Transformation of Random Signals in Dynamic System in Time Domain
- 1. Representation of system work quality and some statistical concepts
- 2. Versatile ergodicity
- 3. Properties of correlation functions
- 4. Experimental approach to determine the correlation function
- 5. Properties of Statically Determinate Random Signals Through Linear Dynamical Systems
- 6. Calculation of the mean square error of the system output
1. Representation of system work quality and some statistical concepts
Let the input of a dynamic system be u ( t ) u(t)u ( t ) , the output isx ( t ) x(t)x ( t ) , then the dynamic error ise ( t ) = u ( t ) − x ( t ) e(t) = u(t) - x(t)e(t)=u(t)−x ( t ) . When inputu ( t ) u(t)When u ( t ) is a random signal, e ( t ) e(t)e ( t ) is the random error.
Generally, under the effect of random input signals, the working quality of the system can be expressed by the random mean square error :
e 2 ‾ ( t ) = lim T → ∞ 1 2 T ∫ − TT e 2 ( t ) dt (1 ) \overline{e^2} (t) = \lim _{T \rightarrow \infty} \frac{1}{2T} \int_{-T} ^T e^2 (t) {\rm d} t \tag{1}e2(t)=T→∞lim2T _1∫−TTe2(t)dt(1)
Next, some basic concepts of statistics are introduced.
(1) Probability distribution function F ( x ) F(x)F ( x ) . The probability distribution function refers to the random variableXXX does not exceed a certain valuexxx(即 X < x X < x X<x ):
F ( x ) = P ( X < x ) F(x) = P (X < x)F(x)=P(X<x ) (2)Probability density distribution functionf ( x ) f(x)f ( x ) . It can be simply and roughly understood as "random variableXXX is equal to some valuexxThe probability of x ", or mathematically understood as "the probability distribution functionF ( x ) F(x)F(x)的导数”即可:
f ( x ) = d F ( x ) d x = lim Δ → 0 P ( x ≤ X ≤ x + Δ x ) Δ x = lim Δ → 0 F ( x + Δ x ) − F ( x ) Δ x f(x) = \frac{ {\rm d} F(x)}{ {\rm d} x} = \lim_{\Delta \rightarrow 0} \frac{ P \left( x \leq X \leq x + \Delta x\right)}{\Delta x} = \lim_{\Delta \rightarrow 0} \frac{F \left( x + \Delta x \right) - F(x)}{\Delta x} f(x)=dxdF(x)=Δ→0limΔx _P(x≤X≤x+Δ x )=Δ→0limΔx _F(x+Δ x )−F(x)This leads to
F ( x ) = ∫ ∞ xf ( x ) dx F(x) = \int _\infty ^xf(x) {\rm d} xF(x)=∫∞xf ( x ) d x random variableXXThe mathematical expectation of X
: x ~ = M [ X ] = ∫ − ∞ ∞ xf ( x ) dx \tilde x = M \left[ X \right] = \int _{-\infty} ^\infty xf(x) {\rm d} xx~=M[X]=∫−∞∞x f ( x ) d x random variableXXmmof Xm – 阶矩:
x ~ m = ∫ − ∞ ∞ x m f ( x ) d x \tilde x^m = \int _{-\infty} ^\infty x^m f(x) {\rm d} x x~m=∫−∞∞xm f(x)dxrandom variableXXmmof Xm – 阶中心矩:
M [ ( X − x ~ ) m ] = ∫ − ∞ ∞ ( X − x ~ ) m f ( x ) d x M \left[ \left( X - \tilde x \right)^m \right] = \int _{-\infty} ^\infty \left( X - \tilde x \right)^m f(x) {\rm d} x M[(X−x~)m]=∫−∞∞(X−x~)mf ( x ) d x variance (in fact, second-order central moment):
