-
Notifications
You must be signed in to change notification settings - Fork 31
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
Showing
12 changed files
with
230 additions
and
19 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Binary file not shown.
This file was deleted.
Oops, something went wrong.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,14 @@ | ||
--- | ||
tags: | ||
- 数学 | ||
dlink: | ||
- "[[--概率论--]]" | ||
author: | ||
- Cyletix | ||
finished: false | ||
chapter: 13 | ||
--- | ||
- [[平稳随机过程的概念]] | ||
- [[各态历经性]] | ||
- [[相关函数]] | ||
- [[平稳随机过程的功率谱密度]] |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,67 @@ | ||
--- | ||
tags: | ||
- 数学 | ||
dlink: | ||
- "[[--概率论--]]" | ||
author: | ||
- PaulSun | ||
--- | ||
考虑宽平稳随机过程 $X\left(t\right)$ 的 Fourier 展开 | ||
$$X\left(t\right)=\dfrac1{2\pi}\sum^{+\infty}_{k=-\infty}\left(\int^{T/2}_{-T/2}X\left(s\right)\exp\left\{-j\omega_ks\right\}\,\mathrm ds\right)\exp\left\{j\omega_kt\right\}\cdot\dfrac{2\pi}{T} | ||
$$ | ||
其中根据 Fourier 展开的要求,$T\to\infty$,但问题出现: | ||
$$\int^{+\infty}_{-\infty}\left|X\left(t\right)\right|\,\mathrm dt<\infty$$ | ||
未必成立,即 Fourier 展开不一定存在。这宣告用 Fourier 展开对宽平稳随机过程进行谱分析的思路失败(这不意味着非宽平稳随机过程也不可以)。 | ||
|
||
# 功率谱密度函数的获得 | ||
|
||
为了避开如上所述的问题, Wiener 和 Khinchine 把将问题改为计算 | ||
$$S_X\left(\omega\right)=\lim_{T\to\infty}\dfrac1T\,\mathrm E\,\left|\int^{T/2}_{-T/2}X\left(t\right)\exp\left\{-j\omega t\right\}\,\mathrm dt\right|^2$$ | ||
计算过程如下 | ||
$$\begin{aligned}&\;\dfrac1T\,\mathrm E\left(\int^{T/2}_{-T/2}X\left(t\right)\exp\left\{-j\omega t\right\}\,\mathrm dt\right)\overline{\left(\int^{T/2}_{-T/2}X\left(s\right)\exp\left\{-j\omega s\right\}\,\mathrm ds\right)} | ||
\\\\ = & \;\dfrac1T\int^{T/2}_{-T/2}\int^{T/2}_{-T/2}\mathrm E\left(X\left(t\right)\,\overline{X\left(s\right)}\right)\exp\left\{-j\omega\left(t-s\right)\right\}\,\mathrm dt\mathrm ds | ||
\\\\ = & \;\dfrac1T\iint R_X\left(u\right)\exp\left\{-j\omega u\right\}\,\dfrac12\,\mathrm du\mathrm dv | ||
\\\\ = & \;\dfrac1T\left(\int^0_{-T}\int^{u+T}_{-u-T}+\int^T_0\int^{-u+T}_{u-T}\right)\,R_X\left(u\right)\exp\left\{-j\omega u\right\}\,\dfrac12\,\mathrm dv\mathrm du | ||
\\\\ = & \;\dfrac1T\int^T_{-T}\int^{-\left|u\right|+T}_{\left|u\right|-T}R_X\left(u\right)\exp\left\{-j\omega u\right\}\,\dfrac12\mathrm dv\mathrm du | ||
\\\\ = & \;\dfrac1T\int^T_{-T}\left(T-\left|u\right|\right)\,R_X\left(u\right)\exp\left\{-j\omega u\right\}\,\mathrm du | ||
