1 定义
- 取样处理就是对连续信号的离散化处理
p ( t ) p(t) p(t) 是开关函数
f s ( t ) = f ( t ) ⋅ p ( t ) f_s(t)=f(t)\cdot p(t) fs(t)=f(t)⋅p(t)
- 当
p
(
t
)
p(t)
p(t) 为周期矩形函数时
该取样为均匀抽样,周期为 T s T_s Ts,则取样角频率为: w s = 2 π f s ( t ) = 2 π T s w_s=2\pi f_s(t)=\frac{2\pi}{T_s} ws=2πfs(t)=Ts2π
p ( t ) ↔ P ( j w ) = 2 π ∑ n = − ∞ + ∞ F n δ ( w − n w s ) p(t)\leftrightarrow P(jw)=2\pi\sum_{n=-\infty}^{+\infty}F_n\delta(w-nw_s) p(t)↔P(jw)=2πn=−∞∑+∞Fnδ(w−nws)
其中:
F
n
=
1
T
s
∫
−
T
s
2
T
s
2
p
(
t
)
e
−
j
n
w
s
t
d
t
=
τ
S
a
(
n
w
s
τ
2
)
T
s
F_n=\frac{1}{T_s}\int_{-\frac{T_s}{2}}^{\frac{T_s}{2}}p(t)e^{-jnw_st}dt=\frac{\tau Sa(\frac{nw_s\tau}{2})}{T_s}
Fn=Ts1∫−2Ts2Tsp(t)e−jnwstdt=TsτSa(2nwsτ)
f
s
(
t
)
=
f
(
t
)
⋅
p
(
t
)
↔
F
s
(
j
w
)
=
F
(
j
w
)
∗
P
(
j
w
)
2
π
f_s(t)=f(t)\cdot p(t)\leftrightarrow F_s(jw)=\frac{F(jw)*P(jw)}{2\pi}
fs(t)=f(t)⋅p(t)↔Fs(jw)=2πF(jw)∗P(jw)
F
s
(
j
w
)
=
1
2
π
F
(
j
w
)
∗
2
π
∑
n
=
−
∞
+
∞
F
n
δ
(
w
−
n
w
s
)
=
F
(
j
w
)
∗
∑
n
=
−
∞
+
∞
τ
S
a
(
n
w
s
τ
2
)
T
s
δ
(
w
−
n
w
s
)
F_s(jw)=\frac{1}{2\pi}F(jw)*2\pi\sum_{n=-\infty}^{+\infty}F_n\delta(w-nw_s)=F(jw)*\sum_{n=-\infty}^{+\infty}\frac{\tau Sa(\frac{nw_s\tau}{2})}{T_s}\delta(w-nw_s)
Fs(jw)=2π1F(jw)∗2πn=−∞∑+∞Fnδ(w−nws)=F(jw)∗n=−∞∑+∞TsτSa(2nwsτ)δ(w−nws)
根据上一篇推导的结论 f ( t ) ∗ δ ( t ) = f ( t ) f(t)*\delta(t)=f(t) f(t)∗δ(t)=f(t) 得 F ( j w ) ∗ δ ( w − n w s ) = F [ j ( w − n w s ) ] F(jw)*\delta(w-nw_s)=F[j(w-nw_s)] F(jw)∗δ(w−nws)=F[j(w−nws)]:
F s ( j w ) = τ T s ∑ n = − ∞ + ∞ S a ( n w s τ 2 ) ⋅ F [ j ( w − n w s ) ] F_s(jw)=\frac{\tau}{T_s}\sum_{n=-\infty}^{+\infty}Sa(\frac{nw_s\tau}{2})\cdot F[j(w-nw_s)] Fs(jw)=Tsτn=−∞∑+∞Sa(2nwsτ)⋅F[j(w−nws)]
处理过后统一为:
F s ( j w ) = ∑ n = − ∞ + ∞ F n ⋅ F [ j ( w − n w s ) ] F_s(jw)=\sum_{n=-\infty}^{+\infty}F_n\cdot F[j(w-nw_s)] Fs(jw)=n=−∞∑+∞Fn⋅F[j(w−nws)]
- 结论:
①【 时域的抽样就是频域的周期延拓 】信号在时域被抽样后,它的频谱函数 F s ( j w ) F_s(jw) Fs(jw) 是连续信号的频谱 F ( j w ) F(jw) F(jw) 以取样角频率 w s w_s ws 为间隔,周期地重复而得到的
②在重复过程后得到的各离散幅度被取样脉冲 p ( t ) p(t) p(t) 的傅立叶系数 F n F_n Fn 所加权
2 奈奎斯特取样频率
不能让第
k
k
k 与第
k
+
1
k+1
k+1 个有重叠,所以
(
w
s
−
w
m
)
≥
w
m
(w_s-w_m)\geq w_m
(ws−wm)≥wm,则
w
s
≥
2
w
m
w_s\geq 2w_m
ws≥2wm ,
T
s
≤
1
2
f
m
T_s\leq \frac{1}{2f_m}
