Scribd
Upload
10 views

Variable Structure Control Mechanical Systems: Harry G. Kwatny

This document summarizes techniques for variable structure control of mechanical systems including stabilization, trajectory tracking, and path tracking. It provides the equations of motion for mechanical systems and outlines steps to design switching controllers to stabilize equilibrium points and track desired trajectories using concepts like Lyapunov stability and sliding mode control. Examples provided include controlling a double pendulum system and path tracking for an automobile model. Wheelset control is also discussed as an application of these concepts.

Uploaded by

Chernet Tuge
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
10 views

Variable Structure Control Mechanical Systems: Harry G. Kwatny

This document summarizes techniques for variable structure control of mechanical systems including stabilization, trajectory tracking, and path tracking. It provides the equations of motion for mechanical systems and outlines steps to design switching controllers to stabilize equilibrium points and track desired trajectories using concepts like Lyapunov stability and sliding mode control. Examples provided include controlling a double pendulum system and path tracking for an automobile model. Wheelset control is also discussed as an application of these concepts.

Uploaded by

Chernet Tuge
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 15

Variable Structure Control ~ Mechanical Systems

Harry G. Kwatny
Department of Mechanical Engineering & Mechanics Drexel University

Outline

Mechanical Systems Stabilization Trajectory Tracking Path Tracking Example: wheelset

Mechanical Systems ~ Stabilization


= V (q) p q kinematics = F ( p, q ) + G ( p, q ) u dynamics M (q) p Assumptions: V ( q ) , G ( p, q ) are square & invertible F ( p, q ) = 0 Problem: Stabilize the equilibrium point p = 0, q = 0 Step 1: Design controller for kinematics - try Lyapunov = qT QV ( q ) p L ( q ) = 1 qT Qq L
2

= qT QV ( q )V T ( q ) Qq < 0 for q 0 choose p = V T ( q ) Qq L

Mechanical Systems 2
Step 2: Design switching controller with surface s ( p, q ) = p V T ( q ) Qq T T Take L = s Rs L = s R p (V ( q ) Qq ) q q = sT RM 1 ( q ) G ( p, q ) u + sT R ( p, q ) L
1 2 T

u = U sgn s* , s* = G T M 1R ( p V T ( q ) Qq ) Another choice for control = sT R {M 1 ( q ) G ( p, q ) u + ( p, q )} L


Get by with smaller discontinuous part, Requires enough control authority to cancel terms

Choose usm = G 1 M , udis U sgn s* , s* = G T M 1R ( p V T ( q ) Qq )

Let u = usm + udis = sT R {M 1 ( q ) G ( p, q )( u + u ) + ( p, q )} L sm dis

Example: Double Pendulum


V (q) = I ,

1
1

2 1 ( m1 + m2 ) + 22 m2 + 21 2 m2 cos 2 M (q) = 2 m2 ( 2 + 1 cos 2 )

2 m2 ( 2 + 1 cos 2 ) 2 m 2 2

m1
2
2

g 1 ( m1 + m2 ) sin 1 g 2 m2 sin (1 + 2 ) + T1 F ( p, q, u ) = g 2 m2 sin (1 + 2 ) + T2

m2

Mechanical Systems ~ Path Tracking


= V (q) p q kinematics = F ( p, q ) + G ( p, q ) u dynamics M (q) p q Q ~ R , p P ~ R n Given: a path q ( ) : R R Problem: find u ( t ) such that q ( t ) q ( )

Example: Auto
q = [ X Y] p= vx vy 0 0 1 V ( ) = 0 cos sin 0 sin cos J 0 0 M = 0 m 0 0 0 m vy 2( ) a b 2 ( a 2 + b2 ) vx 0 f = mv + 4 v y 2( a b) x vx

Vs = V

m, J

vx 2a + 0 2 vx

0 2 0 F 0 0 F

Fl
Fr

Auto-Path following
We can control v x with F , so set v x = const and drop eq = v y + d + ( t ) e = 2 ( a b) J vy vx 2 ( a 2 + b2 ) vy vx 2 ( a b)

vx

+ 2a + 2

y = mv x + 4 mv

vx

Step 1: Choose =

1 = e in sliding (vy + e) e d 1 Step 2: Choose s = ( v y + e ) d 2 = 2 dJ + am 1 ( v + e ) + D V=1 s V y 2 Jmd d = sgn s

Gradient Tracking

Potential Field: U ( q ) , Gradient: F ( q ) = = v ( q ) F ( q ) q F (q )

U q

2 1.5 U 1 0.5 0 -1 -0.5 0 x 0.5 1 -1 -0.5 1 0.5 0 y

U ( x, y ) = x 2 + y 2

U ( r ) = ln ( r )
0 -0.5 U -1 -1.5 -1 -0.5 0 x 0.5 1 -1 -0.5 0 1 0.5 y

r = x2 + y2

Obstacle Avoidance
Create an obstacle security circle centered at D with radius R: r D cos R 1 = U ( r, ) = ln ln , 2 ( r 2 D cos + D 2 ) R+D r

1 U 0 -1 -2 -1 0 -0.5 x 1 2 -1
1

1 0.5 0 y

0.5

-0.5

-1 -1 -0.5 0 0.5 1 1.5 2

Wheelset ~ Setup
= v x cos x = v x sin y = x + N ( v x , ) = F Mv + K ( v x , ) = T J M min < M < M max J min < J < J max N ( v x , ) < N max K ( v x , ) < K max
vx

Wheelset ~ Sliding Mode For perfect tracking with U ( x, y ) = ln ( x + y ) we need


2 2

speed: v x = v ( x, y ) U ( x, y ) direction: = arctan y U x y , ( ) x Switching surfaces s1 = v x v , = arctan U y ( x, y ) s2 = e + e e U x y , ( ) x s2 = + arctan x y What is the motion like in sliding? Solve kinematics along with v x = v , = arctan x y

Wheelset ~ Reaching
1 2 2 s1 + s2 ) ( 2 =ss V 1 1 + s2 s2 V= FN 1 = s1 + + s1s v v v v cos sin x x y x M J + T + K v x ( y cos x sin ) 2 = s2 s2 s + 2 2 K x + y F = Fmax sgn s1 T = Tmax sgn s2

Wheelset ~ Simulations
y

y 0.2

0.03

0.02

0.01

0.1
x 0.05 0.1 0.15 0.2

x -0.2 -0.1 -0.1 0.1 0.2

x+ y v ( x, y ) = 0.1 1+ x + y In all cases: ( 0 ) = 0

-0.2

Wheelset ~ Simulations
vx 0.025 0.02
10 s2 20 15

0.015 0.01

5 t 10 20 30 40 50 -5

0.005 t 10 -0.005 20 30 40 50

-10 -15

You might also like