\\\\ = & \;\int^T_{-T}\left(1-\dfrac{\left|u\right|}T\right)\,R_X\left(u\right)\exp\left\{-j\omega u\right\}\,\mathrm du | ||
\\\\ = & \;\int^{+\infty}_{-\infty} R_X\left(u\right)\exp\left\{-j\omega u\right\}\,\mathrm du \ \ \left(T\to\infty\right) | ||
\end{aligned}$$ | ||
|
||
虽然我们不能找到随机过程的 Fourier 变换,但最后结果暗示找到了随机过程的相关函数的 Fourier 变换,即前文定义的新函数 $S_X\left(\omega\right)$ | ||
$$\begin{aligned} | ||
& S_X\left(\omega\right)=\int^{+\infty}_{-\infty} R_X\left(\tau\right)\exp\left\{-j\omega \tau\right\}\,\mathrm d\tau | ||
\\\\ & R_X\left(\tau\right)=\dfrac1{2\pi}\int^{+\infty}_{-\infty} S_X\left(\omega\right)\exp\left\{-j\omega \tau\right\}\,\mathrm d\omega | ||
\end{aligned}\tag1$$ | ||
称 $S_X\left(\omega\right)$ 为功率谱密度。Bochner 定理是指:一个函数是正定函数,当且仅当它的 Fourier 变换对子恒为正数。从功率谱密度函数不难看出其恒正,那么相关函数是正定函数。 | ||
|
||
> [!theorem] Wiener-Khinchine 定理 | ||
> 任意一个宽平稳随机过程的功率谱密度是其相关函数的 Fourier 变换。 | ||
|
||
## 性质 | ||
|
||
### 性质 1 | ||
|
||
从 (1) 可知 | ||
$$\int^{+\infty}_{-\infty} S_X\left(\omega\right)\,\mathrm d\omega =2\pi R_X\left(0\right)=2\pi\mathrm E\left(X^2\left(t\right)\right)$$ | ||
|
||
### 性质 2 | ||
|
||
考察功率谱密度函数是否具有线性性。假设假设 $\alpha\in\mathbb R$ | ||
$$\begin{aligned} | ||
S_{\alpha X}\left(\omega\right) & = \int^{+\infty}_{-\infty} R_{\alpha X}\left(\tau\right)\exp\left\{-j\omega \tau\right\}\,\mathrm d\tau | ||
\\\\ & = \int^{+\infty}_{-\infty} R_{\alpha X}\left(\alpha X\left(t\right),\alpha X\left(s\right)\right)\exp\left\{-j\omega \tau\right\}\,\mathrm d\tau | ||
\\\\ & = \int^{+\infty}_{-\infty} \mathrm E\left(\alpha X\left(t\right),\alpha X\left(s\right)\right)\exp\left\{-j\omega \tau\right\}\,\mathrm d\tau | ||
\\\\ & = \int^{+\infty}_{-\infty} \left|\alpha\right|^2\,\mathrm E\left(X\left(t\right),X\left(s\right)\right)\exp\left\{-j\omega \tau\right\}\,\mathrm d\tau | ||
\\\\ & = \int^{+\infty}_{-\infty} \left|\alpha\right|^2\,R_{X}\left(\tau\right)\exp\left\{-j\omega \tau\right\}\,\mathrm d\tau | ||
\\\\ & = \left|\alpha\right|^2\,S_X\left(\omega\right) | ||
\end{aligned}$$ | ||
结果不满足 $S_{\alpha X}\left(\omega\right)=\alpha S_X\left(\omega\right)$,所以功率谱密度没有线性性。 | ||
|
||
### 性质 3 | ||
|
||
对功率谱密度的 Fourier 变换表达式使用欧拉公式 | ||
$$\begin{aligned}S_X\left(\omega\right) & =\int^{+\infty}_{-\infty}R_X\left(\tau\right)\,\cos\left(-\omega \tau\right)\,\mathrm d\tau+j\int^{+\infty}_{-\infty}R_X\left(\tau\right)\,\sin\left(-\omega \tau\right)\,\mathrm d\tau\\\\&=\int^{+\infty}_{-\infty}R_X\left(\tau\right)\,\cos\left(\omega \tau\right)\,\mathrm d\tau+0=S_X\left(-\omega\right)\end{aligned}$$ | ||