Ts≤2fm1
【取样率不得小于信号频谱最高频率的两倍】
-
时域取样定理
如果 f ( t ) f(t) f(t) 是带宽有限的连续信号,其频谱 F ( j w ) F(jw) F(jw) 的最高频率为 f m f_m fm ,则以取样间隔 T s ≤ 1 2 f m T_s\leq \frac{1}{2f_m} Ts≤2fm1 对信号 f ( t ) f(t) f(t) 进行等间隔取样,所得的取样信号 f s ( t ) f_s(t) fs(t) 将包含原信号 f ( t ) f(t) f(t) 的全部信息,因而可以从 f s ( t ) f_s(t) fs(t) 完全恢复出原信号 f ( t ) f(t) f(t) -
当 T s = 1 2 f m T_s=\frac{1}{2f_m} Ts=2fm1 为奈奎斯特取样间隔, f s = 2 f m f_s=2f_m fs=2fm 为奈奎斯特取样频率, w s = 2 w m w_s=2w_m ws=2wm 为奈奎斯特取样角频率
例、求下列信号的奈奎斯特取样角频率
- (1)
S
a
(
90
t
)
Sa(90t)
Sa(90t)
解:
{ g τ ( t ) ↔ τ S a ( w τ 2 ) τ S a ( t τ 2 ) ↔ 2 π ⋅ g τ ( w ) \begin{cases} g_\tau(t)\leftrightarrow \tau Sa(\frac{w\tau}{2})\\ \tau Sa(\frac{t\tau}{2})\leftrightarrow 2\pi\cdot g_\tau(w)\\ \end{cases} {gτ(t)↔τSa(2wτ)τSa(2tτ)↔2π⋅gτ(w)
其中:
τ
2
=
90
\frac{\tau}{2}=90
2τ=90,则
τ
=
180
\tau=180
τ=180,则
180
S
a
(
90
t
)
↔
2
π
⋅
g
180
(
w
)
180Sa(90t)\leftrightarrow 2\pi\cdot g_{180}(w)
180Sa(90t)↔2π⋅g180(w),则
S
a
(
90
t
)
↔
π
90
⋅
g
180
(
w
)
Sa(90t)\leftrightarrow \frac{\pi}{90}\cdot g_{180}(w)
Sa(90t)↔90π⋅g180(w)
则
w
m
=
90
w_m=90
wm=90,奈奎斯特取样角频率为:
w s = 180 w_s=180 ws=180
- (2)
S
a
2
(
90
t
)
Sa^2(90t)
Sa2(90t)
解:
由 S a 2 ( 90 t ) = S a ( 90 t ) ⋅ S a ( 90 t ) Sa^2(90t)=Sa(90t)\cdot Sa(90t) Sa2(90t)=Sa(90t)⋅Sa(90t),根据频域卷积性质 f 1 ( t ) ⋅ f 2 ( t ) ↔ 1 2 π F 1 ( j w ) ∗ F 2 ( j w ) f_1(t)\cdot f_2(t)\leftrightarrow \frac{1}{2\pi}F_1(jw)*F_2(jw) f1(t)⋅f2(t)↔2π1F1(jw)∗F2(jw),则:
S
a
2
(
90
t
)
↔
1
2
π
π
90
g
180
(
w
)
∗
π
90
g
180
(
w
)
=
∫
−
90
w
+
90
(
π
90
)
2
d
τ
+
∫
w
−
90
90
(
π
90
)
2
d
τ
Sa^2(90t)\leftrightarrow\frac{1}{2\pi}\frac{\pi}{90}g_{180}(w)*\frac{\pi}{90}g_{180}(w)=\int_{-90}^{w+90}(\frac{\pi}{90})^2d\tau+\int_{w-90}^{90}(\frac{\pi}{90})^2d\tau
Sa2(90t)↔2π190πg180(w)∗90πg180(w)=∫−90w+90(90π)2dτ+∫w−9090(90π)2dτ
π
90
g
180
(
w
)
∗
π
90
g
180
(
w
)
=
(
π
90
)
2
{
(
w
+
180
)
[
u
(
w
+
180
)
−
u
(
w
)
]
+
(
180
−
w
)
[
u
(
w
)
−
u
(
w
−
180
)
]
}
\frac{\pi}{90}g_{180}(w)*\frac{\pi}{90}g_{180}(w)=(\frac{\pi}{90})^2\bigg\{(w+180)\Big[u(w+180)-u(w)\Big]+(180-w)\Big[u(w)-u(w-180)\Big]\bigg\}
90πg180(w)∗90πg180(w)=(90π)2{(w+180)[u(w+180)−u(w)]+(180−w)[u(w)−u(w−180)]}
