arxiv:math/ v2 [math.oc] 8 Nov 2002

Transcription

1 arxiv:math/6121v2 [math.oc] 8 Nov 22 Matching linear systems and the ball and beam F. Andreev 13 D. Auckly 14 L. Kapitanski 124 S. Gosavi 15 W. White 15 A. Kelkar 16 3 Department of Mathematics Western Illinois University Macomb IL USA 4 Department of Mathematics Kansas State University Manhattan KS 6656 USA 5 Department of Mechanical and Nuclear Engineering Kansas State University Manhattan KS 6656 USA 6 Department of Mechanical Engineering Iowa State University Ames IA 511 USA Abstract A recent approach to the control of underactuated systems is to look for control laws which will induce some specified structure on the closed loop system. In this paper we describe one matching condition and an approach for finding all control laws that fit the condition. After an analysis of the resulting control laws for linear systems we present the results from an experiment on a nonlinear ball and beam system. 1 Underactuated systems and the matching condition Over the past five years several researchers have proposed nonlinear control laws for which the closed loop system assumes some special form see the controlled Lagrangian method of [8 9 1] the generalized matching conditions of [ ] the interconnection and damping assignment passivity based control of [7] the λ-method of [6 5] and the references therein. In this paper we describe the implementation of the λ-method of [6] on a ball and beam system. For the readers convenience we start with the statement of the main theorem on λ-method matching control laws Theorem 1. We also present an indicial derivation of the main equations. We then prove a new theorem showing that the family 1 Supported in part by NSF grant CMS Supported in part by NSF grant DMS of matching control laws of any linear time invariant system contains all linear state feedback control laws Theorem 2. We next present the general solution of the matching equations for the Quanser ball and beam system. Note that this system is different from the system analyzed by Hamberg [11]. As always the general solution contains several free functional parameters that may be used as tuning parameters. We chose these arbitrary functions in order to have a fair comparison with the manifacturer s linear control law. Our laboratory tests confirm the predicted stabilization. This was our first experimental test of the λ-method. We later tested this method on an inverted pendulum cart [3]. Consider a system of the form g rj ẍ j + [jk r]ẋ j ẋ k + C r + V x r = u r 1 r = 1...n where g ij denotes the mass-matrix C r the dissipation V the potential energy [i j k] the Christoffel symbol of the first kind [jki] = 1 gij 2 q k + g ki q j g jk q i 2 and u r is the applied actuation. To encode the fact that some degrees of freedom are unactuated the applied forces and/or torques are restricted to satisfy Pj igjk u k = where Pj i is a g-orthogonal projection. The matching conditions come from this restriction together with the requirement that the closed loop 1

2 system takes the form ĝ rj ẍ j + [jk r]ẋ j ẋ k +Ĉr + V = r = 1...n xr for some choice of ĝ Ĉ and V. The matching conditions read Pk r Γ k ij Γ k ij = Pk g r ki C i ĝ ki Ĉ i = Pk r g ki V ĝki V 3 q i q i = whereγ k ij isthechristoffelsymbolofthesecondkind Γ k ij = gkl [ijl]. 4 If the matching conditions 3 hold the control law will be given by u r = g rk Γ k ij Γ k ij q i q j + C r Ĉr +g rk g ki V ĝki V q i q i. 5 The motivation for this method is that Ĥ = 1 2ĝij q i q j + V is a natural candidate for a Lyapunov function because d dtĥ = ĝ ijĉ i q j. Following [6] introduce new variables λ k i = g ijĝ jk. We have Theorem 1 The functions ĝ ij V and Ĉ satisfy 3 in a neighborhood of x if and only if P r k Pk g r ki C i ĝ ki Ĉ i = g ki V ĝki V q i q i = and the following conditions hold. First there exists a hypersurface containing x and transverse to each of the vectorfields λ l i Pi j / xl on which ĝ ij is invertible and symmetric and satisfies g ki P k l = λj k Pk l ĝji. Second λ i j Pj k and ĝ ij satisfy Pk s λ l Pr r t g ls q j +[ljs]λl r [rji]λi s λ i s +g ir q j +[ijr]λi s [sjl]λ l r = 6 λ l rpt r ĝ nm λ l r q l +ĝ Pr t ln q m +ĝ λ l r Pr t lm q n = Pt l partialg nm Pt l P q l ++g ln q mg t l lm qn. 7 Although the proof of this proposition may be found in [6] [4] and [5] for convenience we include an indicial notation derivation of equations 6 and 7. Substitute equations2 4 forboth Γ k ij and Γ k ij into the first of equations 3 and multiply the result by the scalar 2 to obtain: Pkĝ r kl ĝ ij q l Pr kĝ kl ĝ li q j Pr kĝ kl ĝ jl q i = P r k gkl g ij q l Pr k gkl g li q j Pr k gkl g jl q i. Multiply by g rt and use that P is self-adjoint i.e. Pi kg kj = g ik Pj k to get Use P r t λl r ĝ li Pt r ĝ ij ĝ li λl r q l Pr t λl r q j Pr t λl r = P r t g ij q r Pr t q = Pr j t λl rĝli g ri q j Pr t P q ĝ r j t λl r li q j ĝ jl q i g jr q i. and 8 λ l rĝli = g ri 9 in 8 to obtain 7. To derive 6 first differentiate 9 with respect to q j to get λ l ĝ li r q j = g ri q j ĝ li λ l r q j. 1 Substitute equation 1 into equation8 and obtain P r t = P r t ĝ li λ l r ĝ ij q l q j +ĝ λ l r lj q i +λl r gri q j + g rj q i g ri q j g jr q i + g ij q r 11. 2