可见功率谱密度函数是偶函数。并且在该推导基础上,相关函数的 Fourier 变换表达式简化为 | ||
$$R_X\left(\tau\right)=\dfrac1{2\pi}\int^{+\infty}_{-\infty}S_X\left(\omega\right)\,\cos\left(\omega\tau\right)\,\mathrm d\omega$$ |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,65 @@ | ||
--- | ||
tags: | ||
- 数学 | ||
dlink: | ||
- "[[--概率论--]]" | ||
author: | ||
- PaulSun | ||
--- | ||
如果随机过程的某一种性质不随下角标(时间)变化而变化,则称随机过程具有该性质的平稳性。 | ||
|
||
众多平稳性是根据相关函数来建立的,所以有必要先介绍相关函数的定义。 | ||
|
||
### 相关函数 | ||
|
||
相关函数是两个时刻的函数,其为两时刻随机变量的相关性,写作 | ||
$$R\left(t,s\right)=\mathrm E\left[X\left(t\right),X\left(s\right)\right]$$ | ||
务必注意相关函数不是由协方差来定义的,上式与协方差的表达式不同。 | ||
|
||
还有许多根据相关性定义出的相关函数,例如相关系数、互相关函数。我们这里研究的是 “自相关函数”,即随机过程自身在不同时刻的相关性。 | ||
|
||
### 宽平稳 | ||
|
||
> [!definition] 定义 - 宽平稳 | ||
> 随机过程 $X\left(t\right)$ 是宽平稳的,如果对于任意时刻 $t$ 和 $s$,以及任意时长 $T$,有 | ||
> $$R\left(t+T,s+T\right)=R\left(t,s\right)$$ | ||
宽平稳是相关函数具有稳定性的平稳,它是随机过程中最重要的平稳性,是研究出发的基石。根据宽平稳的定义,其暗示我们:随机过程中两个时刻随机变量的相关性只依赖于时刻的相对位置,从此衍生第二种定义 | ||
$$R\left(t,s\right)=R\left(t-s\right)=:R\left(\tau\right)$$ | ||
通常在证明一个随机过程具有宽平稳性质时,目标就是得到第二种定义的表达式。以下用 “相位调频” 为例,说明在证明中如何得到 $R\left(t-s\right)$。 | ||
|
||
假设随机过程 $X\left(t\right)=\cos\left(2\pi f_0t+\theta\right)$,其中 $\theta\sim U\left(0,2\pi\right)$。计算一阶矩 | ||
$$\mathrm E\left(X\left(t\right)\right)=\dfrac1{2\pi}\displaystyle\int^{2\pi}_0\cos\left(2\pi f_0t+\theta\right)\,\mathrm d\theta=0$$ | ||
计算相关函数 | ||
$$\begin{aligned}R\left(t,s\right)&=\mathrm E\left(X\left(t\right)\,X\left(s\right)\right)=\dfrac1{2\pi}\displaystyle\int^{2\pi}_0\cos\left(2\pi\,f_0t+\theta\right)\,\cos\left(2\pi f_0s+\theta\right)\,\mathrm d\theta | ||
\\\\ & =\dfrac1{4\pi}\displaystyle\int^{2\pi}_0\cos\left(2\pi f_0\left(t+s\right)+2\theta\right)+\cos\left(2\pi f_0\left(t-s\right)\right)\,\mathrm d\theta | ||
\\\\ & = \dfrac12\cos\left(2\pi f_0\left(t-s\right)\right)=R\left(t-s\right) | ||
\end{aligned} | ||
$$ | ||
|
||
### 严平稳 | ||
|
||
> [!definition] 定义 - 严平稳 | ||
> 随机过程 $X\left(t\right)$ 是严平稳的,如果对于任意时刻 $t_1,t_2,\dots,t_n$ 及所有时间长 $T$,以下两个随机向量服从相同联合分布 | ||
> $$\begin{array}c\left[X\left(t_1\right),X\left(t_2\right),\dots,X\left(t_n\right)\right]^T\\\\\left[X\left(t_1+T\right),X\left(t_2+T\right),\dots,X\left(t_n+T\right)\right]^T\end{array}$$ | ||
显然,独立同分布的随机过程是宽平稳的。因为这个性质太严格,所以在研究中较少用到。 | ||
|
||
### 循环平稳 | ||
|
||
> [!definition] 定义 - 循环平稳 | ||
