Métodos de Sintonía y Autosintonía de PIDs Fraccionarios
Transcription
Métodos de Sintonía y Autosintonía de PIDs Fraccionarios
Jornadas de Ingeniería de Control Pamplona, 14 y 15 de Marzo de 2006 MÉTODOS DE SINTONÍA Y AUTOSINTONÍA DE PIDs FRACCIONARIOS Blas M. Vinagre, Concepción A. Monje *Escuela de Ingenierías Industriales. Universidad de Extremadura (Badajoz), Spain. e-mail: cmonje@unex.es; bvinagre@unex.es 1 MÉTODOS DE SINTONÍA Y AUTOSINTONÍA DE PIDs FRACCIONARIOS ÍNDICE X Cálculo Fraccionario Y Control Fraccionario Z Sintonía de controladores PID fraccionarios [ Autosintonía \ Conclusiones 2 CÁLCULO FRACCIONARIO • Definiciones (Riemann – Liouville): 3 CÁLCULO FRACCIONARIO • Definición de Gründwald – Letnikov: • Otras definiciones: Weyl (Potencial), Caputo (Condiciones iniciales interpretables). 4 CONTROL FRACCIONARIO 5 CONTROL FRACCIONARIO Acciones básicas de control Acción integral Acción derivativa 6 CONTROLADORES PID FRACCIONARIOS • Dos estructuas: • Sintonía: Encontrar los valores de los parámetros para satisfacer 5 especificaciones de diseño. ki C ( s) = k p + λ + kd s µ s • Autosintonía: Compensación atraso-adelanto. 7 SINTONÍA DE CONTROLADORES PID FRACCIONARIOS Especificaciones Phase Margin (φm) and Gain Crossover Frequency (ωcg) Arg ( F ( jωcg )) = Arg (C ( jωcg )G ( jωcg )) = −π + φm F ( jωcg ) dB = C ( jωcg )G ( jωcg ) dB = 0dB Robustness to Variations in the Gain of the Plant d ( Arg ( F ( jω )) ) =0 dω ω =ωcg Output Disturbance Rejection (Sensitivity Function) S ( jω ) dB = 1 1 + C ( jω )G ( jω ) dB ≤ BdB, ∀ω ≤ ω s rad / sec High Frequency Noise Rejection T ( jω ) dB C ( jω )G ( jω ) = ≤ AdB, ∀ω ≥ ωt rad / sec 1 + C ( jω )G ( jω ) dB 1 − S ( s) = T ( s) Complementary Functions 8 SINTONÍA DE CONTROLADORES PID FRACCIONARIOS Diseño Set of Five Nonlinear Equations with Five Unknown Parameters (kp,kd,ki,λ,µ) Nonlinear Minimization Problem F ( jωcg ) dB =0 Main Function to Minimize Arg ( F ( jωcg )) + π − φm = 0 d ( Arg ( F ( jω )) ) =0 dω ω =ωcg Constraints FMINCON (Matlab) S ( jω s ) − B dB = 0 T ( jωt ) − A dB = 0 9 SINTONÍA DE CONTROLADORES PID FRACCIONARIOS Ejemplo: Simulación First-Order Plant Plus Integrator: k 0.25 = G (s) = s (τs + 1) s ( s + 1) Design Specifications: 1) ωcg = 1rad / sec 2) φm = 48.5º deg 3) Flat Phase 4) T ( jω ) dB ≤ −20dB, ∀ω ≥ ωt = 10rad / sec 5) S ( jω ) dB ≤ −20dB, ∀ω ≤ ω s = 0.01rad / sec Fractional PIλDµ Controller: ki 2.1199 µ C ( s ) = k p + λ + k d s = 3.8159 + 0.6264 + 2.2195s 0.8090 s s 10 SINTONÍA DE CONTROLADORES PID FRACCIONARIOS Ejemplo: Simulación Control of a First-Order Plant Plus an Integrator 1) ωcg = 1rad / sec 2) φm = 48.5º deg 3) Flat Phase 11 SINTONÍA DE CONTROLADORES PID FRACCIONARIOS Ejemplo: Simulación 4) T ( jω ) dB ≤ −20dB, ∀ω ≥ ωt = 10rad / sec 5) S ( jω ) dB ≤ −20dB, ∀ω ≤ ω s = 0.01rad / sec 12 SINTONÍA DE CONTROLADORES PID FRACCIONARIOS Ejemplo: Experimento Design Specifications: G (s) = k e − Ls τs + 1 C ( s ) = 0.0469 + 0.0469 s 0.7333 + 1.4747 s 0.3146 13 SINTONÍA DE CONTROLADORES PID FRACCIONARIOS Otros métodos 14 SINTONÍA DE CONTROLADORES PID FRACCIONARIOS Otros métodos Controladores PID – Optimización (Barbosa, Machado, 2003) 15 AUTOSINTONÍA The idea: automatic tuning of a controller, generally a PID, so that an unknown plant can be controlled, fulfilling several design especifications. ki + kd s C (s) = k p + s Two parts: a) information of the plant (relay test). b) tuning of the controller with that information. There are different auto-tuning methods for conventional PID controllers, currently working on industrial environments. The more complex the controller design method, the more difficult the implementation problem (computer/PLC). 16 AUTOSINTONÍA Auto-Tuning Method for a Fractional PIλDµ Controller. C (s) = k p + ki µ + k s d sλ The introduction of the orders λ and µ allows the fulfillment of a robustness constraint without increasing the complexity of the design method (simple equations). The method uses the relay test to obtain the information of the plant to control, due to its reliability and simplicity. 