Lecture 13: Stability and Control I We are learning how to analyze mechanisms but what we’d like to do is make them do our bidding We want to be able to.

Slides:



Advertisements
Similar presentations
exponential functions
Advertisements

exponential functions
Ch 7.6: Complex Eigenvalues
Differential Equations
Lecture 5: Constraints I
Exponential Functions Logarithmic Functions
Physics 430: Lecture 17 Examples of Lagrange’s Equations
Lecture 7 Rolling Constraints The most common, and most important nonholonomic constraints They cannot be written in terms of the variables alone you must.
Lecture 2. A Day of Principles The principle of virtual work d’Alembert’s principle Hamilton’s principle 1 (with an example that applies ‘em all at the.
Physics 430: Lecture 24 Euler Equations Dale E. Gary NJIT Physics Department.
Lecture 19. The Method of Zs When problems get complicated numerical complexity makes computation SLOW The method of Zs speeds the computation up We will.
Signal Spaces.
Physics 430: Lecture 16 Lagrange’s Equations with Constraints
Lecture 9 Hamilton’s Equations conjugate momentum cyclic coordinates Informal derivation Applications/examples 1.
Solving Linear Equations
Key Stone Problem… Key Stone Problem… next Set 15 © 2007 Herbert I. Gross.
Physics 430: Lecture 11 Oscillations
Ch 7.8: Repeated Eigenvalues
Ch 7.9: Nonhomogeneous Linear Systems
Math 3C Practice Midterm #1 Solutions Prepared by Vince Zaccone For Campus Learning Assistance Services at UCSB.
Vectors Sections 6.6.
Dynamical Systems Analysis II: Evaluating Stability, Eigenvalues By Peter Woolf University of Michigan Michigan Chemical Process Dynamics.
Ch 7.5: Homogeneous Linear Systems with Constant Coefficients
Dynamical Systems Analysis IV: Root Locus Plots & Routh Stability
Quantum Theory of Hydrogen shrödinger's equation for hydrogen separation of variables “A facility for quotations covers the absence of original thought.”—
Mr Barton’s Maths Notes
Lecture 20: Putting it all together An actual mechanism A physical abstraction: links, springs, wheels, motors, etc. A mathematical model of the physical.
To Solve Equations which Involve Exponents.
Dividing Polynomials.
A quadratic equation is a second degree polynomial, usually written in general form: The a, b, and c terms are called the coefficients of the equation,
ME 407 Advanced Dynamics We will learn to model systems that can be viewed as collections of rigid bodies Common mechanical systems Robots Various wheeled.
Eigenvectors and Eigenvalues
CIS 540 Principles of Embedded Computation Spring Instructor: Rajeev Alur
Numerical Computations in Linear Algebra. Mathematically posed problems that are to be solved, or whose solution is to be confirmed on a digital computer.
Pre-Calculus Lesson 7-3 Solving Systems of Equations Using Gaussian Elimination.
Row rows A matrix is a rectangular array of numbers. We subscript entries to tell their location in the array Matrices are identified by their size.
1 Lecture 27. Two Big Examples The eleven link diver The Stanford Arm (Kane & Levinson)
Lecture 12: The Constraint Null Space The matrix C has M rows and N columns, M < N If we have set this up correctly, the rows of C are independent Nonholonomic.