3 Multiply by Pk sλi s use 9 and 1 to obtain λ l r g ij g ls P s k Pr t = P s k Pr t g js q l λi s q j +λl r ĝ ij λ l λ i s r q l λi sĝlj λ l r q i q r. 12 Finally to obtain 6 add to equation 12 an equation obtained from 12 by interchanging k and t r and s l and i. 2 Matching and constant coefficient linear systems In this section we prove that for linear time invariant systems any linear full state feedback control law is a solution to the matching equations. Theorem 2 When applied to linear timeindependent systems the family of matching control laws contains all linear state feedback laws. Choose coordinates q i so that the desired equilibrium is at the origin V = V ij q i q j + v k q k and C i = C ij q j where g ij V ij v k C ij and Pk r are constant and Pk r has rank n u. Clearly ĝ ij V = V ij q i q j and Ĉi = Ĉij q j is a solution to the matching equations when ĝ ij Vij and Ĉij are constant provided ĝ ij and V ij are symmetric Pk g r ki V ij ĝ ki Vij = and Pk g r ki C ij ĝ ki Ĉ ij =. Let u k = v k +a ki q i + b ki q i be an arbitrary linear control law satisfying Pk rgkl u l =. Comparison with equation 3 gives g rk g ki V ij ĝ ki Vij = a rj and Thus and g rk g ki C ij ĝ ki Ĉ ij = b rj. V lj = ĝ lp g pr V rj a rj Ĉ lj = ĝ lp g pr C rj b rj. It remains to check that we can find a symmetric nondegenerate matrix ĝ ki so that the resulting V lj is also symmetric. The symmetry of V lj will follow if we have ĝ lp g pr V rj a rj ĝ jp g pr V rl a rl = and therefore we need to find a symmetric nondegenerate matrix ĝ lp satisfying this equation. The existence of such matrix is guaranteed by the following simple observation. Lemma 1 Given any real n n matrix R there is a nondegenerate symmetric matrix X so that RX X T R T =. Indeed setting X = QYQ T results in the following equation for Y: Q 1 RQY Y T Q 1 RQ T =. Hence without loss of generality we may assume that Q 1 RQ is a real Jordan block see [14] λ 1... λ or a b 1... b a 1 a b 1... b a In each case Y = solves the 1... equation. NotethattheresultofLemma1istrueformatrices with coefficients in any field. This is proved in [15]. 3 Example: The Ball and Beam In order to demonstrate the approach described above we have implemented one of the control laws from the family of control laws described in the first 3