M [ ( X − x ~ ) 2 ] = ∫ − ∞ ∞ ( X − x ~ ) 2 f ( x ) dx M \left[ \left( X - \tilde x \right)^2 \right] = \int _{-\infty} ^\infty \left( X - \tilde x \right)^2 f(x) {\rm d} xM[(X−x~)2]=∫−∞∞(X−x~)2f ( x ) d x When considering time, the mean of the sample can be equated to the mathematical expectation:
mx ( t ) = x ~ ( t ) = M [ X ( t ) ] = ∫ − ∞ ∞ xf ( x , t ) dx , D x ( t ) = D [ X ( t ) ] = ∫ − ∞ ∞ [ X ( t ) − mx ( t ) ] 2 f ( x , t ) dx m_x (t) = \tilde x(t) = M \left[ X(t) \right] = \int _{-\infty} ^\infty xf(x, t) {\rm d} x, \\ D_x (t) = D \left[ X(t ) \right] = \int _{-\infty} ^\infty \left[ X(t) - m_x (t) \right]^2 f(x, t) {\rm d} xmx(t)=x~(t)=M[X(t)]=∫−∞∞xf(x,t)dx,Dx(t)=D[X(t)]=∫−∞∞[X(t)−mx(t)]2f(x,t ) d x and for different timet 1 , t 2 t_1, t_2t1,t2For the random process of , we can also have mathematical expectations:
M [ X ( t 1 ) X ( t 2 ) ] = ∫ − ∞ ∞ ∫ − ∞ ∞ x 1 x 2 f ( x 1 , t 1 , x 2 , t 2 ) dx 1 dx 2 = R ( t 1 , t 2 ) M \left[ X\left( t_1 \right) X\left( t_2 \right) \right] = \int_{-\infty} ^\infty \ int_{-\infty} ^\infty x_1 x_2 f \left( x_1, t_1, x_2, t_2 \right) {\rm d} x_1 {\rm d} x_2 = R \left( t_1, t_2 \right)M[X(t1)X(t2)]=∫−∞∞∫−∞∞x1x2f(x1,t1,x2,t2)dx1dx2=R(t1,t2) is calledthe correlation function. The correlation function characterizesthe connection of random variables between different moments.
For a random variable xxx , its correlation function at different timesR x ( t 1 , t 2 ) R_x \left( t_1, t_2 \right)Rx(t1,t2) calledxxAutocorrelation functionof x . And for two different random variablesx , yx,yx,y , the correlation function R at different timesR xy ( t 1 , t 2 ) R_{xy} \left( t_1, t_2 \right)Rxy(t1,t2) is calledthe cross-correlation function. Obviously, whent 2 = t 1 + τ t_2 = t_1 + \taut2=t1+When τ , the autocorrelation function can also be expressed as
R ( τ ) = M [ X ( t 1 ) X ( t 1 + τ ) ] = ∫ − ∞ ∞ dx 1 ∫ − ∞ ∞ x 1 x 2 f ( x 1 , x 2 , τ ) dx 2 R (\tau) = M \left[ X\left( t_1 \right) X\left( t_1 + \tau \right) \right] = \int_{-\infty} ^\infty {\rm d} x_1 \int_{-\infty} ^\infty x_1 x_2 f \left( x_1, x_2, \tau \right) {\rm d} x_2R ( τ )=M[X(t1)X(t1+) ] _=∫−∞∞dx1∫−∞∞x1x2f(x1,x2,t )dx2
2. Versatile ergodicity
For a statically determinate process with ergodicity, the following formula holds:
x ~ = x ˉ , x 1 x 2 ~ = x 1 x 2 ‾ \tilde x = \bar x, \quad \widetilde{x_1 x_2} = \ overline{x_1 x_2}x~=xˉ,x1x2
=x1x2At the same time, due to the ergodicity of various states, the mean value of the random variable will not change due to the sampling period:
x ˉ = lim T → ∞ 1 2 T ∫ − TT x ( t ) dt = lim T → ∞ 1 T ∫ 0 T x ( t ) dt \bar x = \lim _{T \rightarrow \infty} \frac{1}{2T} \int_{-T} ^T x (t) {\rm d} t = \lim _{T \rightarrow \infty} \frac{1}{T} \int_{0} ^T x (t) {\rm d} txˉ=T→∞lim2T _1∫−TTx(t)dt=T→∞limT1∫0Tx ( t ) d t so the autocorrelation function is