17 AUTOSINTONÍA Relay Auto-Tuning Scheme: Process: G(s) Condition for Oscillation: πa 1 = arg(G ( jωc ) ) = −π + ωcθ a , G ( jωc ) = 4d N ( a ) ωc: Frequency of interest How to select the right value of θa which corresponds to ωc? Iterative Process θn = ωc − ωn −1 (θ n −1 − θ n − 2 ) + θ n −1 ωn −1 − ωn − 2 n: Curent iteration number (θ-1, θ0 ); (ω-1,ω0): Two Initial Values of the Delay and Their Corresponding Frequencies 18 AUTOSINTONÍA Método Lag Compensator with pole at the origin=PI SPECIFICATIONS OF DESIGN: Crossover Frequency ωc Phase Margin ϕm Robustness Property (flat phase) Lead Compensator: Noise Filter 19 AUTOSINTONÍA Método 1) Robustness Constraint: flat phase of the open-loop system Constant overshoot for gain variations λ λ Slope of the phase of the plant, estimated by the relay test 20 AUTOSINTONÍA Método 2) * Crossover frequency especification, ωc * Phase margin especification, ϕm Lead Compensator + ROBUSTNESS CONSTRAINT (AGAIN) 21 AUTOSINTONÍA Método LEAD COMPENSATOR (α>0) α α s +1/ λ α λs + 1 C ( s) = kc = kc x , 0 < x < 1, α ∈ ℜ + s x x s + 1 / λ λ 1 C(s) = C' (s) α ω=ωm kc x ω=ω m α (λω )2 +1 1 α m = = 2 ( xλωm ) +1 x 1− x Arg (C ' ( s )) ω =ω m = φ m = α sin −1 1+ x ω zero = 1 /λ ω pole = 1 / xλ ωm = 1 λ x •α: Flexibility in the Design Modulation of the Phase Curve •φm: Does Not Impose the Value of x 22 AUTOSINTONÍA Método LEAD COMPENSATOR (α>0) α≥αmin (α,a1,b1) αmin x= Doing Some Calculations: Lead Region (α,x,λ); kc=kss/kxα xmin: Maximum Distance Between Zero-Pole (Phase Flatness) a −1 a (a − 1) + b 2 a(a − 1) + b 2 λ= bωc 23 AUTOSINTONÍA Método 24 AUTOSINTONÍA Método • The relay test: *magnitude and phase of the plant at ω1, ωc, ω2 *estimation of the slope of the phase of the plant • PIλ(s): cancell the slope of the phase of the plant, giving the minimum lag phase possible=robustness constraint • PDµ(s): fulfills the frequency especifications of ωc and ϕm, following a robustness constraint • The parameters of the fractional PIλDµ(s) controller are obtained by simple equations that can be solved by a PLC (industrial application) • The implementation of the fractional controller: we are currently working on it 25 AUTOSINTONÍA Ejemplo: Simulación UNKNOWN PLANT: G (s) = 0.55 e −0.05 s s (0.6 s + 1) SPECIFICATIONS OF DESIGN: Robustness Constraint Crossover Frequency ωc =2.3rad /sec Phase Margin ϕm =72o RELAY TEST: Slope of the phase of the plant=0.2566 26 AUTOSINTONÍA Ejemplo: Simulación UNKNOWN PLANT: G (s) = 0.55 e −0.05 s s (0.6 s + 1) SPECIFICATIONS OF DESIGN: Robustness Constraint Crossover Frequency ωc =2.3rad /sec Phase Margin ϕm =72o RELAY TEST: Slope of the phase of the plant=0.2566 27 AUTOSINTONÍA Ejemplo: Simulación UNKNOWN PLANT: G (s) = 0.55 e −0.05 s s (0.6 s + 1) SPECIFICATIONS OF DESIGN: Robustness Constraint Crossover Frequency ωc =2.3rad /sec Phase Margin ϕm =72o RELAY TEST: Slope of the phase of the plant=0.2566 28 AUTOSINTONÍA Ejemplo: Simulación UNKNOWN PLANT: G (s) = 0.55 e −0.05 s s (0.6 s + 1) SPECIFICATIONS OF DESIGN: Robustness Constraint Crossover Frequency ωc =2.3rad /sec Phase Margin ϕm =72o 29 AUTOSINTONÍA Ejemplo: Simulación UNKNOWN PLANT: G (s) = 0.55 e −0.05 s s (0.6 s + 1) SPECIFICATIONS OF DESIGN: Robustness Constraint Crossover Frequency ωc =2.3rad /sec Phase Margin ϕm =72o 30 AUTOSINTONÍA Ejemplo: Simulación 31 AUTOSINTONÍA Ejemplo: Experimento G (s) = k s (τs + 1) 0.4348s + 1 PI ( s ) = s λ 0.8486 4.0350 s + 1 PD µ ( s ) = s + 0 . 0039 1 0.8160 32 CONCLUSIONES Y TRABAJO FUTURO Conclusiones Método de sintonía: optimización: determinar el valor de los 5 parámetros para satisfacer 5 especificaciones de diseño. Método de autosintonía: especificaciones en frecuencia: Test del relé + compensación en atraso (conseguir fase plana) + compensación en adelanto (conseguir especificaciones respetando robustez). Implantaciones en PC y PLC. Trabajo futuro Otras especificaciones Aproximaciones adecuadas para autosintonía: pulsar un botón. 33