DESIGN AND ANALYSIS OF FUZZY CONTROLLERS FOR DC-DC CONVERTERS Ahmed Rubaai, Mohamed F.

Chouikha Electrical & Computer Engineering Department Howard University 2300 6'h Street, Northwest Washington, DC 20059, USA mchouikha[&howard.edu aruhaaiGilhoward.edu Abstract - A successful implementation of fuzzy controllers for DC-DC converters is presented in this paper. Two different fuzzy logic control topologies are developed and implemented using different types of DC-DC converters such as the buck, the boost, the buck-boost, and the sepic converters. Issues of sudden changes in the load or parametric uncertainties control and communication interface, among many other issues, are discussed and presented. The fundamentals governing the design, control and performance of the DC-DC converters are also illustrated. Properties of the proposed controllers are: 1) robustness around the operating point, 2 ) good performance of transient responses under varying loading conditions and/or input voltage, and 3) invariant dynamic performance in the presence of varying operating conditions. Simulation results have been obtained using appropriate scaling factors associated with the input variables of the fuzzy controller. INTRODUCTION DC-DC switching converters are a traditional benchmark for testing nonlinear controllers, due to their inherent nonlinear characteristics. After the pioneering studies of Middlebrock [I], a great deal of research has been directed at developing techniques for averaged modeling of different classes of switching converters [2] and for an automatic generation of the averaged models [3]. The motivation of such studies was the selection of continuous models as simple as possible, but adequate to capture all the main features of the switching converters in terms of stability, dynamic characteristics and effectiveness for designing closed loop regulators. A large number of possible nonlinear controllers have been proposed: among others sliding mode control strategies [4], nonlinear PI controllers based on the method of extended linearization [5] and nonlinear H, controllers [6]. A recent interesting paper [7] presents the results of an experimental comparison of five control algorithms on a boost converter: linear averaged controller, feedback linearizing controller, passivity-based controller, sliding mode controller, sliding mode plus passivity-based controller are compared along with their adaptive versions in order to cope with the parameter uncertainty due to a load resistance change. Advantages and drawbacks of the proposed control strategies are tested under a fixed output voltage with load variations. All the quoted literature comply with the more general problem of applying nonlinear control techniques to complex real world technical problems: such classical approach has undoubtedly the advantage of designing analytical controllers and to evaluate quantitatively their stability bounds. The major problem of the classical approach remains that as the complexity of system increases, our ability to make precise and yet significant statements about its behavior diminishes [8]. In our opinion the control of switching converter constitutes at the present time a borderline problem, which can be handled both with conventional nonlinear control strategies and with fuzzy logic-based technologies. Why can be fuzzy logic chosen as an alternative design method to nonlinear controllers? An important answer was given in [9]: a nonlinear controller such as

fuzzy logic can be inexpensively implemented with DSP-based

micrc-controller. As a matter of fact many researchers focused their efforts on the application of fuzzy technology for controlling switching converters. In [9] the advantages of a low cost microcontroller implementation of a fuzzy direct control were pointed out. A model- based fuzzy controller (fuzzy indirect control) for a Buck converter was proposed in [lo]. Bonissone [1 11 proposed a successful application for resonant converters, by using suitable scaling factors. In [ 1 1 1 the fuzzy controller performs a variable action depending on the difference between the desired and the actual output voltage. Such implementation considers an optimization of the scaling factors around a single output operating point. Our goal is to implement a robust fuzzy controller that can achieve the following properties: 1) Robusmess around the operating point (e.g. in the case of a load change; 2) Good dynamic performance (i.e. rise time, overshoot, settling time and limited output ripple) in the presence of input voltage variations (and load changes); and 3) Invariant dynamic performance in presence of varying operating conditions. To the best of our knowledge, property 1 has been hlfilled in all related literature. Property 2 requires the synthesis of a complex controller (fuzzy or nonlinear) able to optimize the transient performance. Property 3 (along with I and 2) implies the synthesis of a global controller, with optimized parameters for varying operating conditions. Such task seems to be extremely hard; however we believe that a complex nonlinear controller could be accomplished using --based controller. In this paper, two distinct topologies-based fuzzy logic controllers (FLCs) using different types of DC-DC converters and at different operating modes are developed and presented. In topology I, the fuzzy controller requires only sensing of one inductor current and the output voltage. However, for topology 11, the output voltage is the only variable to be monitored. For this topology, two categories of tests that cover the two basic performance areas, load regulation and line regulation, are carried out to evaluate the controller's performance. FUZZY LOGIC -BASED CONTROL TOPOLOGY I In this topology, the proposed fuzzy controller uses three input variables: 1 ) Output voltage error, e,, 2 ) Inductor current error, ei and 3) Inductor current, iL. A block diagram of the fuzzy controller structure is shown in Fig. 1 While the output voltage reference is usually available as an external signal, the inductor current reference depends on the operating point. For this reason it is computed by means of a low-pass filter in the assumption that the dc value of the current is automatically adjusted by the converter according to the power balance condition. The controller output variable is the switch duty cycle controller, which is obtained by adding the outputs of two fuzzy controllers. One fuzzy (P) gives the proportional part 6, of the duty cycle as a function ofe, = I , - i, ,eu = U , , - U , and iL. The other fuzzy (I) gives as increment of 61, which is then

0-7803-8379-6/04/$20.00 02004 IEEE.