R ( τ ) = x 1 x 2 ‾ = lim T → ∞ 1 T ∫ 0 T x ( t ) x ( t + τ ) dt R (\tau) = \ overline{x_1 x_2} = \lim _{T \rightarrow \infty} \frac{1}{T} \int _0 ^T x(t) x(t + \tau) {\rm d} tR ( τ )=x1x2=T→∞limT1∫0Tx(t)x(t+τ)dt
3. Properties of correlation functions
Some properties of the correlation function are given below:
R xy ( τ ) = R yx ( − τ ) ; R_{xy} (\tau) = R_{yx} (-\tau);Rxy( t )=Ryx(−τ); R y x ( τ ) = lim T → ∞ 1 2 T ∫ − T T y ( t ) x ( t + τ ) d t ; R_{yx} (\tau) = \lim _{T \rightarrow \infty} \frac{1}{2T} \int _{-T} ^T y(t) x(t + \tau) {\rm d} t; Ryx( t )=T→∞lim2T _1∫−TTy(t)x(t+τ)dt; R y x ( − τ ) = lim T → ∞ 1 2 T ∫ − T T y ( t ) x ( t − τ ) d t R_{yx} (-\tau) = \lim _{T \rightarrow \infty} \frac{1}{2T} \int _{-T} ^T y(t) x(t - \tau) {\rm d} t Ryx( − τ )=T→∞lim2T _1∫−TTy(t)x(t−τ ) d t and if sett 1 = t − τ t_1 = t - \taut1=t−τ则
R y x ( − τ ) = lim T → ∞ 1 2 T ∫ − T T y ( t 1 + τ ) x ( t 1 ) d t = R x y ( τ ) R_{yx} (-\tau) = \lim _{T \rightarrow \infty} \frac{1}{2T} \int _{-T} ^T y(t_1 + \tau) x(t_1) {\rm d} t = R_{xy} (\tau) Ryx( − τ )=T→∞lim2T _1∫−TTy(t1+t ) x ( t1)dt=Rxy( τ )当τ = 0 \tau=0t=0时:
R x ( 0 ) = M [ X 2 ( t ) ] = x 2 ‾ = D x R_x (0) = M \left[ X^2 (t) \right] = \overline{x^2} = D_x Rx(0)=M[X2(t)]=x2=Dx当τ → ∞ \tau \rightarrow \inftyt→∞时:
R x ( τ → ∞ ) = ( x ~ ) 2 = ( x ˉ ) 2 R_x ( \tau \rightarrow \infty ) = \left( \tilde x \right) ^2 = \left( \bar x \right) ^2 Rx( t→∞)=(x~)2=(xˉ)2 In particular, the correlation function of white noise is:
R x ( τ ) = N 2 δ ( τ ) R_x (\tau) = N^2 \delta (\tau)Rx( t )=N2 d(t)
4. Experimental approach to determine the correlation function
For a statically determinate ergodic system, its correlation function:
R x ( τ ) = X ( t ) X ( t + τ ) ‾ = lim T → ∞ 1 T ∫ 0 T x ( t ) x ( t + τ ) dt R_x(\tau) = \overline{X(t) X(t + \tau)} = \lim_{T \rightarrow \infty} \frac{1}{T} \int_0 ^T x(t) x( t + \tau) {\rm d} tRx( t )=X(t)X(t+t )=T→∞limT1∫0Tx(t)x(t+τ ) d t its estimated value is
R ^ x ( τ ) = 1 T ∫ 0 T x ( t ) x ( t − τ ) dt \hat R_x (\tau) = \frac{1}{T} \int_0 ^ T x(t) x(t - \tau) {\rm d} tR^x( t )=T1∫0Tx(t)x(t−τ ) d t In practice, in order to make the estimated value as accurate as possible, the experimental timeTTT as long as possible.
5. Properties of Statically Determinate Random Signals Through Linear Dynamical Systems
Suppose a random signal u ( t ) u(t)u ( t ) , its correlation function isR u ( τ ) R_u (\tau)Ru( τ ) , then it passes through the pulse signalK ( t ) K(t)K ( t ) system, the output is still a random signal:
x ( t ) = ∫ − ∞ ∞ K ( λ ) u ( t − λ ) d λ = K ( λ ) ∗ u ( λ ) (2) x(t) = \int_{-\infty} ^\infty K(\lambda) u(t - \lambda) {\rm d} \lambda = K(\lambda) * u( \lambda) \tag{2 }x(t)=∫−∞∞K ( λ ) u ( t−l ) d l=K ( λ )∗u ( λ )( 2 ) is the convolution of the two.
