K n s − s n = ∑ i = 1 n K i s − s i \begin{aligned} F(s)&=\frac{K_1}{s-s_1}+\frac{K_2}{s-s_2}+\cdots+\frac{K_n}{s-s_n} \\&=\sum_{i=1}^{n}\frac{K_i}{s-s_i} \end{aligned} F(s)=ss1K1+ss2K2++ssnKn=i=1nssiKi
其中係數 K i K_i Ki的求法有很多,常見以下三種:

  • 係數平衡法:即將 ∑ i = 1 n K i s − s i \sum_{i=1}^{n}\frac{K_i}{s-s_i} i=1nssiKi進行通分,得到帶有係數 K i K_i Ki的分子,通過對比等式左邊的 F ( s ) F(s) F(s)分子的係數,得到有關 K i K_i Ki的方程組,從而計算得到係數的值。

    如求解: F ( s ) = 1 s 2 + 3 s + 2 F(s)=\frac{1}{s^2+3s+2} F(s)=s2+3s+21

    其待定係數的分部分式為: K 1 s + 1 + K 2 s + 2 \frac{K_1}{s+1}+\frac{K_2}{s+2} s+1K1+s+2K2

    通分後得: K 1 ( s + 1 ) + K 2 ( s + 2 ) ( s + 1 ) ( s + 2 ) \frac{K_1(s+1)+K_2(s+2)}{(s+1)(s+2)} (s+1)(s+2)K1(s+1)+K2(s+2)

    對比 F ( s ) = 1 s 2 + 3 s + 2 F(s)=\frac{1}{s^2+3s+2} F(s)=s2+3s+21,可得方程組: K 1 + K 2 = 0 , K 1 + 2 K 2 = 1 K_1+K_2=0, K_1+2K_2=1 K1+K2=0,K1+2K2=1

    進而求得: K 1 = 1 , K 2 = − 1 K_1=1, K_2=-1 K1=1,K2=1

  • 利用 K i = ( s − s i ) ⋅ F ( s ) K_i=(s-s_i)\cdot F(s) Ki=(ssi)F(s)求得

    如求解: F ( s ) = 1 s 2 + 3 s + 2 F(s)=\frac{1}{s^2+3s+2} F(s)=s2+3s+21

    其待定係數的分部分式為: K 1 s + 1 + K 2 s + 2 \frac{K_1}{s+1}+\frac{K_2}{s+2} s+1K1+s+2K2,且 s 1 = − 1 , s 2 = − 2 s_1 = -1, s_2=-2 s1=1,s2=2

    進而求得:

    K 1 = F ( s ) ⋅ ( s − s 1 ) ∣ s = s 1 = [ 1 s 2 + 3 s + 2 ] ⋅ ( s + 1 ) ∣ s = − 1 = 1 K_1 =F(s)\cdot (s-s_1)|_{s=s_1}= [\frac{1}{s^2+3s+2}]\cdot (s+1)|_{s=-1} = 1 K1=F(s)(ss1)s=s1=[s2+3s+21](s+1)s=1=1

    K 2 = F ( s ) ⋅ ( s − s 2 ) ∣ s = s 2 = [ 1 s 2 + 3 s + 2 ] ⋅ ( s + 2 ) ∣ s = − 2 = − 1 K_2 =F(s)\cdot (s-s_2)|_{s=s_2} =[\frac{1}{s^2+3s+2}]\cdot (s+2)|_{s=-2} = -1 K2=F(s)(ss2)s=s2=[s2+3s+21](s+2)s=2=1

  • 利用 K i = N ( s ) D ′ ( s ) ∣ s = s i K_i = \frac{N(s)}{D'(s)}|_{s=s_i} Ki=D(s)N(s)s=si

    如求解: F ( s ) = 1 s 2 + 3 s + 2 F(s)=\frac{1}{s^2+3s+2} F(s)=s2+3s+21

    其待定係數的分部分式為: K 1 s + 1 + K 2 s + 2 \frac{K_1}{s+1}+\frac{K_2}{s+2} s+1K1+s+2K2,且 s 1 = − 1 , s 2 = − 2 s_1 = -1, s_2=-2 s1=1,s2=2

    進而求得:

    K 1 = 1 ( s 2 + 3 s + 2 ) ′ ∣ s = s 1 = 1 2 s + 3 ∣ s = − 1 = 1 K_1 = \frac{1}{(s^2+3s+2)'}|_{s=s_1}=\frac{1}{2s+3}|_{s=-1}=1 K1=(s2+3s+2)1s=s1=2s+31s=1=1

    K 2 = 1 ( s 2 + 3 s + 2 ) ′ ∣ s = s 2 = 1 2 s + 3 ∣ s = − 2 = − 1 K_2 = \frac{1}{(s^2+3s+2)'}|_{s=s_2}=\frac{1}{2s+3}|_{s=-2}=-1 K2=(s2+3s+2)1s=s2=2s+31s=2=1

