Describing function analysis

Limit cycles are oscillations of 2

fixed amplitude and period 0

-2

Undesirable in many cases -4

0 20 40 60 80 100

Prediction of existence of LC

Phase portraits Describing function analysis

Index Theorem

Bendixon Theorem
Application domain
Nonlinear
element Linear element

Almost linear systems

𝑤=𝑓 ( 𝑥 ) 𝐺(𝑠)
Designed using linear control 𝑟 ( 𝑡 ) =0 𝑥(𝑡) 𝑤(𝑡) 𝑦 (𝑡)
theory and then analyzed for +
hard nonlinearities. -
Genuinely nonlinear systems
 Application domain
Nonlinear
element Linear element

Almost linear systems

𝑤=𝑓 ( 𝑥 ) 𝐺(𝑠)
Designed using linear control 𝑟 ( 𝑡 ) =0 𝑥(𝑡) 𝑤(𝑡) 𝑦 (𝑡)
theory and then analyzed for +
hard nonlinearities. -
Genuinely nonlinear systems
𝑢(𝑡) ~
𝑢(𝑡 )
0
~ + 𝐺𝑐 𝐺 𝑠 𝐺𝑝 (𝑠)
𝑥(𝑡) 𝑢(𝑡) 𝑤(𝑡 ) 𝑢(𝑡 )
𝐺𝑐 (𝑠) 𝐺𝑝 (𝑠)
𝑟 ( 𝑡 ) =0 𝑦 (𝑡 ) -

+
𝑟 ( 𝑡 ) =0
-
+
𝐺 𝑠 (𝑠) 𝑢(𝑡)
~
𝑢(𝑡 )
𝐺𝑝 (𝑠)
𝑦 (𝑡)
𝐺 𝑠 (𝑠) - 𝑥(𝑡)
𝐺𝑐 (𝑠)
 Application domain
Nonlinear
element Linear element

Almost linear systems

𝑤=𝑓 ( 𝑥 ) 𝐺(𝑠)
Designed using linear control 𝑟 ( 𝑡 ) =0 𝑥(𝑡) 𝑤(𝑡) 𝑦 (𝑡)
theory and then analyzed for +
hard nonlinearities. -
Genuinely nonlinear
systems
Van der Pol Equation
𝑠 2 𝑋 ( 𝑠 ) −𝛼 𝑠𝑋 ( 𝑠 ) + 𝑋 ( 𝑠 )=𝛼 𝑊 ( 𝑠)
𝑥+𝛼
¨ ( 𝑥 − 1 ) 𝑥+
2
˙ 𝑥= 0
( 𝑠 2 −𝛼 𝑠+1 ) 𝑋 ( 𝑠 ) =𝛼 𝑊 ( 𝑠) 𝑤(𝑡) 𝛼
𝑥(𝑡)
2 𝑤= 𝑓 ( 𝑥)
𝑥¨ −𝛼 𝑥+
˙ 𝑥=− 𝛼 𝑥 𝑥˙ + ( 𝑠 2 − 𝛼 𝑠+1 )
− 𝑥 (𝑡)
𝑋 ( 𝑠) 𝛼 -
𝑥¨ −𝛼 𝑥+𝑥=𝛼
˙ 𝑤 = 2
𝑊 (𝑠 )   ( 𝑠 − 𝛼 𝑠 +1 )
𝑤=− 𝑥 2 𝑥˙
Basic Idea
Nonlinear
element Linear element
Assume a periodic solution
Approximate the nonlinearity
𝑤=𝑓 ( 𝑥 ) 𝐺(𝑠)
𝑟 ( 𝑡 ) =0 𝑥(𝑡) 𝑤(𝑡) 𝑦 (𝑡)
with a quasi linear function
+
Check whether the assumed -
solution satisfies the system
 Nonlinear
element Linear element

Basic Idea cos2 𝜔 𝑡=cos ( 𝜔 𝑡+𝜔 𝑡 )

𝑤=𝑓 ( 𝑥 ) 𝐺(𝑠)
2 𝑟 ( 2𝑡 ) =0 𝑥(𝑡) 𝑤(𝑡) 𝑦 (𝑡)
 Assume a periodic solution ¿ cos 𝜔 𝑡 −sin 𝜔 𝑡
+
¿ 1− sin2 𝜔 𝑡 − sin2 𝜔- 𝑡
 Approximate the nonlinearity with a quasi linear function
¿ 1− 2 sin2 𝜔 𝑡
 Check whether the assumed solution satisfies the system
2
Van der Pol Equation ⇒2 sin 𝜔 𝑡=1− cos2 𝜔 𝑡
− 𝑥 (𝑡) 𝑤(𝑡) 𝑥(𝑡)
¿ cos ( 2 𝜔 𝑡+𝜔
𝑤= 𝑓 ( 𝑡
𝑥) ) +cos ( 2 𝛼 𝑡 −𝜔𝑡 )
𝜔
2
𝑤 =− 𝑥 𝑥
˙ + ( 𝑠 2 − 𝛼 𝑠+1 )
( 𝑥 − 1 ) 𝑥+
¿cos2𝜔𝑡cos𝜔𝑡−sin2𝜔𝑡 sin𝜔𝑡
2
𝑥+𝛼
¨ ˙ 𝑥= 0 -
¿ − ( 𝐴 sin 𝜔 𝑡 ) ( 𝐴 𝜔 cos 𝜔 𝑡 )
2 2