pair outputxxx求数学期望:
M [ x ( t ) ] = ∫ − ∞ ∞ K ( λ ) M [ u ( t − λ ) ] d λ M \left[ x(t) \right] = \int _{-\infty} ^\infty K(\lambda) M \left[ u(t - \lambda) \right] {\rm d} \lambda M[x(t)]=∫−∞∞K ( λ ) M[u(t−l ) ]d λwhile att + τ t+\taut+τ time:
x ( t + τ ) = ∫ − ∞ ∞ K ( η ) u ( t + τ − η ) d η x(t + \tau) = \int_{-\infty} ^\infty K(\eta ) u(t + \tau - \eta) {\rm d} \etax(t+t )=∫−∞∞K ( η ) u ( t+t−η ) d η can calculatexxx correlation function:
R x ( τ ) = M [ x ( t ) x ( t + τ ) ] = M [ ∫ − ∞ ∞ K ( λ ) u ( t − λ ) d λ ∫ − ∞ ∞ K ( η ) u ( t + τ − η ) d η ] = ∫ − ∞ ∞ K ( λ ) d λ ∫ − ∞ ∞ M [ u ( t − λ ) u ( t + τ − η ) ] K ( η ) d η \begin{aligned} R_x (\tau) &= M \left[ x(t) x(t + \tau) \right] \\ &= M \left[ \int_{-\infty} ^\infty K( \lambda) u(t - \lambda) {\rm d} \lambda \int_{-\infty} ^\infty K(\eta) u(t + \tau - \eta) {\rm d} \eta \ right] \\ &= \int_{-\infty} ^\infty K(\lambda) {\rm d} \lambda \int_{-\infty} ^\infty M \left[ u(t - \lambda) u (t + \tau - \eta) \right] K(\eta) {\rm d} \eta \end{aligned}Rx( t )=M[x(t)x(t+) ] _=M[∫−∞∞K ( λ ) u ( t−l ) d l∫−∞∞K ( η ) u ( t+t−h ) d h ]=∫−∞∞K ( λ ) d λ∫−∞∞M[u(t−λ ) u ( t+t−h ) ]K ( η ) d η设 t ′ = t − λ t' = t - \lambda t′=t−λ,则
M [ u ( t ′ ) u ( t ′ + λ + τ − η ) ] = R u ( τ + λ − η ) M \left[ u(t') u(t' + \lambda+ \tau - \eta) \right] = R_u (\tau + \lambda - \eta)M[u(t′)u(t′+l+t−h ) ]=Ru( t+l−η )代入上式有
R x ( τ ) = ∫ − ∞ ∞ K ( λ ) d λ ∫ − ∞ ∞ R u ( τ + λ − η ) K ( η ) d η (3) R_x (\tau) = \int_{-\infty} ^\infty K(\lambda) {\rm d} \lambda \int_{-\infty} ^\infty R_u (\tau + \lambda - \eta) K(\eta) {\ rm d} \eta \tag{3}Rx( t )=∫−∞∞K ( λ ) d λ∫−∞∞Ru( t+l−h ) K ( h ) d h( 3 ) In addition,xxx sumuuu的互相关函数为(用到式(2)):
R x u ( τ ) = M [ x ( t ) u ( t − τ ) ] = M { [ ∫ − ∞ ∞ K ( λ ) u ( t − λ ) d λ ] u ( t − τ ) } = ∫ − ∞ ∞ K ( λ ) M [ u ( t − τ ) u ( t − λ ) ] d λ \begin{aligned} R_{xu} (\tau) &= M \left[ x(t) u(t - \tau ) \right] \\ &= M \left\{ \left[ \int_{-\infty} ^\infty K(\lambda) u(t - \lambda) {\rm d} \lambda \right] u(t - \tau) \right\} \\ &= \int_{-\infty} ^\infty K(\lambda) M \left[ u(t - \tau) u(t - \lambda) \right] {\rm d} \lambda \end{aligned} Rxu( t )=M[x(t)u(t−) ] _=M{
[∫−∞∞K ( λ ) u ( t−l ) d l ]u(t−) } _=∫−∞∞K ( λ ) M[u(t−t ) u ( t−l ) ]dλ任t − τ = t ′ t - \tau = t't−t=t′ ,in the case of
