You are on page 1of 4

Random Vibration

Lecturer: Cheng Rui-Hwa


Homework#4
Andy/M10605804

1. A stationary random process X(t) has an autocorrelation


function
𝑅𝑥𝑥 (𝜏) = 𝑘 𝑒 −𝑢|𝜏| for 𝑢 > 0 and 𝑘 > 0
a. Find power spectral density function 𝑆𝑥𝑥 (𝜔)
1 ∞
𝑆𝑥𝑥 (𝜔) = ∫ 𝑅𝑥𝑥 (𝜏) ∙ 𝑒 −𝑖𝑤𝜏 𝑑𝜏
2𝜋 −∞
𝑅𝑥𝑥 (𝜏) is an even function therefore the multiplication with odd function will
equal with zero after integration.
1 ∞
𝑆𝑥𝑥 (𝜔) = ∫ 𝑘 ∙ 𝑒 −𝑢|𝜏| ∙ cos 𝜔𝜏 𝑑𝜏
2𝜋 −∞
1 ∞
𝑆𝑥𝑥 (𝜔) = ∫ 𝑘 ∙ 𝑒 −𝑢𝜏 ∙ cos 𝜔𝜏 𝑑𝜏
𝜋 0
Using integration by parts
𝑢 = 𝑒 −𝑢𝜏 𝑑𝑣 = cos 𝜔𝜏 𝑑𝜏
sin 𝜔𝜏
𝑑𝑢 = −𝑢 ∙ 𝑒 −𝑢𝜏 𝑑𝜏 𝑣=
𝜔
sin 𝜔𝜏 𝑢
∫ 𝑒 −𝑢𝜏 ∙ cos 𝜔𝜏 𝑑𝜏 = 𝑒 −𝑢𝜏 + ∙ ∫ 𝑒 −𝑢𝜏 sin 𝜔𝜏 𝑑𝜏
𝜔 𝜔
𝑢 = 𝑒 −𝑢𝜏 𝑑𝑣 = sin 𝜔𝜏 𝑑𝜏
−𝑢𝜏
cos 𝜔𝜏
𝑑𝑢 = −𝑢 ∙ 𝑒 𝑑𝜏 𝑣=−
𝜔
sin 𝜔𝜏 𝑢 𝑢2
∫ 𝑒 −𝑢𝜏 ∙ cos 𝜔𝜏 𝑑𝜏 = 𝑒 −𝑢𝜏 − 2 𝑒 −𝑢𝜏 cos 𝜔𝜏 − 2 ∫ 𝑒 −𝑢𝜏 cos 𝜔𝜏 𝑑𝜏
𝜔 𝜔 𝜔
𝑢2 sin 𝜔𝜏 𝑢
(1 + 2 ) ∫ 𝑒 −𝑢𝜏 ∙ cos 𝜔𝜏 𝑑𝜏 = 𝑒 −𝑢𝜏 − 2 𝑒 −𝑢𝜏 cos 𝜔𝜏
𝜔 𝜔 𝜔
𝑒 −𝑢𝜏 𝜔 sin 𝜔𝜏 − 𝑢 cos 𝜔𝜏
∫ 𝑒 −𝑢𝜏 ∙ cos 𝜔𝜏 𝑑𝜏 =
𝜔2 𝑢2
(1 + 2 )
𝜔
𝜏=∞
−𝑢𝜏
𝑘 𝑒 𝜔 sin 𝜔𝜏 − 𝑢 cos 𝜔𝜏
𝑆𝑥𝑥 (𝜔) = [ 2 ]
𝜋 𝜔 𝑢2
(1 + 2 )
𝜔 𝜏=0
𝑘 −𝑢 𝑘𝑢
𝑆𝑥𝑥 (𝜔) = (0 − ( 2 )) = #
𝜋 𝜔 + 𝑢2 𝜋(𝜔 2 + 𝑢2 )
b. Plot 𝑅𝑥𝑥 (𝜏) and 𝑆𝑥𝑥 (𝜔)for 𝑘 = 1 and 𝑢 = 1,3,5,7, 𝑎𝑛𝑑 9
Tendency of 𝑅𝑥𝑥 (𝜏) and 𝑆𝑥𝑥 (𝜔) when u increases