> 随机过程 $X\left(t\right)$ 是循环平稳的,如果对于任意时刻 $t$ 和 $s$,存在时长 $T$,有 | ||
> $$R\left(t+T,s+T\right)=R\left(t,s\right)$$ | ||
如果说宽平稳的相关函数像一条平稳的线,那么循环平稳的相关函数则像正弦函数上下波动。考虑有没有什么办法让正弦函数被压成一条线呢?从技术上是有的,可以加入一个与之对冲的随机变量,把循环平稳随机过程处理成宽平稳随机过程,这是常见的手法。 | ||
|
||
### 增量平稳 | ||
|
||
> [!definition] 定义 - 增量平稳 | ||
> 随机过程 $X\left(t\right)$ 是增量平稳的,如果增量服从时间差的分布函数,即 | ||
> $$X\left(t\right)-X\left(s\right)\sim F\left(t-s\right)$$ | ||
常见的增量平稳过程有布朗运动 | ||
$$B\left(t\right)-B\left(s\right)\sim \mathrm{N}\left(t-s\right)$$ | ||
和泊松过程 | ||
$$N\left(t\right)-N\left(s\right)\sim\mathrm{Poisson}\left(t-s\right)$$ |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,79 @@ | ||
--- | ||
tags: | ||
- 数学 | ||
dlink: | ||
- "[[---平稳随机过程---]]" | ||
author: | ||
- PaulSun | ||
--- | ||
相关函数我们主要研究的是宽平稳随机过程的相关函数,即能被记作 $R\left(t-s\right)$ 或 $R\left(\tau\right)$ 的相关函数,这样的相关函数具有五条基本性质和正定性。 | ||
|
||
# 基本性质 | ||
|
||
> [!property] 相关函数的五条基本性质 | ||
> 1. $R\left(0\right)\ge0$,即 $R\left(t,t\right)\ge0$ | ||
> 2. $R\left(\tau\right)=R\left(-\tau\right)$,即 $R\left(t,s\right)=R\left(s,t\right)$ | ||
> 3. $\left|R\left(\tau\right)\right|\le R\left(0\right)$ | ||
> 4. 存在 $\tau$,使得 $R\left(\tau\right)=R\left(0\right)$,那么对于任意 $t$,$R\left(t+\tau\right)=R\left(t\right)$ | ||
> 5. 如果 $R\left(\tau\right)$ 在 $\tau=0$ 连续,那么 $R\left(\tau\right)$ 在定义域内连续 | ||
前两条性质是符合直觉的基础性质。性质 3 说明 $\tau=0$ 处相关函数取得全局最大值,在邻域 $B\left(0,\delta\right)$ 内,相关函数先增后减。但相关函数在定义域内并非一定先增后减,例如 $\left(\delta,+\infty\right)$ 内可以为非单调函数。如果在 $\left(\delta,+\infty\right)$ 的子区间内相关函数递增,且函数值达到 $R\left(0\right)$,那么相关函数一定是周期函数,这由性质 4 保证。 | ||
|
||
### 证明 - 性质 3 | ||
|
||
根据 Cauchy-Schwarz 不等式有 | ||
$$\left| R\left( \tau \right) \right|=\left| \mathrm{E}\left( X\left( t \right) \,X\left( t+\tau \right) \right) \right|\le \left( \mathrm{E}X^2\left( t \right) \,\mathrm{E}X^2\left( t+\tau \right) \right) ^{1/2} | ||
$$ | ||
其中 | ||
$$\begin{aligned}&\mathrm EX^2\left(t\right)=\mathrm E\left(X\left(t\right)\,X\left(t\right)\right)=R\left(t,t\right)=R\left(0\right)\\\\&\mathrm EX^2\left(t+\tau\right)=\mathrm E\left(X\left(t+\tau\right)\,X\left(t+\tau\right)\right)=R\left(0\right)\end{aligned} | ||
$$ | ||
因此 | ||
$$ | ||
\left| R\left( \tau \right) \right|\le \left( \mathrm{E}X^2\left( t \right) \,\mathrm{E}X^2\left( t+\tau \right) \right) ^{1/2}=\left(R^2\left(0\right)\right)^{1/2}=R\left(0\right) | ||
$$ | ||
|
||
### 证明 - 性质 4 | ||
|
||
考虑一个新式子 | ||
$$\begin{aligned} | ||