根據冪函式的拉式變換:
L { e α t ε ( t ) } = 1 s − α ⟺ L − 1 { 1 s − α } = e α t ε ( t ) L\{e^{\alpha t}\varepsilon(t)\}=\frac{1}{s-\alpha} \Longleftrightarrow L^{-1}\{\frac{1}{s-\alpha}\}=e^{\alpha t}\varepsilon(t) L{eαtε(t)}=sα1L1{sα1}=eαtε(t)
因此:
L − 1 { F ( s ) } = L − 1 { ∑ i = 1 n K i s − s i } = L − 1 ∑ i = 1 n { K i s − s i } = ∑ i = 1 n K i e s i t ε ( t ) \begin{aligned} L^{-1}\{F(s)\}&=L^{-1}\{\sum_{i=1}^{n}\frac{K_i}{s-s_i}\} \\&=L^{-1}\sum_{i=1}^{n}\{\frac{K_i}{s-s_i}\} \\&=\sum_{i=1}^{n}K_ie^{s_it}\varepsilon(t) \end{aligned} L1{F(s)}=L1{i=1nssiKi}=L1i=1n{ssiKi}=i=1nKiesitε(t)
舉例,如求解下列拉普拉斯變換的拉普拉斯反變換:
F ( s ) = 1 s 2 + 3 s + 2 ,    收 斂 區 間 為 : R e ( s ) > − 1 F(s)=\frac{1}{s^2+3s+2},\space \space 收斂區間為:Re(s)>-1 F(s)=s2+3s+21,  Re(s)>1
F ( s ) F(s) F(s)根據分母進行因式分解,進而得到待定係數分部分式:
F ( s ) = K 1 s + 1 + K 2 s + 2 F(s)=\frac{K_1}{s+1}+\frac{K_2}{s+2} F(s)=s+1K1+s+2K2
D ( s ) = 0 D(s)=0 D(s)=0的根為:
s 1 = − 1 , s 2 = − 2 s_1 = -1, s_2=-2 s1=1,s2=2
係數 K i K_i Ki為:
K 1 = 1 , K 2 = − 1 K_1 = 1, K_2=-1 K1=1,K2=1
求出求解分部分式的拉普拉斯的反變換:
L − 1 { 1 s + 1 } ⟺ e − t ε ( t ) L − 1 { − 1 s + 2 } ⟺ − e − 2 t ε ( t ) \begin{aligned} L^{-1}\{\frac{1}{s+1}\} \Longleftrightarrow e^{-t}\varepsilon(t) \\L^{-1}\{\frac{-1}{s+2}\} \Longleftrightarrow -e^{-2t}\varepsilon(t) \end{aligned} L1{s+11}etε(t)L1{s+21}e2tε(t)
因此,最終的 F ( s ) F(s) F(s)的拉普拉斯反變換為:
f ( t ) = e − t ε ( t ) − e − 2 t ε ( t ) f(t) =e^{-t}\varepsilon(t) -e^{-2t}\varepsilon(t) f(t)=etε(t)e2tε(t)

