Lecture A3 : State space models for continuous-time systems Jan Swevers July 2006 Aim of this lecture : Teach you the procedure to derive a state space model starting from the system equations: • using the natural state variables approach • using the direct simulation diagram 0-0 Lecture A3 : State space models for continuous-time systems Contents of this lecture • The state space model of a continuous-time system • The state of a system • How to derive a state space model : – using the natural state variables – using the direct simulation diagram • Example Control Theory [H04X3a] 1 Lecture A3 : State space models for continuous-time systems The state space model of a continuous-time system • It is the description of the system dynamic as a set of first order differential equations. • They are derived by introducing a set of state variables. • The number of state variables is (in most cases) equal to the order of the differential equation. • ẋ = Fx + Gu : state equation y = Hx + Ju : input/output − equation Control Theory [H04X3a] 2 Lecture A3 : State space models for continuous-time systems Control Theory [H04X3a] 3 Lecture A3 : State space models for continuous-time systems The state of a system • If we know the state of a system at a certain time instance and its future input, we can calculate its future output unambiguously. The state of a system is the minimal information that is required of its past history/trajectory to predict its future. How the system arrived in this state is not relevant. • Examples : – the voltage over a capacitor, – the current through an inductance, – the force through a spring, – the velocity of a mass. Control Theory [H04X3a] 4 Lecture A3 : State space models for continuous-time systems • The choice of the state variables in not unique. • We can choose the system variables that are responsible for the energy storage in the system: the natural state variables. • The number of state variables is equal to the number of independent energy storing elements in the system. For certain systems, called degenerated systems, this can be different from the number of energy storing elements in the system. In such cases, an arbitrary specification of the values of the variables that determine the energy storage in these elements in not possible because of the geometry of the system (interconnection of certain elements). Control Theory [H04X3a] 5 Lecture A3 : State space models for continuous-time systems How to determine the state space equations using the natural state variables approach Example : An electro-mechanical system: Control Theory [H04X3a] 6 Lecture A3 : State space models for continuous-time systems 7 • Loop equation for the electrical part of the system: di u(t) = Ri + L + eb , with eb = Kω dt • Node equation for node 4 yields: dω I + Cω + T = 0, with T = −Ki dt • Natural state variables: • This yields: ẋ1 ẋ2 = y −R L +K I −K L x1 x2 = i ω x1 + − CI x2 h i x 1 + 0 . u(t) = 0 1 x2 Control Theory [H04X3a] 1/L 0 u(t) Lecture A3 : State space models for continuous-time systems How to determine the state space equations from the direct simulation diagram Example • Consider the following system: ÿ(t) + 7ẏ(t) + 12y(t) = u̇(t) + 2u(t) • Introduce the following operator s = d/dt: (s2 + 7s + 12)y(t) = (s + 2)u(t) of B(s) s+2 u(t) = 2 u(t) y(t) = G(s)u(t) = A(s) s + 7s + 12 Control Theory [H04X3a] 8 Lecture A3 : State space models for continuous-time systems 9 • Remark that: d 1 s= → = dt s Z t dτ −∞ • Introducing an auxiliary variable y1 (t): µ ¶ B(s) 1 y(t) = u(t) = B(s) u(t) = B(s)y1 (t) A(s) A(s) and consequently : A(s)y1 (t) = (s2 + 7s + 12)y1 (t) = ÿ1 (t) + 7ẏ1 (t) + 12y1 (t) = u(t) .. . y1(t) u(t) + - - y1(t) 1/s y1(t) 1/s 7 12 Control Theory [H04X3a] Lecture A3 : State space models for continuous-time systems 10 • This yields : y(t) = B(s)y1 (t) = (s + 2)y1 (t) = ẏ1 (t) + 2y1 (t) 1 .. . y1(t) u(t) + - - y1(t) 1/s + y1(t) 1/s 2 7 12 Control Theory [H04X3a] + + y(t) Lecture A3 : State space models for continuous-time systems 11 • Introducing the state variables x1 and x2 : the outputs of both integrator elements: 1 u(t) + - 1/s - x1(t) x2(t) 1/s 2 + + + 7 12 • This yields the following state space description: ẋ1 (t) = −7x1 (t) − 12x2 (t) + u(t) ẋ2 (t) = x1 (t) y(t) = x1 (t) + 2x2 (t) Control Theory [H04X3a] y(t) Lecture A3 : State space models for continuous-time systems • This yields the control canonical form of the state space model. −7 −12 1 , G = F = 1 0 0 i h H = 1 2 , J=0 Control Theory [H04X3a] 12 Lecture A3 : State space models for continuous-time systems 13 General form for the control canonical state space model : B(s) b1 sn−1 + b2 sn−2 + . . . + bn u(t) y(t) = u(t) = A(s) sn + a1 sn−1 + . . . + an yields : F = G = −a1 −a2 1 .. . . . . −an 0 ... 0 0 1 0 0 1 0 .. . 0 0 ... Control Theory [H04X3a] Lecture A3 : State space models for continuous-time systems H = J = h 0 b1 b2 . . . bn 14 i In Matlab : B = [0 1 2]; A = [1 7 12]; [F,G,H,J]=tf2ss(B,A); Control Theory [H04X3a] Lecture A3 : State space models for continuous-time systems Example: quarter car • Equations of motion (1) kw (r − x) − m1 ẍ − ks (x − y) − b(ẋ − ẏ) = 0 (2) ks (x − y) + b(ẋ − ẏ) − m2 ÿ = 0 Control Theory [H04X3a] 15 Lecture A3 : State space models for continuous-time systems 16 • Natural state variables x1 = x x2 = ẋ x3 = x−y x4 = ẏ • This yields the following equations: ẋ1 ẋ2 ẋ3 ẋ4 = x2 b ks kw kw = ẍ = − ẋ3 − x3 − x1 + r m1 m1 m1 m1 = x2 − x4 b ks = ÿ = ẋ3 + x3 m2 m2 Control Theory [H04X3a] Lecture A3 : State space models for continuous-time systems 17 • or ẋ1 0 ẋ − kw 2 = m1 ẋ3 0 0 ẋ4 1 0 − mb1 ks −m 1 1 0 b m2 ks m2 y = h 1 0 x1 0 b x kw 2 m1 m1 r + −1 x3 0 − mb2 0 x4 0 x1 i x 2 −1 0 x3 x4 Control Theory [H04X3a] Lecture A3 : State space models for continuous-time systems Aim of this lecture • You must be able to derive a state space model starting from the system equations: – using the natural state variables approach – using the direct simulation diagram Control Theory [H04X3a] 18
© Copyright 2025 Paperzz