各态历经性的判定

  • Post author:
  • Post category:其他




1.均值函数具有各态历经性的充要条件





X

=

{

X

t

,

<

t

<

+

}

X=\{X_t,-\infty<t<+\infty \}






X




=








{




X










t


















,












<








t




<








+





}





是平稳过程,则均值函数具有各态历经性的充要条件是




l

i

m

T

>

1

2

T

2

T

+

2

T

(

1

τ

2

T

)

C

x

(

τ

)

d

τ

=

0

lim_{T->\infty}\frac{1}{2T}\int_{-2T}^{+2T}(1-\frac{|\tau|}{2T})C_x(\tau)d\tau=0






l


i



m











T





>

































2


T














1






































2


T










+


2


T



















(


1
























2


T

















τ























)



C










x


















(


τ


)


d


τ




=








0






其中



C

x

(

τ

)

C_x(\tau)







C










x


















(


τ


)





是协方差函数

证明:





P

(

<

x

t

>

=

m

x

(

t

)

)

=

1

D

[

<

x

t

>

]

=

0

P(<x_t>=m_x(t))=1可得:D[<x_t>]=0






P


(


<









x










t




















>






=









m










x


















(


t


)


)




=








1











D


[


<









x










t




















>






]




=








0






(因为



X

t

X_t







X










t





















是常数,所以



X

t

X_t







X










t





















的方差为零)

,又




D

[

<

x

t

>

]

=

D

[

l

i

m

T

>

1

2

T

T

+

T

X

t

d

t

]

=

l

i

m

T

>

D

[

1

2

T

T

+

T

X

t

d

t

]

0

D[<x_t>]=D[lim_{T->\infty}\frac{1}{2T}\int_{-T}^{+T}X_tdt]=lim_{T->\infty}D[\frac{1}{2T}\int_{-T}^{+T}X_tdt]0






D


[


<









x










t




















>






]




=








D


[


l


i



m











T





>

































2


T














1






































T










+


T






















X










t


















d


t


]




=








l


i



m











T





>






















D


[













2


T














1






































T










+


T






















X










t


















d


t


]


0






对其进行计算




D

[

<

x

t

>

]

=

E

1

2

T

+

X

t

d

t

E

[

1

2

T

T

+

T

X

t

d

t

]

2

D[<x_t>]=E|\frac{1}{2T}\int_{-\infty}^{+\infty}X_tdt-E[\frac{1}{2T}\int_{-T}^{+T}X_tdt]|^2






D


[


<









x










t




















>






]




=








E
















2


T














1

















































+

























X










t


















d


t













E


[













2


T














1






































T










+


T






















X










t


















d


t


]














2

















=

E

1

2

T

T

+

T

X

t

m

x

d

t

2

=E|\frac{1}{2T}\int_{-T}^{+T}X_t-m_xdt|^2






=








E
















2


T














1






































T










+


T






















X










t






























m










x


















d


t














2

















=

E

[

1

4

T

2

T

+

T

(

X

s

m

x

)

d

s

T

+

T

(

X

t

m

x

)

d

t

]

=E[\frac{1}{4T^2}\int_{-T}^{+T}(\overline{X_s-m_x})ds\int_{-T}^{+T}({X_t-m_x})dt]






=








E


[













4



T










2





















1






































T










+


T



















(











X










s


























m










x







































)


d


s




















T










+


T



















(




X










t


























m










x



















)


d


t


]










=

1

4

T

2

T

+

T

T

+

T

E

[

(

X

s

m

x

)

(

X

t

m

x

)

]

d

s

d

t

=\frac{1}{4T^2}\int_{-T}^{+T}\int_{-T}^{+T}E[(\overline{X_s-m_x})({X_t-m_x})]dsdt






=



















4



T










2





















1






































T










+


T





































T










+


T





















E


[


(











X










s


























m










x







































)


(




X










t


























m










x



















)


]


d


s


d


t










=

1

4

T

2

T

+

T

T

+

T

C

x

(

t

s

)

d

s

d

t

=\frac{1}{4T^2}\int_{-T}^{+T}\int_{-T}^{+T}C_x(t-s)dsdt






=



















4



T










2





















1






































T










+


T





































T










+


T






















C










x


















(


t













s


)


d


s


d


t









u

=

t

s

,

v

=

t

+

s

u=t-s,v=t+s






u




=








t













s


,




v




=








t




+








s





,上式可以化简为




1

4

T

2

D

C

x

(

u

)

J

d

u

d

v

,

J

=

1

/

2

1

/

2

1

/

2

1

/

2

\frac{1}{4T^2}\int\int_{D}C_x(u)|J|dudv,|J|=\left |\begin{array}{cccc} -1/2 &1/2 \\ 1/2 &1/2 \end{array}\right|

















4



T










2





















1








































D






















C










x


















(


u


)