1.1.2 沒有重根,且根存在共軛復根時

具有待定係數的分部分式展開後,當 D ( s ) = 0 D(s)=0 D(s)=0的根具有共軛復根時:
F ( s ) = K 1 s − s 1 + K 2 s − s 2 + ⋯ + K n s − s n F(s)=\frac{K_1}{s-s_1}+\frac{K_2}{s-s_2}+\cdots+\frac{K_n}{s-s_n} F(s)=ss1K1+ss2K2++ssnKn
s 1 s_1 s1 s 2 s_2 s2為一對共軛復根, s 3 ⋯ s n s_3 \cdots s_n s3sn為實數根,即:
s 1 = a 1 + j ω s 2 = a 1 − j ω s 3 = a 3 ⋅ ⋅ ⋅ s n = a n a 1 ⋯ a n 為 實 數 , a 1 = a 2 \begin{aligned} s_1&=a_1+j\omega \\s_2&=a_1-j\omega \\s_3&=a_3 \\\cdot \\\cdot \\\cdot \\s_n&=a_n \\a_1 &\cdots a_n為實數, a_1=a_2 \end{aligned} s1s2s3sna1=a1+jω=a1jω=a3=anan,a1=a2
則係數為:
K 1 = N ( s ) D ′ ( s ) ∣ s = s 1 = N ( s ) D ′ ( s ) ∣ s = a 1 + j ω K 2 = N ( s ) D ′ ( s ) ∣ s = s 2 = N ( s ) D ′ ( s ) ∣ s = a 1 − j ω K 3 = N ( s ) D ′ ( s ) ∣ s = s 3 = N ( s ) D ′ ( s ) ∣ s = a 3 ⋅ ⋅ ⋅ K n = N ( s ) D ′ ( s ) ∣ s = s n = N ( s ) D ′ ( s ) ∣ s = a n \begin{aligned} K_1 = \frac{N(s)}{D'(s)}|_{s=s_{1}}&= \frac{N(s)}{D'(s)}|_{s=a_1+j\omega} \\K_2 = \frac{N(s)}{D'(s)}|_{s=s_{2}}&= \frac{N(s)}{D'(s)}|_{s=a_1-j\omega} \\K_3 = \frac{N(s)}{D'(s)}|_{s=s_{3}}&=\frac{N(s)}{D'(s)}|_{s=a_3} \\\cdot \\\cdot \\\cdot \\K_n = \frac{N(s)}{D'(s)}|_{s=s_{n}}&=\frac{N(s)}{D'(s)}|_{s=a_n} \end{aligned} K1=D(s)N(s)s=s1K2=D(s)N(s)s=s2K3=D(s)N(s)s=s3Kn=D(s)N(s)s=sn=D(s)N(s)s=a1+jω=D(s)N(s)s=a1jω=D(s)N(s)s=a3=D(s)N(s)s=an
舉例:求解下列拉普拉斯變換的反變換:
F ( s ) = s + 3 s 2 + 2 s + 5 F(s)=\frac{s+3}{s^2+2s+5} F(s)=s2+2s+5s+3
D ( s ) = 0 D(s)=0 D(s)=0的根為:
s 2 + 2 s + 5 = 0 ⇒ ( s + 1 ) 2 + 4 = 0 ⇒ s 1 = − 1 + j 2 , s 2 = − 1 − j 2 \begin{aligned} &s^2+2s+5=0 \\\Rightarrow &(s+1)^2+4=0 \\\Rightarrow &s_1=-1+j2, s_2=-1-j2 \end{aligned} s2+2s+5=0(s+1)2+4=0s1=1+j2,s2=1j2
因此,其對應的係數為:
K 1 = N ( s ) D ′ ( s ) ∣ s = s 1 = s + 3 2 s + 2 ∣ s = − 1 + j 2 = 0.5 − j 0.5 = 0.5 2 e − j 4 5 ∘ K 2 = N ( s ) D ′ ( s ) ∣ s = s 2 = s + 3 2 s + 2 ∣ s = − 1 − j 2 = 0.5 + j 0.5 = 0.5 2 e j 4 5 ∘ \begin{aligned} K_1=\frac{N(s)}{D'(s)}|_{s=s_{1}}=\frac{s+3}{2s+2}|_{s=-1+j2}=0.5-j0.5=0.5\sqrt{2}e^{-j45^\circ} \\K_2=\frac{N(s)}{D'(s)}|_{s=s_{2}}=\frac{s+3}{2s+2}|_{s=-1-j2}=0.5+j0.5=0.5\sqrt{2}e^{j45^\circ} \end{aligned} K1=D(s)N(s)s=s1=2s+2s+3s=1+j2=0.5j0.5=0.52 ej45K2=D(s)N(s)s=s2=2s+2s+3s=1j2=0.5+j0.5=0.52 ej45
因此係數也是共軛複數,並且係數引入了一個相角,即 K 2 = K 1 ∗ K_2=K_1^* K2=K1,則分部分式展開為:
F ( s ) = 0.5 − j 0.5 s − ( − 1 + j 2 ) + 0.5 + j 0.5 s − ( − 1 − j 2 ) = 0.5 − j 0.5 s − ( − 1 + j 2 ) + 0.5 + j 0.5 s − ( − 1 − j 2 ) \begin{aligned} F(s)&=\frac{0.5-j0.5}{s-(-1+j2)}+\frac{0.5+j0.5}{s-(-1-j2)} \\&=\frac{0.5-j0.5}{s-(-1+j2)}+\frac{0.5+j0.5}{s-(-1-j2)} \end{aligned} F(s)=s(1+j2)0.5j0.5+s(1j2)0.5+j0.5=s(1+j2)0.5j0.5+s(1j2)0.5+j0.5
因此, F ( s ) F(s) F(s)的拉普拉斯反變換為:
L − 1 { F ( s ) } = L − 1 { 0.5 − j 0.5 s − ( − 1 + j 2 ) + 0.5 + j 0.5 s − ( − 1 − j 2 ) } = ( 0.5 − j 0.5 ) e ( − 1 + j 2 ) t ε ( t ) + ( 0.5 + j 0.5 ) e ( − 1 − j 2 ) t ε ( t ) = ( 0.5 − j 0.5 ) e ( − 1 + j 2 ) t ε ( t ) + ( 0.5 + j 0.5 ) e ( − 1 − j 2 ) t ε ( t ) = 0.5 2 e − j 4 5 ∘ ⋅ e ( − 1 + j 2 ) t ε ( t ) + 0.5 2 e j 4 5 ∘ ⋅ e ( − 1 − j 2 ) t ε ( t ) = 0.5 2 e − t ⋅ [ e ( j 2 t − j 4 5 ∘ ) + e ( j 2 t + j 4 5 ∘ ) ] 利 用 歐 拉 公 式 : ⇒ 2 e − t ⋅ c o s ( 2 t + 4 5 ∘ ) \begin{aligned} L^{-1}\{F(s)\}&=L^{-1}\{\frac{0.5-j0.5}{s-(-1+j2)}+\frac{0.5+j0.5}{s-(-1-j2)}\} \\&=(0.5-j0.5)e^{(-1+j2)t}\varepsilon(t)+(0.5+j0.5)e^{(-1-j2)t}\varepsilon(t) \\&=(0.5-j0.5)e^{(-1+j2)t}\varepsilon(t)+(0.5+j0.5)e^{(-1-j2)t}\varepsilon(t) \\&=0.5\sqrt{2}e^{-j45^\circ}\cdot e^{(-1+j2)t}\varepsilon(t) + 0.5\sqrt{2}e^{j45^\circ} \cdot e^{(-1-j2)t}\varepsilon(t) \\&=0.5\sqrt{2}e^{-t}\cdot [e^{(j2t-j45^{\circ})}+e^{(j2t+j45^{\circ})}] \\利用尤拉公式:&\Rightarrow \sqrt{2}e^{-t}\cdot cos(2t+45^{\circ}) \end{aligned} L1{F(s)}=L1{s(1+j2)0.5j0.5+s(1j2)0.5+j0.5}=(0.5j0.5)e(1+j2)tε(t)+(0.5+j0.5)e(1j2)tε(t)=(0.5j0.5)e(1+j2)tε(t)+(0.5+j0.5)e(1j2)tε(t)=0.52 ej45e(1+j2)tε(t)+0.52 ej45e(1j2)tε(t)=0.52 et[e(j2tj45)+e(j2t+j45)]2 etcos(2t+45)
通過上式可知:

D ( s ) = 0 D(s)=0 D(s)=0有一對共軛復根 s 1 , s 2 s_{1}, s_{2} s1,s2,且 s 2 = s 1 ∗ s_{2}=s^*_{1} s2=s1,則與之相關的係數 K 1 , K 2 K_1, K_2 K1,K2也為複數,且也為共軛關係,即
K 2 = N ( s ) D ′ ( s ) ∣ s = s 2 = N ( s ) D ′ ( s ) ∣ s = s 1 ∗ = [ N ( s ) D ′ ( s ) ] ∗ ∣ s = s 1 = K 1 ∗ K_{2} =\frac{N(s)}{D'(s)}|_{s=s_{2}}=\frac{N(s)}{D'(s)}|_{s=s^*_{1}}=[\frac{N(s)}{D'(s)}]^*|_{s=s_{1}} = K_1^* K2=D(s)N(s)s=s2=D(s)N(s)s=s1=[D(s)N(s)]s=s1=K1
因此帶有一對復根的拉普拉斯變換 F ( s ) F(s) F(s)為:
F ( s ) = K 1 s − s 1 + K 2 s − s 2 = K 1 s − s 1 + K 1 ∗ s − s 1 ∗ F(s) = \frac{K_1}{s-s_1}+\frac{K_2}{s-s_2}= \frac{K_1}{s-s_1}+\frac{K_1^*}{s-s_1^*} F(s)=ss1K1+ss2K2=ss1K1+ss1K1
其拉普拉斯反變換為:
K 1 e s 1 t ε ( t ) + K 2 e s 2 t ε ( t ) = [ K 1 e s 1 t + K 1 ∗ e s 1 ∗ t ] ε ( t ) = [ ∣ K 1 ∣ e j ϕ 1 e ( σ 1 + j ω 0 ) t + ∣ K 1 ∣ e − j ϕ 1 e ( σ 1 − j ω 0 ) t ] ε ( t ) = [ e j ω 0 t + ϕ 1 + e − j ω 0 t + ϕ 1 ] ⋅ ∣ K 1 ∣ e σ 1 t ε ( t ) = 2 ∣ K 1 ∣ e σ 1 t cos ⁡ ( ω 0 t + ϕ 1 ) ε ( t ) \begin{aligned} K_{1} e^{s_{1} t} \varepsilon(t)+K_{2} e^{s_{2} t} \varepsilon(t) &=\left[K_{1} e^{s_{1} t}+K_{1}^{*} e^{s_{1}^{*} t}\right] \varepsilon(t) \\&={\left[\left|K_{1}\right| e^{j \phi_{1}} e^{\left(\sigma_{1}+j \omega_{0}\right) t}+\left|K_{1}\right| e^{-j \phi_{1}} e^{\left(\sigma_{1}-j \omega_{0}\right) t}\right] \varepsilon(t)} \\&=\left[e^{j \omega_{0} t+\phi_{1}}+e^{-j \omega_{0} t+\phi_{1}}\right] \cdot\left|K_{1}\right| e^{\sigma_{1} t} \varepsilon(t) \\&=2\left|K_{1}\right| e^{\sigma_{1} t} \cos \left(\omega_{0} t+\phi_{1}\right) \varepsilon(t) \end{aligned} K1es1tε(t)+K2es2tε(t)=[K1es1t+K1es1t]ε(t)=[K1ejϕ1e(σ1+jω0)t+K1ejϕ1e(σ1jω0)t]ε(t)=[ejω0t+ϕ1+ejω0t+ϕ1]K1eσ1tε(t)=2K1eσ1tcos(ω0t+ϕ1)ε(t)
上式中, ϕ 1 \phi_1 ϕ1為複數 K 1 K_1 K1的相角,由上述推導可知,最終,當存在一對共軛復根時,其拉普拉斯反變換仍然是一個實數訊號,因此其計算方式與實數根計算方式可以統一考慮