4 Table 1: The physical parameters of the system. l b = length of the beam =.43m I B = 2 5 m BrB 2 = ball inertia = Kg m 2 l l = length of the link =.11m I b = inertia of the beam =.1Kg m 2 r g = radius of the gear =.3m I s = effective servo inertia =.2Kg m 2 r B = radius of the ball =.1m g = gravitational acceleration = 9.8m/s 2 m B = mass of the ball =.7Kg s = desired equilibrium position =.22m m b = mass of the beam =.15Kg c = inherent servo dissipation = Kg m 2 /s m l = mass of the link =.1Kg s θ α and the dissipation is C 1 = C 2 = c θ. After rescaling we get 1 cosα a 2 1 cosθ 2 +sinα+a 1 a 2 sinθ 2 = a 2 1 T = 1 2 ṡ a 4+a 3 +5/2s 2 α θ 2 θ 2 +α θṡ θ C 1 = C 2 = Figure 1: The ball and beam system section on a ball and beam system Figure 1 this system is commercially available from Quanser Consulting Ontario Canada. The s coordinate is unactuated the θ coordinate is actuated by the servo and the objective is to bring the ball to the center of the beam. The physical parameters of the system are given in Table 1. One can express α as a concrete function of θ from the kinematic relation l b 1 cosα r g 1 cosθ 2 +l b sinα+l l r g sinθ 2 = l 2 l. The kinetic energy of the system is T = 1 2 m bs 2 α I B α+ 1 r B ṡ I b α I s θ 2. The potential energy is V = 1 2 m l g r g sinθ+ 1 2 m b +m l g l b sinα +m B g ssinα V = a 5 sinθ+s+a 6 sinαθ C 1 = and C 2 = a 7 θ where the ak are the dimensionless parameters a 1 = l l l b a 2 = r g l b a 3 = I b +I B I B a 5 = m lr g a 6 = l bm b +m l 2m B r B 2m B r B c a 7 = 2 rb 3 g. m B a 4 = I s I B The notation is used to denote a derivative of a function of one variable. For general underactuated systems the use of the powerful λ-method to solve the matching equations is discussed in [6 5]. For systems with two degrees of freedom the λ-method produces the general solution to the matching equations in an explicit form [4]. When applied to the ball and beam system the explicit family of control laws is given by equation 5 with the following expressions for ĝ V and Ĉ where ĝ 11 = ψ 2 α hysθ+1 dϕ µ 1 ϕψ2 ϕ ĝ 12 = 1 µ g 11 σĝ 11 ĝ 22 = 1 µ g 12 σĝ 12 4

5 where Vsθ =wy+5y +s 5 sinϕ µ 1 ϕψϕ ϕ µ 1 sinϕ dϕ ϕψϕ ψτdτ dϕ Ĉ 1 = g 1i ĝ i1 1 C 1 g 1j ĝ j2 Ĉ 2 µsθ = µ 1 αθ 5sg 12 σsθ = µ 1 α 1 5s µ 1 α y = ψαs s + ψτdτ µ 1κ µ dκ}. 1 κ Here hy wy µ 1 α are arbitrary functions of one variable and Ĉ2 is an arbitraryfunction which is odd in velocities. 4 Experimental Results Our experiments were conducted on the Quanser ball andbeamsystem. Thecontrolsignalisavoltagesuppliedtothe servoandthesensedoutputofthe system is s and θ sampled at 3 Hz. A Quanser MULTIQ R data acquisition card is used for the analog signal input and output. The velocities are computed via numerical differentiation using the forward difference algorithm. The control law produces a voltage signal and is supplied through the D/A converterto the DC servomotor via an amplifier. The relation between the control voltage v in and the torque u = u 2 in equation 5 is K 2 m N2 g θ where R m = armature resistance = 2.6 Ω N g = gear ratio = 7.5 K m = motor torque constant =.767 Volt sec. Any stabilizing linear control law for this system is specified by four constants. The nonlinear control laws in our family are specified by the four arbitrary functions: µ 1 α hy wy and Ĉ2sθṡ θ. We chose µ 1 α = 1.849exp4.7845sinα hy = wy =.23y 2 Ĉ 2 sθṡ θ = ĝ 12 1+ṡ 2 +1 θ 2 µṡ+σ θ. These functions produce the control law u in rescaledunits. Thevaluesoftheconstantsa 1 through a 7 are as follows a 1 =.2547 a 5 =.1889 a 2 =.588 a 6 = 42 a 3 = a 7 = a 4 = The final control signal is obtained by converting back into MKS units and using the formula in the preceding paragraph to get the input voltage. These choices were made from the following considerations. The form of the function µ 1 was chosen to simplify the integralsin the expressions for y ψ and ĝ 11. The form ofĉ2 waschosento ensurethat Ĉ1ṡ+Ĉ2 θ would be positive for Ĥ to be a Lyapunov function. Finally the coefficients in these functions were chosen so that the linearization of the nonlinear control law would agree with the linear control law provided by the manufacturer. Extensive numerical simulations done using Matlab R confirm that the nonlinear control law stabilizes the system. The linear control law appears to stabilize the system for a wider range of initial conditions than the nonlinear control law. This is an empirical observation not a mathematical fact. Finding an adequate mathematical framework to compare different control laws is a very interesting unresolved problem see[4]. Usually given two locally stabilizing control laws there exist initial conditions stabilized by one but not by the other. For example one set of physically unrealistic initial conditions with a large angular velocity θ = 3.6 or 158 rad/sec in physical units is stabilized by our nonlinear control law but not by the linear one. We have implemented the nonlinear control law in the laboratory. The laboratory tests confirm the predicted behavior of the nonlinear controller. Figures 2 and 3 show a comparison of the time histories of the ball position s and angular displacement θ for the 5