𝑋 ( 𝑠) 𝛼
𝐺 ( 𝑠 )= = 2
𝑊 (𝑠)   ( 𝑠 − 𝛼 𝑠+ 1 )
(
¿ − 𝐴2
1 − cos 2 𝜔 𝑡
2 )( 𝐴 𝜔 cos 𝜔 𝑡 )

𝑤 = 𝑓 ( 𝑥 )=− 𝑥 2 𝑥˙
𝐴3 𝜔
¿− ( cos 𝜔 𝑡 − cos 2 𝜔 𝑡 × cos 𝜔 𝑡 )
2
Assume

𝑥 ( 𝑡 ) = 𝐴sin 𝜔 𝑡
¿−
𝐴3 𝜔
2 (
cos 𝜔 𝑡 −
cos 3 𝜔 𝑡 +cos 𝜔 𝑡
2 )
 Nonlinear
element Linear element

Basic Idea
𝑤=𝑓 ( 𝑥 ) 𝐺(𝑠)
𝑟 ( 𝑡 ) =0 𝑥(𝑡) 𝑤(𝑡) 𝑦 (𝑡)
 Assume a periodic solution
+
 Approximate the nonlinearity with a quasi linear function -
 Check whether the assumed solution satisfies the system

Van der Pol Equation

− 𝑥 (𝑡) 𝑤(𝑡) 𝛼
𝑥(𝑡)
𝑤= 𝑓 ( 𝑥)
𝑤 =− 𝑥 2 𝑥
˙ + ( 𝑠 2 − 𝛼 𝑠+1 )
𝑥+𝛼
¨ ( 𝑥 − 1 ) 𝑥+
2
˙ 𝑥= 0 -
¿ − ( 𝐴 sin 𝜔 𝑡 ) ( 𝐴 𝜔 cos 𝜔 𝑡 )
2 2

𝑋 ( 𝑠) 𝛼
𝐺 ( 𝑠 )= = 2
𝑊 (𝑠)   ( 𝑠 − 𝛼 𝑠+ 1 )
(
¿ − 𝐴2
1 − cos 2 𝜔 𝑡
2 )( 𝐴 𝜔 cos 𝜔 𝑡 )
𝑤=−
𝐴3 𝜔
( cos 𝜔 𝑡 − cos 3 𝜔 𝑡 )
4
𝑤 = 𝑓 ( 𝑥 )=− 𝑥 2 𝑥˙
𝐴3 𝜔 𝐴3 𝜔
¿− ( cos 𝜔 𝑡 − cos 2 𝜔 𝑡 × cos 𝜔 𝑡 ) 𝑤≈− cos 𝜔 𝑡
2 4
Assume
higher order harmonics
𝑥 ( 𝑡 ) = 𝐴sin 𝜔 𝑡
¿−
𝐴3 𝜔
2 (
cos 𝜔 𝑡 −
cos 3 𝜔 𝑡 +cos 𝜔 𝑡
2 ) are suppressed
 Nonlinear
element Linear element

Basic Idea ( 1 − 𝜔2 ) + 𝑗 ( 𝛼 𝜔 ) 2
𝛼 𝑟 ( 𝑡 ) =0 𝐴 ( 𝑗𝑥(𝑡)
𝜔)
𝑤=𝑓 ( 𝑥 ) 𝐺(𝑠)
1+ × =0 𝑤(𝑡) 𝑦 (𝑡)
 Assume a periodic solution ( 1 − 𝜔 ) + 𝑗 ( 𝛼 𝜔 ) ( 1 − 𝜔 ) − 𝑗 ( 𝛼 𝜔+ )
2 2
4
 Approximate the nonlinearity with a quasi2 linear function 2 -
( 1 − 𝜔 ) + 𝑗 ( 𝛼 𝜔 ) 𝛼 𝐴 ( 𝑗 𝜔)
 Check whether the assumed solution
1+ satisfies the system =0
2 2 4
( 1− 𝜔 ) + ( 𝛼 𝜔 ) 2

Van der Pol Equation 𝑥=𝐺 ( 𝑗 𝜔 ) 𝑤=𝐺 ( 𝑗 𝜔 ) 2𝑁 ( 𝐴 , 𝜔 ) ( − 𝑥 ) ( 𝛼 𝜔) 𝛼 𝐴 2 (𝜔)