因此:在 **當 m < n m<n m<n,且 D ( s ) = 0 D(s)=0 D(s)=0沒有重根(包括共軛復根)**條件下,最終反變換的結果一定是一個實數,在計算時,要最終化簡到沒有虛數為止

1.2 m < n m<n m<n,且 D ( s ) = 0 D(s)=0 D(s)=0有重根時

D ( s ) = 0 D(s)=0 D(s)=0的重根為 s 1 s_1 s1的p次重根,則多項式分母 D ( s ) D(s) D(s)可通過因式分解表示為:
D ( s ) = ( s − s 1 ) p ( s − s 2 ) ⋯ ( s − s n ) D(s)=(s-s_1)^p(s-s_2)\cdots(s-s_n) D(s)=(ss1)p(ss2)(ssn)
因此, F ( s ) F(s) F(s)表示為:
F ( s ) = K 1 p ( s − s 1 ) p + K 1 ( p − 1 ) ( s − s 1 ) p − 1 + ⋯ + K 12 ( s − s 1 ) 2 + K 11 ( s − s 1 ) 1 + ∑ i = 2 n K i s − s i \begin{aligned} F(s)&=\frac{K_{1p}}{(s-s_1)^{p}}+\frac{K_{1(p-1)}}{(s-s_1)^{p-1}}+\cdots+\frac{K_{12}}{(s-s_1)^{2}}+\frac{K_{11}}{(s-s_1)^{1}} +\sum_{i=2}^{n}\frac{K_i}{s-s_i} \end{aligned} F(s)=(ss1)pK1p+(ss1)p1K1(p1)++(ss1)2K12+(ss1)1K11+i=2nssiKi
上式中:

  • K 1 p ( s − s 1 ) p + K 1 ( p − 1 ) ( s − s 1 ) p − 1 + ⋯ + K 12 ( s − s 1 ) 2 + K 11 ( s − s 1 ) 1 \frac{K_{1p}}{(s-s_1)^{p}}+\frac{K_{1(p-1)}}{(s-s_1)^{p-1}}+\cdots+\frac{K_{12}}{(s-s_1)^{2}}+\frac{K_{11}}{(s-s_1)^{1}} (ss1)pK1p+(ss1)p1K1(p1)++(ss1)2K12+(ss1)1K11 p p p階重根 s 1 s_1 s1對應的分部分式展開;
  • K 1 p K_{1p} K1p K 1 ( p − 1 ) K_{1(p-1)} K1(p1) K 1 ( p − 1 ) K_{1(p-1)} K1(p1) ⋯ \cdots K 11 K_{11} K11為根 s 1 s_1 s1 p p p階重根對應分部展開式的係數;
  • ∑ i = 2 n K i s − s i \sum_{i=2}^{n}\frac{K_i}{s-s_i} i=2nssiKi、代表非重根 s i s_i si對應的分部分式展開;
  • K i , i ≥ 2 K_i, i\geq 2 Ki,i2代表非重根 s i s_i si對應分部展開式的係數。