6 Linear control law Nonlinear control law 1 Linear control law Nonlinear control law Position m.1 Angle rad Time sec Time sec Figure 2: Ball position response Figure 3: Angular displacement response linear and nonlinear control laws. In both cases the control signal reached the saturation limit for a short duration during the initial rise of the response. The difference in the steady-state values of the responses is attributed to a lack of sensitivity of the resistive strip used to measure the ball position. 5 Conclusions The λ-method produces explicit infinite-dimensional families of control laws and simultaneously provides a natural candidate for a Lyapunov function. When this method is applied to linear time-invariant systems the resulting family contains all linear state feedback control laws Proposition 2. In this paper we also present the results of the first implementation of a λ-method matching control law on a concrete physical device the ball and beam system. The experimental results agree with theoretical predictions and numerical simulations. In our experiments we observe that the linear control law performs better than our nonlinear control law for the ball and beam system. However in a later experiment with an inverted pendulum cart [3] we found that a properly tuned λ-method matching control law performed better than the corresponding linear one. At the moment it is not known for which systems matching control laws will perform better. This is an important problem that must be resolved. References [1] F. Andreev D. Auckly S. Gosavi L. Kapitanski A. Kelkar and W. White Matching linear systems and the ball and beam [2] F. Andreev D. Auckly L. Kapitanski A. Kelkar W. White Matching control laws for a ball and beam system Proc. IFAC Workshop on Lagrangian and Hamiltonian Methods for Nonlinear Control Princeton ; [3] F. Andreev D. Auckly L. Kapitanski A. Kelkar W. White Matching and digital control implementation for underactuated systems Proceedings of the American Control Conference Chicago IL [4] D. Auckly L. Kapitanski Mathematical problems in the control of underactuated systems CRM Proceedings and Lecture Notes [5] D. Auckly L. Kapitanski On the λ-equations for matching control laws submitted 6

7 [6] D. Auckly L. Kapitanski and W. White Control of nonlinear underactuated systems Communications on Pure Appl. Math [7] G. Blankenstein R. Ortega and A. J. van der Schaft The matching conditions of controlled Lagrangians and interconnection and damping assignment passivity based control Preprint 21 [8] A. Bloch N. Leonard and J. Marsden Stabilization of mechanical systems using controlled Lagrangians Proc. IEEE Conference on Decision and Control San Diego CA [9] A. Bloch D. Chang N. Leonard and J. E. Marsden Controlled Lagrangians and the stabilization of mechanical systems II: Potential shaping Trans IEEE on Auto. Control [1] A. Bloch N. Leonard and J. Marsden Controlled Lagrangians and a stabilization of mechanical systems I: The first matching theorem IEEE Trans. Automat. Control [11] J. Hamberg General matching conditions in the theory of controlled Lagrangians in Proc. IEEE Conference on Decision and Control Phoenix AZ 1999 [12] J. Hamberg Controlled Lagrangians symmetries and conditions for strong matching Proc. IFAC Workshop on Lagrangian and Hamiltonian methods for nonlinear control Princeton NJ [13] J. Hamberg Simplified conditions for matching and for generalized matching in the theory of controlled Lagrangians Proc. ACC Chicago Illinois [14] R. Horn and C. Johnson Matrix analysis Cambridge University Press Cambridge [15] O. Taussky and H. Zassenhaus On the similarity transformation between a matrix and its transpose Pacific J. Math