R xu ( τ ) = ∫ − ∞ ∞ K ( λ ) M [ u ( t ′ ) u ( t ′ + τ − λ ) ] d λ R_{xu} (\tau) = \int_ {-\infty} ^\infty K(\lambda) M \left[ u(t') u(t' + \tau - \lambda) \right] {\rm d} \lambdaRxu( t )=∫−∞∞K ( λ ) M[u(t′)u(t′+t−l ) ]dλ又由于 M [ u ( t ′ ) u ( t ′ + τ − λ ) ] = R u ( τ − λ ) M \left[ u(t') u(t' + \tau - \lambda) \right] = R_u (\tau - \lambda) M[u(t′)u(t′+t−l ) ]=Ru( t−λ ),代入上式有
R xu ( τ ) = ∫ − ∞ ∞ K ( λ ) R u ( τ − λ ) d λ (4) R_{xu} (\tau) = \int_{-\infty} ^ \infty K(\lambda) R_u (\tau - \lambda) {\rm d} \lambda \tag{4}Rxu( t )=∫−∞∞K ( λ ) Ru( t−l ) d l( 4 ) In the actual situation, whenλ < 0 \lambda < 0l<0时K ( λ ) ≡ 0 K(\lambda) \equivK ( λ )≡0,故式(3)(4) can also be written as
R x ( τ ) = ∫ 0 ∞ K ( λ ) d λ ∫ 0 ∞ R u ( τ + λ − η ) K ( η ) d η (3) R_x (\tau) = \int_0 ^\infty K(\lambda) {\rm d} \lambda \int_0 ^\infty R_u (\tau + \lambda - \eta) K(\eta) {\rm d} \ eta \tag{3}Rx( t )=∫0∞K ( λ ) d λ∫0∞Ru( t+l−h ) K ( h ) d h( 3 ) R xu ( τ ) = ∫ 0 ∞ K ( λ ) R u ( τ − λ ) d λ (4) R_{xu} (\tau) = \int_0 ^\infty K(\lambda) R_u (\ tau - \lambda) {\rm d} \lambda \tag{4}Rxu( t )=∫0∞K ( λ ) Ru( t−l ) d l(4)
6. Calculation of the mean square error of the system output
As mentioned earlier, when τ = 0 \tau=0t=0时:
R x ( 0 ) = M [ X 2 ( t ) ] = x 2 ‾ = D x R_x (0) = M \left[ X^2 (t) \right] = \overline{x^2} = D_x Rx(0)=M[X2(t)]=x2=Dx即语电影方差。上式即(用到式(4)):
R x ( 0 ) = R x ( τ ) ∣ τ = 0 = ∫ 0 ∞ K ( λ ) d λ ∫ 0 ∞ R u ( τ + λ − η ) K ( η ) d η ∣ τ = 0 = ∫ 0 ∞ K ( λ ) d λ ∫ 0 ∞ R u ( λ − η ) K ( η ) d η ⏟ R xu ( λ ) = ∫ 0 ∞ K ( λ ) R xu ( λ ) d λ (5) \begin{aligned} R_x (0) &= R_x (\tau) \big\rvert _{\tau = 0} = \int_0 ^\infty K( \lambda) {\rm d} \lambda \int_0 ^\infty R_u (\tau + \lambda - \eta) K(\eta) {\rm d} \eta \Big\rvert _{\tau = 0} \ \ &= \int_0 ^\infty K(\lambda) {\rm d} \lambda \underbrace{\int_0 ^\infty R_u ( \lambda - \eta) K(\eta) {\rm d} \eta}_ {R_{xu} (\lambda)} \\ &= \int_0 ^\infty K(\lambda) R_{xu} (\lambda) {\rm d} \lambda \tag{5} \end{aligned}Rx(0)=Rx( t )
τ = 0=∫0∞K ( λ ) d λ∫0∞Ru( t+l−h ) K ( h ) d h
τ = 0=∫0∞K ( λ ) d λRxu( l )
∫0∞Ru( l−η ) K ( η ) d η=∫0∞K ( λ ) Rxu( λ ) d λ( 5 ) promptD
x = x 2 ‾ = ∫ 0 ∞ K ( λ ) R xu ( λ ) d λ (6) D_x = \overline{x^2} = \int_0 ^\infty K(\lambda) R_{ xu} (\lambda) {\rm d} \lambda \tag{6}Dx=x2=∫0∞K ( λ ) Rxu( l ) d l(6)