其中,非重根項的係數可根據1.1中的方法求得,而重根系數可通過下述方法求得:
K 1 p = [ ( s − s 1 ) p ⋅ F ( s ) ] ∣ s = s 1 K 1 ( p − 1 ) = d d s [ ( s − s 1 ) p ⋅ F ( s ) ] ∣ s = s 1 K 1 ( p − 2 ) = 1 2 ! d 2 d s 2 [ ( s − s 1 ) p ⋅ F ( s ) ] ∣ s = s 1 ⋅ ⋅ ⋅ K 11 = 1 ( p − 1 ) ! d p − 1 d s p − 1 [ ( s − s 1 ) p ⋅ F ( s ) ] ∣ s = s 1 \begin{aligned} K_{1p}&=[(s-s_1)^p\cdot F(s)]|_{s=s_1} \\K_{1(p-1)}&=\frac{d}{ds}[(s-s_1)^p\cdot F(s)]|_{s=s_1} \\K_{1(p-2)}&=\frac{1}{2!}\frac{d^2}{ds^2}[(s-s_1)^p\cdot F(s)]|_{s=s_1} \\\cdot \\\cdot \\\cdot \\K_{11}&=\frac{1}{(p-1)!}\frac{d^{p-1}}{ds^{p-1}}[(s-s_1)^p\cdot F(s)]|_{s=s_1} \end{aligned} K1pK1(p1)K1(p2)K11=[(ss1)pF(s)]s=s1=dsd[(ss1)pF(s)]s=s1=2!1ds2d2[(ss1)pF(s)]s=s1=(p1)!1dsp1dp1[(ss1)pF(s)]s=s1
為方便表示通式,將 s 1 s_1 s1重根項的第 k k k階對應的分式的係數記為 K 1 k K_{1k} K1k,則係數 K 1 k K_{1k} K1k的通式可以表示為
K 1 k = 1 ( p − k ) ! d p − k d s p − k [ ( s − s 1 ) p ⋅ F ( s ) ] ∣ s = s 1 K_{1k}=\frac{1}{(p-k)!}\frac{d^{p-k}}{ds^{p-k}}[(s-s_1)^p\cdot F(s)]|_{s=s_1} K1k=(pk)!1dspkdpk[(ss1)pF(s)]s=s1
如:

K 11 K_{11} K11 ( s − s 1 ) (s-s_1) (ss1)的一階分式 1 ( s − s 1 ) 1 \frac{1}{(s-s_1)^1} (ss1)11的係數,則 K 11 = 1 ( p − 1 ) ! d p − 1 d s p − 1 [ ( s − s 1 ) p ⋅ F ( s ) ] ∣ s = s 1 K_{11}=\frac{1}{(p-1)!}\frac{d^{p-1}}{ds^{p-1}}[(s-s_1)^p\cdot F(s)]|_{s=s_1} K11=(p1)!1dsp1dp1[(ss1)pF(s)]s=s1

K 12 K_{12} K12 ( s − s 1 ) (s-s_1) (ss1)的二階分式 1 ( s − s 1 ) 2 \frac{1}{(s-s_1)^2} (ss1)21的係數,則 K 12 = 1 ( p − 2 ) ! d p − 2 d s p − 2 [ ( s − s 1 ) p ⋅ F ( s ) ] ∣ s = s 1 K_{12}=\frac{1}{(p-2)!}\frac{d^{p-2}}{ds^{p-2}}[(s-s_1)^p\cdot F(s)]|_{s=s_1} K12=(p2)!1dsp2dp2[(ss1)pF(s)]s=s1

K 1 p K_{1p} K1p ( s − s 1 ) (s-s_1) (ss1)的p階分式 1 ( s − s 1 ) p \frac{1}{(s-s_1)^p} (ss1)p1的係數,則 K 1 p = [ ( s − s 1 ) p ⋅ F ( s ) ] ∣ s = s 1 K_{1p}=[(s-s_1)^p\cdot F(s)]|_{s=s_1} K1p=[(ss1)pF(s)]s=s1