Fig.1. 𝑅𝑥𝑥 (𝜏) with k=1 and u=1 Fig.2. 𝑅𝑥𝑥 (𝜏) with k=1 and u=3

Fig.3. 𝑅𝑥𝑥 (𝜏) with k=1 and u=5 Fig.4. 𝑅𝑥𝑥 (𝜏) with k=1 and u=7

Fig.5. 𝑅𝑥𝑥 (𝜏) with k=1 and u=9

Fig.6. 𝑆𝑥𝑥 (𝜔) with k=1 and u=1 Fig.7. 𝑆𝑥𝑥 (𝜔) with k=1 and u=3
Fig.8. 𝑆𝑥𝑥 (𝜔) with k=1 and u=5 Fig.9. 𝑆𝑥𝑥 (𝜔) with k=1 and u=7

Fig.10. 𝑆𝑥𝑥 (𝜔) with k=1and u=9

Can be observed from Fig.1 to Fig.5 that 𝑅𝑥𝑥 (𝜏) becomes narrower as u
becomes larger. This is because of the equation of 𝑅𝑥𝑥 (𝜏) itself. Exponential
equation will decay faster if the coefficient infront of the variable becomes
larger.
From Fig.6 to Fig.10 can be conclude that 𝑆𝑥𝑥 (𝜔) becomes wider as u becomes
larger. Power density function will be wider as coefficient correlation function
become narrower. Variation of frequency inside the correlation function
becomes higher as the autocorrelation function is narrow. Then, the maximum
value for PSD function tend to be lower with larger value of u.

c. Find the mean square value based on 𝑅𝑥𝑥 (𝜏) and 𝑆𝑥𝑥 (𝜔)
From 𝑅𝑥𝑥 (𝜏)
𝐸(𝑥 2 ) = 𝑅𝑥𝑥 (0) = 𝑘 𝑒 −𝑢|0| = 𝑘 #
From 𝑆𝑥𝑥 (𝜔)
∞ ∞
𝑘𝑢
𝐸(𝑥 2 ) = ∫ 𝑆𝑥𝑥 (𝜔)𝑑𝜔 = ∫ 𝑑𝜔
−∞ −∞ 𝜋(𝜔 2+ 𝑢2 )
𝑘 𝜔 𝜔=∞ 𝑘 𝜋 𝜋
𝐸(𝑥 2 ) = [arctan ] = ( + )=𝑘#
𝜋 𝑢 𝜔=−∞ 𝜋 2 2
d. Find the coefficient correlation between 𝑋(10) and 𝑋(15)
𝐸[𝑋(10)𝑋(15)] − 𝜇𝑋(10) 𝜇𝑋(15)
𝜌𝑋(10)𝑋(15) =
𝜎𝑋(10) 𝜎𝑋(15)
𝐸[𝑋(10). 𝑋(15)] = 𝑅𝑥𝑥 (5) = 𝑘 𝑒 −5𝑢
𝜇𝑋(10) = 𝜇𝑋(15) = 𝜇𝑋
𝜎𝑋(10) = 𝜎𝑋(15) = 𝜎𝑋
𝑘 𝑒 −5𝑢 − 𝜇𝑋2
𝜌𝑋(10)𝑋(15) =
𝜎𝑋2
𝑘 𝑒 −5𝑢 − 𝑅𝑥𝑥 (0) + 𝜎𝑋2 𝑘 (𝑒 −5𝑢 − 1) + 𝜎𝑋2
𝜌𝑋(10)𝑋(15) = = #
𝜎𝑋2 𝜎𝑋2

You might also like