J





d


u


d


v


,







J







=















































































1


/


2








1


/


2





























1


/


2








1


/


2

















































































确认新的积分区域




{

v

+

u

=

2

T

v

u

=

2

T

\begin{cases} |v+u|=2T \\ |v-u|=2T \end{cases}








{

















v




+




u







=




2


T











v









u







=




2


T



























用之前变量的界表示新的变量的积分区域


在这里插入图片描述





=

1

4

T

2

1

2

[

2

T

0

d

u

2

T

u

2

T

+

u

C

x

(

u

)

d

v

+

0

2

T

d

u

2

T

+

u

2

T

u

C

x

(

u

)

d

v

]

=\frac{1}{4T^2}\cdot\frac{1}{2}\left [\int_{-2T}^0du\int_{-2T-u}^{2T+u} C_x(u)dv+\int_0^{2T}du\int_{-2T+u}^{2T-u}C_x(u)dv\right ]






=



















4



T










2





















1










































2














1
























[



















2


T









0




















d


u




















2


T





u










2


T


+


u






















C










x


















(


u


)


d


v




+
















0









2


T





















d


u




















2


T


+


u










2


T





u






















C










x


















(


u


)


d


v



]









分为左右两边分别积分





=

1

8

T

2

[

2

T

0

C

x

(

u

)

(

4

T

+

2

u

)

d

u

+

0

2

T

C

x

(

u

)

(

4

T

2

u

)

d

u

]

=\frac{1}{8T^2}\left [\int_{-2T}^0C_x(u)(4T+2u)du+\int_0^{2T}C_x(u)(4T-2u)du\right ]






=



















8



T










2





















1
























[



















2


T









0





















C










x


















(


u


)


(


4


T




+




2


u


)


d


u




+
















0









2


T






















C










x


















(


u


)


(


4


T









2


u


)


d


u



]












=

1

2

T

[

2

T

0

C

x

(

u

)

(

1

+

u

2

T

)

d

u

+

0

2

T

C

x

(

u

)

(

1

u

2

T

)

d

u

]

=\frac{1}{2T}\left [\int_{-2T}^0C_x(u)(1+\frac{u}{2T})du+\int_0^{2T}C_x(u)(1-\frac{u}{2T})du\right ]






=



















2


T














1
























[



















2


T









0





















C










x


















(


u


)


(


1




+















2


T














u




















)


d


u




+
















0









2


T






















C










x


















(


u


)


(


1




















2


T














u




















)


d


u



]












=

1

2

T

[

2

T

2

T

C

x

(

u

)

(

1

u

2

T

)

d

u

]

=\frac{1}{2T}\left [\int_{-2T}^{2T}C_x(u)(1-\frac{|u|}{2T})du\right ]






=



















2


T














1
























[



















2


T










2


T






















C










x


















(


u


)


(


1




















2


T

















u























)


d


u



]












=

1

2

T

[

2

T

2

T

C

x

(

τ

)

(

1

τ

2

T

)

d

τ

]

=\frac{1}{2T}\left [\int_{-2T}^{2T}C_x(\tau)(1-\frac{|\tau|}{2T})d\tau\right ]






=



















2


T














1
























[



















2


T










2


T






















C










x


















(


τ


)


(


1




















2


T

















τ























)


d


τ



]








😫😫😫😫😫😫终于整完啦!



2.实过程





X

X






X





是实过程,则各态历经性的充要条件是




l

i

m

T

>

+

1

T

[

0

2

T

C

x

(

τ

)

(

1

τ

2

T

)

d

τ

]

=

0

lim_{T->+\infty}\frac{1}{T}\left [\int_{0}^{2T}C_x(\tau)(1-\frac{\tau}{2T})d\tau\right ]=0






l


i



m











T





>


+

































T














1
























[
















0










2


T






















C










x


















(


τ


)


(


1




















2


T














τ




















)


d


τ



]






=








0






此时



C

x

(

τ

)

C_x(\tau)为偶函数







C










x


















(


τ


)


















因为是偶函数,所以直接运用偶函数的性质,就可得到上式。



3.



t

0

t\ge0






t













0





的情况

积分区域变为

在这里插入图片描述

所以各态历经的充要条件为:




l

i

m

T

>

+

1

T

[

T

T

C

x

(

τ

)

(

1

τ

T

)

d

τ

]

=

0

lim_{T->+\infty}\frac{1}{T}\left [\int_{-T}^{T}C_x(\tau)(1-\frac{\tau}{T})d\tau\right ]=0






l


i



m











T





>


+

































T














1
























[



















T










T






















C










x


















(


τ


)


(


1




















T














τ




















)


d


τ



]






=








0






此时



C

x

(

τ

)

C_x(\tau)为偶函数







C










x


















(


τ


)


















如果



X

X






X





实平稳过程,则相应结论为




l

i

m

T

>

+

2

T

[

0

T

C

x

(

τ

)

(

1

τ

T

)

d

τ

]

=

0

lim_{T->+\infty}\frac{2}{T}\left [\int_{0}^{T}C_x(\tau)(1-\frac{\tau}{T})d\tau\right ]=0






l


i



m











T





>


+

































T














2
























[
















0










T






















C










x


















(


τ


)


(


1




















T














τ




















)


d


τ



]






=








0







在这里插入图片描述

高木同学也很好看



版权声明:本文为xd15010130025原创文章,遵循 CC 4.0 BY-SA 版权协议,转载请附上原文出处链接和本声明。