將係數求出後,即可得到拉普拉斯變換的分部分解展開形式:
F ( s ) = K 1 p ( s − s 1 ) p + K 1 ( p − 1 ) ( s − s 1 ) p − 1 + ⋯ + K 12 ( s − s 1 ) 2 + K 11 ( s − s 1 ) 1 + ∑ i = 2 n K i s − s i \begin{aligned} F(s)&=\frac{K_{1p}}{(s-s_1)^{p}}+\frac{K_{1(p-1)}}{(s-s_1)^{p-1}}+\cdots+\frac{K_{12}}{(s-s_1)^{2}}+\frac{K_{11}}{(s-s_1)^{1}} +\sum_{i=2}^{n}\frac{K_i}{s-s_i} \end{aligned} F(s)=(ss1)pK1p+(ss1)p1K1(p1)++(ss1)2K12+(ss1)1K11+i=2nssiKi
上式中,第 k k k重根項的拉普拉斯反變換可通過下列關係求得:
L − 1 { K 1 k ( s − s 1 ) k } = K 1 k ( k − 1 ) ! ⋅ t k − 1 e s 1 t ε ( t ) L^{-1}\{\frac{K_{1k}}{(s-s_1)^k}\}=\frac{K_{1k}}{(k-1)!}\cdot t^{k-1}e^{s_1 t}\varepsilon(t) L1{(ss1)kK1k}=(k1)!K1ktk1es1tε(t)
如:
L − 1 { K 1 p ( s − s 1 ) p } = K 1 p ( p − 1 ) ! ⋅ t p − 1 e s 1 t ε ( t ) L − 1 { K 1 ( p − 1 ) ( s − s 1 ) p − 1 } = K 1 ( p − 1 ) ( p − 2 ) ! ⋅ t p − 2 e s 1 t ε ( t ) ⋅ ⋅ ⋅ L − 1 { K 11 ( s − s 1 ) 1 } = e s 1 t ε ( t ) \begin{aligned} L^{-1}\{\frac{K_{1p}}{(s-s_1)^p}\}=&\frac{K_{1p}}{(p-1)!}\cdot t^{p-1}e^{s_1 t}\varepsilon(t) \\L^{-1}\{\frac{K_{1(p-1)}}{(s-s_1)^{p-1}}\}=&\frac{K_{1(p-1)}}{(p-2)!}\cdot t^{p-2}e^{s_1 t}\varepsilon(t) \\&\cdot \\&\cdot \\&\cdot \\L^{-1}\{\frac{K_{11}}{(s-s_1)^{1}}\}=&e^{s_1 t}\varepsilon(t) \end{aligned} L1{(ss1)pK1p}=L1{(ss1)p1K1(p1)}=L1{(ss1)1K11}=(p1)!K1ptp1es1tε(t)(p2)!K1(p1)tp2es1tε(t)es1tε(t)

舉例:求解下列拉普拉斯變換的原函式:
F ( s ) = s + 2 s ( s + 3 ) ( s + 1 ) 2 F(s)=\frac{s+2}{s(s+3)(s+1)^2} F(s)=s(s+3)(s+1)2s+2
首先令 D ( s ) = s ( s + 3 ) ( s + 1 ) 2 = 0 D(s)=s(s+3)(s+1)^2=0 D(s)=s(s+3)(s+1)2=0,則 D ( s ) = 0 D(s)=0 D(s)=0具有四個根,包括兩個單根 s 1 = 0 , s 2 = − 3 s_1=0, s_2=-3 s1=0,s2=3,以及兩個重根 s 3 = s 4 = − 1 s_3=s_4=-1 s3=s4=1

因此部分分式展開為:
F ( s ) = s + 2 s ( s + 3 ) ( s + 1 ) 2 = K 1 s − 0 + K 2 s − ( − 3 ) + [ K 32 ( s + 1 ) 2 + K 31 ( s + 1 ) 1 ] F(s)=\frac{s+2}{s(s+3)(s+1)^2}=\frac{K_1}{s-0}+\frac{K_2}{s-(-3)}+[\frac{K_{32}}{(s+1)^2}+\frac{K_{31}}{(s+1)^1}] F(s)=s(s+3)(s+1)2s+2=s0K1+s(3)K2+[(s+1)2K32+(s+1)1K31]
根據之前所述的求解係數的方法,
K 1 = [ s N ( s ) D ( s ) ] s = 0 = [ s + 2 ( s + 3 ) ( s + 1 ) 2 ] s = 0 = 2 3 K 2 = [ ( s + 3 ) N ( s ) D ( s ) ] s = − 3 = [ s + 2 s ( s + 1 ) 2 ] s = − 3 = 1 12 K 32 = [ ( s + 1 ) 2 N ( s ) D ( s ) ] s = − 1 = [ s + 2 s ( s + 3 ) ] s = − 1 = − 1 2 K 31 = d d s [ ( s + 1 ) 2 N ( s ) D ( s ) ] s = − 1 = [ s ( s + 3 ) − ( s + 2 ) ( 2 s + 3 ) s 2 ( s + 3 ) 2 ] s = − 1 = − 3 4 \begin{aligned} K_{1}&=\left[s \frac{N(s)}{D(s)}\right]_{s=0}&=&\left[\frac{s+2}{(s+3)(s+1)^{2}}\right]_{s=0}&=\frac{2}{3} \\K_{2}&=\left[(s+3) \frac{N(s)}{D(s)}\right]_{s=-3}&=&\left[\frac{s+2}{s(s+1)^{2}}\right]_{s=-3}&=\frac{1}{12} \\K_{32}&=\left[(s+1)^{2} \frac{N(s)}{D(s)}\right]_{s=-1}&=&\left[\frac{s+2}{s(s+3)}\right]_{s=-1}&=-\frac{1}{2} \\K_{31}&=\frac{d}{ds}\left[(s+1)^{2} \frac{N(s)}{D(s)}\right]_{s=-1}&=&\left[\frac{s(s+3)-(s+2)(2s+3_)}{s^2(s+3)^2}\right]_{s=-1}&=-\frac{3}{4} \end{aligned} K1K2K32K31=[sD(s)N(s)]s=0=[(s+3)D(s)N(s)]s=3=[(s+1)2D(s)N(s)]s=1=dsd[(s+1)2D(s)N(s)]s=1====[(s+3)(s+1)2s+2]s=0[s(s+1)2s+2]s=3[s(s+3)s+2]s=1[s2(s+3)2s(s+3)(s+2)(2s+3)]s=1=32=121=21=43
因此完成的分部分式為:
F ( s ) = 2 3 s + 1 12 ( s + 3 ) + [ − 1 2 ( s + 1 ) 2 − 3 14 ( s + 1 ) ] F(s)=\frac{2}{3s}+\frac{1}{12(s+3)}+[-\frac{1}{2(s+1)^2}-\frac{3}{14(s+1)}] F(s)=3s2+12(s+3)1+[2(s+1)2114(s+1)3]
所以原函式為:
L 1 { F ( s ) } = [ 2 3 + 1 12 e − 3 t − 1 2 ( t + 3 2 e − t ) ] ⋅ ε ( t ) L_1\{{F(s)}\}=[\frac{2}{3}+\frac{1}{12}e^{-3t}-\frac{1}{2}(t+\frac{3}{2}e^{-t})]\cdot \varepsilon(t) L1{F(s)}=[32+121e3t21(t+23et)]ε(t)

