Acta Montanistica Slovaca Ročník 8 (2003), číslo 4

Modulus optimum for digital controllers

Miluše Vítečková1 and Antonín Víteček1

Optimálny modul pre digitálne regulátory

The paper extends the popular modulus optimum with the original approach for the conventional digital controllers. The use of the D-
transform allows obtaining the uniform computing formulas for the values of the digital and analog controller tuning parameters. The use is
shown on an example.

Key words: modulus optimum, controller tuning, PID control.

Introduction

For low order controlled plants without the time delay the modulus optimum (absolute value optimum
criterion, Betrags Optimum) is often used for the conventional analog controller tuning (Åström & Hägglund,
1995; Kalaš, Jurišica & Žalman, 1978). It is so because of its simplicity and a suitable course of the step
response from the practical point of view. It is widely used in the electrical drive control where the small time
constants are substituted by their sum (Kalaš, Jurišica & Žalman, 1978; Szklarski, Jaracz & Víteček, 1989).

Modulus Optimum

The control system on the Fig. 1 is considered, where GC is the controller transfer function, GP - the plant
transfer function, W, V and Y – the transforms of the desired w, disturbance v and controlled y variables.
For control systems with analog controllers the modulus optimum is based on the requirement for the
control system transfer function to be in a form
Gwy ( s ) → 1 ⇒ Awy (ω ) → 1 ⇒ Awy
2
(ω ) → 1 (1)
where: Gwy (s ) - the control system L-transfer function, s – the complex variable in the L-transform, Awy (ω ) -
the modulus of the control system F-transfer function Gwy ( jω ) , ω - the angular frequency.
The last relation in (1) is very important because it
V allows avoiding the radical, and additionally the
W Y following holds for the modulus square
GC GP 2
Awy (ω ) = G wy ( jω )G wy (− jω ) (2)
This paper considers only cases (see Tab. 2) in
which the control systems with the analog controllers
Fig.1. Control system.
lead to the standard form of the control system L-transfer
function for the modulus optimum after a compensation
of the plant biggest time constants have been applied to them (Åström & Hägglund, 1995; Kalaš, Jurišica &
Žalman, 1978; Szklarski, Jaracz & Víteček, 1989).
1 1
Gwy ( s ) = 2 2 , ξw = =&0,707, Tw = Ti 2 (3)
Tw s + 2ξ wTw s + 1 2
where: ξ w is the damping coefficient, Tw - the control system time constant (i = 1 for 1st and 2nd row, i = 2 for
3rd and 4th row and i = 3 for 5th row in Tab. 2). It is interesting that the characteristic polynomial coefficients in
(3) correspond to the coefficients of the standard forms by Whiteley, Naslin and ITAE criterion with the step
response relative overshoot about 4.3%.
Further it is supposed that the control system D-transfer function is in the form (Vítečková at al., 2002)
β γ m + Κ + β1γ + β 0
Gwy (γ ) = m n , n≥m (4)
α nγ + Κ + α1γ + α 0
where: γ is the complex variable in the D-transform.

1
Miluše Vítečková, Antonín Víteček: Department of Control Systems and Instrumentation, VSB - Technical University of Ostrava,
Ostrava, Czech Republic [email protected], [email protected]
(Recenzovaná a revidovaná verzia dodaná 19.11.2003)

214
 Acta Montanistica Slovaca Ročník 8 (2003), číslo 4

For a small value of the sampling period T the two terms of the Taylor expansion for the approximation of
the exponential function can be used
e Ts − 1
eTs ≈ 1 + Ts ⇒ γ = ≈s (5)
T
then the relation (2) can be written as
B ω 2 m + Κ + B1ω 2 + B0
2
Awy (ω ) ≈ m 2 n (6)
Anω + Κ + A1ω 2 + A0

Tab.1. Conventional controller transfer function.