( 1 − 𝜔2 ) 𝛼 𝐴 ( 𝑗 𝜔) 1− − 𝑥 (𝑡) 𝑤(𝑡) 𝛼=0 𝑥(𝑡)
=0 2 2 𝑤= 𝑓 ( 𝑥)
( 1 − 𝜔 ) +( 𝛼 𝜔 ) 2 4
2
( 1 − 𝜔 ) + ( 𝛼𝛼𝜔 )
2 2 4 2
𝐴 ( 𝑗 𝜔) + 2
( 𝑠 − 𝛼 𝑠+1 )
𝑥+𝛼
¨ ( 𝑥 − 1 ) 𝑥+ 2
˙ 𝑥= 0 𝑥= (− 𝑥 ) -( 𝛼 ) 𝛼 𝐴 2
( 𝑗 𝜔 ) 2 −𝛼 ( 𝑗 𝜔 ) +1 4
𝜔=0 , 𝜔=1 1− =0 4 𝛼2 − 𝐴 2 𝛼2=0
𝐴 𝜔 3
(𝛼 ) 2
4
𝑤≈− cos 𝜔 𝑡

≈
2
4 𝛼 𝐴 ( 𝑗 𝜔) 𝐴=2
1+ =0
𝐴2 𝑑 ( 𝑗 𝜔 )2 −𝛼 ( 𝑗 𝜔 ) +1 4 − 𝑥 (𝑡) 𝐴2 𝑤(𝑡) 𝛼
𝑥(𝑡)
¿ ( − 𝐴sin 𝜔 𝑡 ) + 4
𝑠
( 𝑠 2 − 𝛼 𝑠+1 )
4 𝑑𝑡 2
𝛼 𝐴 ( 𝑗 𝜔)
1+ =0 -
2
𝐴 𝑑 ( 1 − 𝜔 ) −𝛼 ( 𝑗 𝜔 )
2
4
¿ ( − 𝑥 (𝑡 ) )
4 𝑑𝑡
⇒ 𝐴=2 , 𝜔=1
𝑤=𝑁 ( 𝐴 , 𝜔 ) ( − 𝑥(𝑡 ) ) Describing function: May depend
upon amplitude and frequency of
the input
 Nonlinear
element Linear element

Basic Idea
𝑤=𝑓 ( 𝑥 ) 𝐺(𝑠)
𝑟 ( 𝑡 ) =0 𝑥(𝑡) 𝑤(𝑡) 𝑦 (𝑡)
 Assume a periodic solution
+
 Approximate the nonlinearity with a quasi linear function -
 Check whether the assumed solution satisfies the system

Van der Pol Equation 𝑥=𝐺 ( 𝑗 𝜔 ) 𝑤=𝐺 ( 𝑗 𝜔 ) 𝑁 ( 𝐴 , 𝜔 ) ( − 𝑥 )

− 𝑥 (𝑡) 𝑤(𝑡) 𝛼
𝑥(𝑡)
𝑤= 𝑓 ( 𝑥)
𝛼 𝐴 2 ( 𝑗 𝜔) + ( 𝑠 2 − 𝛼 𝑠+1 )
𝑥+𝛼
¨ ( 𝑥 − 1 ) 𝑥+
2
˙ 𝑥= 0 𝑥= (− 𝑥 ) -
( 𝑗 𝜔 ) 2 −𝛼 ( 𝑗 𝜔 ) +1 4
𝐴3 𝜔
𝑤≈− cos 𝜔 𝑡

≈
4 𝛼 𝐴 2 ( 𝑗 𝜔)
1+ =0
𝐴2 𝑑 ( 𝑗 𝜔 )2 −𝛼 ( 𝑗 𝜔 ) +1 4 − 𝑥 (𝑡) 𝐴2 𝑤(𝑡) 𝛼
𝑥(𝑡)
¿ ( − 𝐴sin 𝜔 𝑡 ) + 4
𝑠
( 𝑠 2 − 𝛼 𝑠+1 )
4 𝑑𝑡 2
𝛼 𝐴 ( 𝑗 𝜔)
1+ =0 -
2
𝐴 𝑑 ( 1 − 𝜔 ) −𝛼 ( 𝑗 𝜔 )
2
4
¿ ( − 𝑥 (𝑡 ) )
4 𝑑𝑡
⇒ 𝐴=2 , 𝜔=1
𝑤=𝑁 ( 𝐴 , 𝜔 ) ( − 𝑥(𝑡 ) ) Describing function: May depend
It is more likely that a LC exists in upon amplitude and frequency of
the nonlinear system the input
Find describing function for the nonlinearity

Solve 1+𝐺 ( 𝑗 𝜔 ) 𝑁 ( 𝐴 , 𝜔 )=0 to find and