则 w m = 180 w_m=180 wm=180,奈奎斯特取样角频率为:
w s = 360 w_s=360 ws=360
- (3)
S
a
(
90
t
)
+
S
a
(
50
t
)
Sa(90t)+Sa(50t)
Sa(90t)+Sa(50t)
w 1 = 90 w_1=90 w1=90, w 2 = 50 w_2=50 w2=50,则 w m = 90 w_m=90 wm=90,奈奎斯特取样角频率为:
w s = 180 w_s=180 ws=180
- (4)
S
a
(
100
t
)
∗
S
a
(
70
t
)
Sa(100t)*Sa(70t)
Sa(100t)∗Sa(70t)
根据时域卷积性质 f 1 ( t ) ∗ f 2 ( t ) ↔ F 1 ( j w ) ⋅ F 2 ( j w ) f_1(t)*f_2(t)\leftrightarrow F_1(jw)\cdot F_2(jw) f1(t)∗f2(t)↔F1(jw)⋅F2(jw),则:
S
a
(
100
t
)
∗
S
a
(
70
t
)
↔
π
100
g
200
(
w
)
⋅
π
70
g
140
(
w
)
Sa(100t)*Sa(70t)\leftrightarrow \frac{\pi}{100}g_{200}(w)\cdot\frac{\pi}{70}g_{140}(w)
Sa(100t)∗Sa(70t)↔100πg200(w)⋅70πg140(w)
g
140
(
w
)
g_{140}(w)
g140(w) 就像一个滤波器将
70
<
∣
w
∣
≤
100
70<|w|\leq100
70<∣w∣≤100 部分滤掉,
w
m
=
70
w_m=70
wm=70,奈奎斯特取样角频率为:
w s = 140 w_s=140 ws=140
3 p ( t ) p(t) p(t) 为 δ T ( t ) \delta_T(t) δT(t) 时的取样
δ T ( t ) = ∑ n = − ∞ n = + ∞ δ ( t − n T s ) \delta_T(t)=\sum_{n=-\infty}^{n=+\infty}\delta(t-nT_s) δT(t)=∑n=−∞n=+∞δ(t−nTs) 其实就是 τ → 0 \tau\to{0} τ→0 的周期矩阵信号
δ T ( t ) \delta_T(t) δT(t) 的傅里叶系数 F n = 1 T s F_n=\frac{1}{T_s} Fn=Ts1,则 δ T ( t ) ↔ 2 π T s ∑ n = − ∞ + ∞ δ ( w − n w 0 ) \delta_T(t)\leftrightarrow{\frac{2\pi}{T_s}\sum_{n=-\infty}^{+\infty}\delta(w-nw_0)} δT(t)↔Ts2π∑n=−∞+∞δ(w−nw0)
f
s
(
t
)
=
f
(
t
)
⋅
δ
T
(
t
)
=
f
(
t
)
⋅
∑
n
=
−
∞
+
∞
δ
(
t
−
n
T
s
)
=
∑
n
=
−
∞
n
=
+
∞
f
(
n
T
s
)
δ
(
t
−
n
T
s
)
f_s(t)=f(t)\cdot\delta_T(t)=f(t)\cdot\sum_{n=-\infty}^{+\infty}\delta(t-nT_s)=\sum_{n=-\infty}^{n=+\infty}f(nT_s)\delta(t-nT_s)
fs(t)=f(t)⋅δT(t)=f(t)⋅n=−∞∑+∞δ(t−nTs)=n=−∞∑n=+∞f(nTs)δ(t−nTs)
F
s
(
j
w
)
=
1
2
π
F
(
j
w
)
∗
2
π
T
s
∑
n
=
−
∞
+
∞
δ
(
w
−
n
w
0
)
=
1
T
s
∑
n
=
−
∞
+
∞
F
[
j
(
w
−
n
w
0
)
]
F_s(jw)=\frac{1}{2\pi}F(jw)*{\frac{2\pi}{T_s}\sum_{n=-\infty}^{+\infty}\delta(w-nw_0)}=\frac{1}{T_s}\sum_{n=-\infty}^{+\infty}F[j(w-nw_0)]
Fs(jw)=2π1F(jw)∗Ts2πn=−∞∑+∞δ(w−nw0)=Ts1n=−∞∑+∞F[j(w−nw0)]
与使用周期矩阵信号取样得到的 F s ( j w ) = ∑ n = − ∞ + ∞ F n ⋅ F [ j ( w − n w s ) ] F_s(jw)=\sum_{n=-\infty}^{+\infty}F_n\cdot F[j(w-nw_s)] Fs(jw)=∑n=−∞+∞Fn⋅F[j(w−nws)] 相比:
等效于 F n = 1 T s F_n=\frac{1}{T_s} Fn=Ts1 ,即没有系数对频谱的离散幅度进行加权
例、在时域对连续信号进行理想采样,对其频谱描述正确的是(A)
A、频谱离散化
B、频谱周期化
取样就是周期性地采样数据