1.2 m ≥ n m\geq n mn

m ≥ n m\geq n mn時,則分子的階數高於分子的階數,因此需要將 F ( s ) F(s) F(s)通過長除法,化簡為一個真分式,和一個多項式的和,即:
F ( s ) = N ( s ) D ( s ) = M ( s ) + N 1 ( s ) D ( s ) F(s)=\frac{N(s)}{D(s)}=M(s)+\frac{N_1(s)}{D(s)} F(s)=D(s)N(s)=M(s)+D(s)N1(s)

  • 其中 M ( s ) M(s) M(s)為多項式,可以利用 L − 1 { s n } = δ n ( t ) L^{-1}\{s^n\}=\delta^n(t) L1{sn}=δn(t)求解。
  • N 1 ( s ) D ( s ) \frac{N_1(s)}{D(s)} D(s)N1(s)為真分式,根據這個真分式的分母和分子的階數關係以及根的情況,可以根據1.1和1.2中的原則和方法求解原函式,

舉例:求解下列拉普拉斯變換的反變換
F ( s ) = s 3 s 2 + 3 s + 2 F(s)=\frac{s^3}{s^2+3s+2} F(s)=s2+3s+2s3
通過長除法,上式可以化簡為:
F ( s ) = s − 3 + 7 s + 6 s 2 + 3 s + 2 F(s)=s-3+\frac{7s+6}{s^2+3s+2} F(s)=s3+s2+3s+27s+6
再對 7 s + 6 s 2 + 3 s + 2 \frac{7s+6}{s^2+3s+2} s2+3s+27s+6進行分部分式展開:
7 s + 6 s 2 + 3 s + 2 = − 1 s + 1 + 8 s + 2 \frac{7s+6}{s^2+3s+2} = \frac{-1}{s+1}+\frac{8}{s+2} s2+3s+27s+6=s+11+s+28
分別對上式中的多項式和真分式求解拉式反變換:
L − 1 { s } = δ ′ ( t ) L − 1 { − 3 } = − 3 δ ( t ) L − 1 { 7 s + 6 s 2 + 3 s + 2 } = − e − t ε ( t ) + 8 e − 2 t ε ( t ) \begin{aligned} &L^{-1}\{s\}&=&\delta'(t)& \\&L^{-1}\{-3\}&=&-3\delta(t)& \\&L^{-1}\{\frac{7s+6}{s^2+3s+2}\}&=&-e^{-t}\varepsilon(t) +8e^{-2t}\varepsilon(t)& \end{aligned} L1{s}L1{3}L1{s2+3s+27s+6}===δ(t)3δ(t)etε(t)+8e2tε(t)
因此最終的拉式反變換為:
f ( t ) = δ ′ ( t ) − 3 δ ( t ) − e − t ε ( t ) + 8 e − 2 t ε ( t ) f(t)=\delta'(t)-3\delta(t)-e^{-t}\varepsilon(t) +8e^{-2t}\varepsilon(t) f(t)=δ(t)3δ(t)etε(t)+8e2tε(t)

謝謝閱讀,若有不當之處歡迎批評指正!

相關文章