Lecture 8: Rolling Constraints II
Lecture 14: Stability and Control II Reprise of stability from last time The idea of feedback control Remember that our analysis is limited to linear systems.
Lecture 18. Electric Motors simple motor equations and their application 1.
Lecture 23 The Spherical Bicycle II 1 The story so far: We found the constraint matrix and the null space matrix We “cheated” a little bit in assessing.
Control Engineering Lecture# 10 & th April’2008.
LURE 2009 SUMMER PROGRAM John Alford Sam Houston State University.
2.5 Implicit Differentiation
Lecture 26. Control of the Diver In order for a diver to do what he or she does the diver applies effective torques at the joints We want to find a recipe.
Chapter 7 Stability and Steady-State Error Analysis
Physics 430: Lecture 26 Lagrangian Approach Dale E. Gary NJIT Physics Department.
Physics 430: Lecture 25 Coupled Oscillations
Zero Product Property. Zero Product Property If ab = 0, then ??? If ab = 0 then either a = 0 or b = 0 (or both). If the product 0f two numbers is 0, then.
MECHATRONICS Lecture 07 Slovak University of Technology Faculty of Material Science and Technology in Trnava.
Writing and Graphing Linear Equations Linear equations can be used to represent relationships.
Boyce/DiPrima 9 th ed, Ch 7.6: Complex Eigenvalues Elementary Differential Equations and Boundary Value Problems, 9 th edition, by William E. Boyce and.
Catching up from Yesterday
Ch 9.2: Autonomous Systems and Stability In this section we draw together and expand on geometrical ideas introduced in Section 2.5 for certain first order.
Physics - Harmonic Motion We have been dealing with straight line motion or motion that is circular. There are other types of motion that must be dealt.
, Free vibration Eigenvalue equation EIGENVALUE EQUATION
1 Lecture 15: Stability and Control III — Control Philosophy of control: closed loop with feedback Ad hoc control thoughts Controllability Three link robot.
ME451 Kinematics and Dynamics of Machine Systems Dynamics of Planar Systems December 6, 2011 Equilibrium Analysis & Inverse Dynamics Analysis ME451 Wrap.
Lecture 10 Reprise and generalized forces The Lagrangian Holonomic constraints Generalized coordinates Nonholonomic constraints Euler-Lagrange equations.
1 Lecture 24: Controlling the Uncontrollable Building the Unbuildable How do we deal with this uncontrollable system? Can we extend the method of Zs to.
Lecture 22 The Spherical Bicycle 1. 2 Some relative dimensions with the wheel radius and mass as unity sphere radius 2, mass 50 fork length 4, radius.
Ch 9.6: Liapunov’s Second Method In Section 9.3 we showed how the stability of a critical point of an almost linear system can usually be determined from.
CIS 540 Principles of Embedded Computation Spring Instructor: Rajeev Alur
Boyce/DiPrima 9th ed, Ch 9.6: Liapunov’s Second Method Elementary Differential Equations and Boundary Value Problems, 9th edition, by William E. Boyce.
Mechanical Engineering at Virginia Tech
Boyce/DiPrima 10th ed, Ch 7.6: Complex Eigenvalues Elementary Differential Equations and Boundary Value Problems, 10th edition, by William E. Boyce and.
Matrices are identified by their size.
Presentation transcript:

Lecture 13: Stability and Control I We are learning how to analyze mechanisms but what we’d like to do is make them do our bidding We want to be able to control them — to make a robot trace a particular path, say A full study of this is well beyond the scope of this course we’ll just do enough to give you the flavor of the thing and to let you do some things 1

2 You might think that if we knew the forces/torques to hold a robot somewhere that we could just set them to those values and we’d be home free Not true, and the key lies in stability We want to have a system that is close to where it should be when the forces are close to where they should be Eventually we can design transient forces that will control our systems

3 Restrict the discussion (at least for now) to holonomic systems Industrial robots are holonomic I will use Hamilton’s equations in the form The system is holonomic, so there is no Lagrange multiplier term

4 Equilibrium There are equilibria for which the system moves (and we’ll look at those) but all of our equilibria have constant generalized (conjugate) momenta In many cases the equilibrium conjugate momentum is equal to zero This discussion of stability is limited to those cases, making the left hand side of the first equation equal to zero

5 It is generally possible to solve the equilibrium equations for the equilibrium (generalized) forces I’ll do this eventually using a robot example Let’s suppose we’ve done that The big question is: What happens if we don’t get the forces exactly right? Will the position of the robot, say, be close to what we want or will it be completely different?

6 In other words Is the robot stable or unstable? I need some vocabulary (all for linear stability) I’m talking about small errors, which I can address using linear stability Nonlinear stability is beyond the scope of the course stable: if we move the system away from equilibrium, the system will go back unstable: if we move the system away from equilibrium, the error will grow (initially) exponentially marginally stable: if we move the system away from equilibrium, the error will oscillate about its reference position

7 No system without damping (dissipation) can be stable without control The best you can hope for is marginal stability. How do we assess (linear) stability? We have the equations and we suppose that we have found the equilibrium forces

8 Note that  is a placeholder. It has no physical meaning; it merely marks what is “small” Let Substitute

9 The two functions need to be expanded to get their O(1) and O(  ) terms separated Plug all of this in and collect like powers of , and we can drop the O(  2 ) terms

10 The  is no longer necessary, and we have the perturbation equation for q’ The last term will be zero for the current argument

11 Repeat for the momentum equation

12 If we restrict the discussion to cases where p 0 = 0, then we have a pair of coupled homogeneous linear equations with constant coefficients These admit exponential solutions, and the real part of the exponents determine stability

13 Denote the exponents by s If Re(s) < 0 for all s, the system is asymptotically stable If Re(s) > 0 for any s, the system is unstable If Re(s) = 0 for all s, the system is marginally stable

14 ??

15 Let’s see how this plays out a simple setting 

16 If we apply the various holonomic constraints we arrive at a one DOF system

17 I like state space for this sort of problem We have two equilibria: q = 0, q = π We can cover both at the same time by using  0 to denote the equilibrium We linearize

18 We can expand the sine function, noting that sinq 0 is zero, to find

19 We seek exponential solutions

20 if q 0 = 0, the cosine is positive and the system is marginally stable if q 0 = π, the cosine is negative and the system is unstable There’s no damping, so the unforced system cannot be stable

21 We can generalize a bit... by adding a torque we can place the equilibrium anywhere This doesn’t change the perturbation analysis Q 0 does not enter that

22 The system is unstable if the equilibrium position has  in the second or third quadrants More colloquially: if the pendulum is above the horizontal We can illustrate this more physically by an artificial example

23 Add a wheel and a counterweight  m1 m1 m2m2 r z constraint

24 equilibrium is at It is stable down and unstable up Can we envision how this goes?

25  z up increases restoring torque down decreases restoring torque STABLE  z up decreases restoring torque down increases restoring torque UNSTABLE

26 ??

Inverted pendulum on a cart M m l  y 27 (y 1, z 1 )

28 Equilibrium requires sin  = 0 constraint (nonsimple holonomic) Euler-Lagrange equations (I won’t use these for stability here) New Lagrangian

29 Hamilton’s equations. Start with the Lagrangian again. As before, equilibrium requires sinq 2 = 0 assign the generalized coordinates; find the conjugate momenta The evolution equations (velocity and momentum)

30 Horizontal momentum is conserved, we can use this to reduce the problem If we start from rest If the bob starts to fall forward, the cart starts to move backwards

31 To understand stability we need to go a little further on the path to solution We can throw out the p 1 equation, reducing the problem to three equations— two velocity and one momentum We can build a three dimensional state for this problem

32 To understand stability we need to linearize this system The nonlinear state equation

33 Let me introduce a nice new trick for linearizing Expand

34 Seek exponential solutions to the perturbation problem The determinant is a polynomial in s from which we can find the possible values

35 where the bullets stand for fairly complicated terms The characteristic polynomial is a cubic, and it has only cubic and linear terms (the cubic term is unity — the leading term in these is always unity We can review this in Mathematica shortly In the present case

36 For l = 1, m = 1, M = 10 we can plot the linear term as a function of  0 and p 1 Unstable if  0 π/2

37 ??

38 What about something a little more real: the three link robot. We’ll look at this one in Mathematica as well, but we can say some things Recall that we have three generalized coordinates after applying all the holonomic constraints

39 The components of the momentum are too big to transcribe here the form is as below The linearizing matrix has the form

40 The characteristic polynomial is a quintic equation with no constant term The zero eigenvalue means that we are at best marginally stable All but the quintic and cubic terms drop out when the robot is stationary so I have three zero eigenvalues and two nonzero eigenvalues I can plot their square

41 unstable marginally stable

42 ?? On to Mathematica: the inverted pendulum and the three-link robot