TYPE GC (γ ) GC (s ) GC (z )
1 P kP kP kP
γT + 1 1 T z
2 I
TI γ TI s TI z − 1
 T γ   T z −1
3 PD k P 1 + D  k P (1+TD s ) k P 1 + D 
 γ T + 1  T z 
 γT + 1   1   T z 
4 PI k P 1 +  k P 1 +  k P 1 + 
 TI γ   TI s 

 TI z − 1 
 γT + 1 TDγ   1   T z T z −1 
5 PID k P 1 + +  k P 1 + + TD s  k P 1 + + D 
 TI γ γT + 1   TI s   TI z − 1 T z 


Tab.2. Recommended controllers and their adjustable parameters for modulus optimum.
CONTROLLER < ANALOG DIGITAL T = 0 T > 0-
CONTROLLED PLANT
TYPE k P* TI* TD*
k1
1 I 2k1 (T1 − 0,5T )
T1s + 1
k1 1
2 P - -
s (T1s + 1) 2k1T1
k1
T1 − 0,5T
3 (T1s + 1)(T2 s + 1) PI T1 − 0,5T -
2k1T2
T1 ≥ T2
k1
1
4 s (T1 s + 1)(T2 s + 1) PD - T1 − 0,5T
2k1 (T2 + 0,5T )
T1 ≥ T2
k1
(T1s + 1)(T2 s + 1)(T3 s + 1) T1 + T2 − T 4T1T2 − 2T (T1 + T2 ) + T 2
5 PID T1 + T2 − T
2k1 (T3 + 0,5T ) 4(T1 + T2 − T )
T1 ≥ T2 ≥ T3

In the case where the controller integral time TI* and derivative time TD* are determined from the
conditions for the compensation, the controller gain k P* can be determined on the basis of the formula (Kalaš,
Jurišica & Žalman, 1978)
A1 = B1 ⇒ α12 − 2α 0α 2 = β12 − 2 β 0 β 2 (7)
The Tab. 2 shows relations of the controller's adjustable parameters after the first order Páde approximation
has been used on them.
The sampling period value T should not be bigger than about one third of the plant smallest compensated
time constant.
The relations for the controller's adjustable parameters for T > 0 are valid for the digital controllers and for
T = 0 are valid for the analog ones.

215
Vítečková and Víteček: Modulus optimum for digital controllers

Conclusions

In the paper the popular modulus optimum is derived and simplified for the conventional digital controllers.
Because the D-transform is used, the obtained formulas are uniform for digital and analog controllers. The
simulation results show the practical usefulness of the proposed approach.

The paper was supported by grant MSM 272300012.

References

ÅSTRÖM, K. & HÄGGLUND, T.: Controllers: Theory, Design, and Tuning. Instrument Society of America, 2nd
Edition. Research Triangle Park, 1995.
FILASOVÁ, A.: Robust Decentralized Regulator for System with Unknown Uncertainty Boundaries. In Process
Control ´99: Proceedings of the 12th Conference, Vol. 2. June 1999, Tatranské Matliare, STU Bratislava,
1999. pp. 86-90.
HUDZOVIČ, P. & KOZÁKOVÁ, A.: A Contribution to the Synthesis of PI Controllers. In Sborník abstraktů
konference “Kybernetika a informatika”. 5. – 6. dubna 2001, STU Bratislava, Piešťany, 2001. str. 31-34.
KALAŠ, V., JURIŠICA, L. & ŽALMAN, M.: Technical Cybernetics of Electrical Drives (in Czech). ALFA,
Bratislava, 1978.
KROKAVEC, D. & FILASOVÁ, A.: Optimal Stochastic Systems (in Slovak). 2nd Edition. Elfa, Košice, 2002.
SZKLARSKI, L., JARACZ, K. & VÍTEČEK, A.: Optimisation of Electric Drive (in Polish). Państwowe
Wydawnictwo Naukowe, Warszawa, 1989.
VÍTEČKOVÁ, M., at al.: Control System Synthesis Methods Based on the Delta Models (in Czech). Technical
Report of Grant No 102/00/0186. Ostrava, December 2002.

216