\mathrm E\,\left|X\left(t\right)-X\left(t+\tau\right)\right|^2 | ||
& = \mathrm E\left(X^2\left(t\right)+X^2\left(t+\tau\right)-2X\left(t\right)\,X\left(t+\tau\right)\right) | ||
\\\\ & = R\left(0\right)+R\left(0\right)-2R\left(\tau\right)=0 | ||
\end{aligned}$$ | ||
其中使用了 $R\left(\tau\right)=R\left(0\right)$ 的条件 | ||
$$\begin{aligned} | ||
\left| R\left( t \right) -R\left( t+\tau \right) \right|&=\left| \mathrm{E}\left( X\left( 0 \right) \,X\left( t \right) \right) -\mathrm{E}\left( X\left( 0 \right) \,X\left( t+\tau \right) \right) \right|\\\\ | ||
&=\left| \mathrm{E}\left( X\left( 0 \right) \left( X\left( t \right) -X\left( t+\tau \right) \right) \right) \right|\\\\ | ||
&\le \left( \mathrm{E}\left( X^2\left( 0 \right) \right) \,{\mathrm{E}\left( X\left( t \right) -X\left( t+\tau \right) \right) }^2 \right) ^{1/2} | ||
\end{aligned} | ||
$$ | ||
因此 $0\le\left| R\left( t \right) -R\left( t+\tau \right) \right|\le0$,$R\left(t\right)=R\left(t+\tau\right)$ 得证。 | ||
|
||
### 证明 - 性质 5 | ||
|
||
当 $\tau\to0$ 时 | ||
$$ | ||
\mathrm E\,\left|X\left(t\right)-X\left(t+\tau\right)\right|^2= R\left(0\right)+R\left(0\right)-2R\left(\tau\right)\to0 | ||
$$ | ||
后续证明同性质 (4) 的证明,此处省略。 | ||
|
||
# 正定性 | ||
|
||
一个函数是正定函数,如果如下形式的矩阵是正定函数 | ||
$$ | ||
A=\left(f\left(t_i-t_j\right)\right)_{ij}=\left[ \begin{matrix} | ||
f\left( 0 \right)& \cdots& f\left( t_1-t_j \right)& \cdots& f\left( t_1-t_n \right)\\\\ | ||
\vdots& & \vdots& & \vdots\\\\ | ||
f\left( t_i-t_1 \right)& \cdots& f\left( t_i-t_j \right)& \cdots& f\left( t_i-t_n \right)\\\\ | ||
\vdots& & \vdots& & \vdots\\\\ | ||
f\left( t_n-t_1 \right)& \cdots& f\left( t_n-t_j \right)& \cdots& f\left( 0 \right)\\ | ||
\end{matrix} \right] _{n\times n} | ||
$$ | ||
|
||
考察宽平稳随机过程的相关函数的正定性,引入向量记号 $X=\left(X\left(t_1\right),\dots,X\left(t_n\right)\right)^T$,那么 | ||
$$\left(R\left(t_i-t_j\right)\right)_{ij}=\left(\mathrm EX\left(t_i\right)\,\mathrm EX\left(t_j\right)\right)_{ij}=\mathrm E\left(XX^T\right)=R | ||
$$ | ||
对于 $\forall\,\alpha\in\mathbb R^n$,有 | ||
$$\alpha^TR\,\alpha=\alpha^T\,\mathrm E\left(XX^T\right)\,\alpha=\mathrm E\left(\alpha^TX\cdot\left(\alpha ^TX\right)^T\right)=\big\|\alpha^TX\big\|^2\ge0 | ||
$$ | ||
宽平稳随机过程的相关函数通过了正定性的验证。 |
File renamed without changes.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
|
@@ -2,7 +2,7 @@ | |
tags: | ||
- 数学 | ||
dlink: | ||
- "[[---随机过程及统计描述---]]" | ||
- "[[---随机过程---]]" | ||
aliases: | ||
- Stochastic Process | ||
--- | ||
|