479

integrated to provide an integral term cycle6 = 6 + 6, '

Funq-P

6 1

of the duty

be kept constant to prevent overshoot, 4) IF the set point is reached AND the output still changing, the duty cycle must be changed a little bit to prevent the output from moving away, 5) IF the set point is reached AND the output is steady, the duty cycle remains unchanged, and 6) IF the output is above the set point, the sign of the change of the duty cycle must be negative, and vice versa.
I

Fig. lFuzzy Controller-based Topology I A. Fuzzy Rules for Topology I A. lFar from the set point: When the output voltage is far from the set point (e, is PB or NB), the corrective action must be strong meaning F, should be NB or PB, while 61should be zero. The basic control rules are: IF e, is PB AND iL is NORM, THEN 6, is PB AND 6, is ZE. IF e, is NB AND iL is NORM THEN 6, is NB AND SI is ZE. This shows that far from the set point, the control action is denoted by the output voltage error, provided the existence of the current limit. A.2 Close to the Set Point: The current error must be taken properly into account in order to ensure stability and speed of response. The goal in this region is centered in achieving a satisfactory dynamic performance with small sensitivity to parameter variations. The control rules are according to energy balance and inductor current is far from the limit. IF e, AND ei are both Zero, 6, AND nd SI must be zero too (steady state condition). 1 I F the output voltage error e,is Negative AND inductor current is greater than the reference value (ei < 0), 6, and should be negative. 1 I F output voltage error is Positive AND the inductor current is greater its reference value, THEN 6 and 6 I must be kept to zero to prevent undershoot and overshoot. IF the output voltage is Positive AND the current is lower than its reference value (ei > 0), 6, and 6 must be positive, the system energy increases in this condition.

Data Base

Fig. 2 Block Diagram of Fuzzy Controller for Topology I1 SIMULATION RESULTS A. Control Topology I This control topology is tested using two different types of DC-DC converters, namely, the buck-boost converter and the sepic converter subject to step load change. For the buck-boost converter the system information are as follows: U, = 12 V, Uoref = 20 V, Iref = 3.5 A, Load changes from 20 R to 150 R and back = 20 - u,(t), El (t) = I,f -IL(t) = 3.5 to 20a, E, (t) = Uoref-Uo(t) IL(t). Figs. 3-5 show the output voltage, the inductor current, and the duty cycle as functions of the time.

FUZZY LOGIC-BASED CONTROL TOPOLOGY I1 The block diagram of the fuzzy logic control scheme of topology I1 for the DC-DC converter is shown in Fig.2. The output is the duty cycle, Fk. For this topology, there are two inputs, the voltage error e = U , -U, and the change of the voltage errorce = e,

Fig. 3 Output voltage for the buck-boost converter

- ek-,. The

term U, is the present output

voltage, and U,f is the reference output voltage. Fuzzy Rules for Topology I1 The derivation of the fuzzy control rules is heuristic and based on the following criteria: 1) IF the output of the converter is far from the set point, the change of the duty cycle must be large to bring the output to the set point quickly, 2) IF the output of the converter is approaching the set point, a small change of duty cycle is necessary, 3) IF the output of the converter is near the set point AND is approaching it rapidly, the duty cycle must

Time. m i

Fig. 4 Inductor current for the buck-boost converter

480

Case A: Load regulation of buck-boost converter; load resistance has step changes from 10 R to 5 C2 and back to 10 R Case B: Line Regulation of Buck-Boost Converter; Input has step changes from 15 V to 20 V and back to 15 V. Figs. 9 and 10 show the output voltage and the duty cycle of the buck-boost converter subject to step load change from 10 R to 5 R and back to IOR. It is shown that when the load voltage decreases from its set point at time 0.03 seconds due to increase in the load resistance, the duty cycle increases severing to stabilize the output voltage. As the output voltage increases from the set point at time 0.0325 seconds due to increase in the load resistance, the duty cycle decreases trying to stabilize the voltage again at the set point. Figs. 11-13 show the input voltage change, output voltage and the duty cycle of the buck-boost converter subject to step input voltage change from 15 V to 20 V and back to 15 V. It is noticed in this case there is more ripples than the other cases.

Fig. 5 Duty cycle of the buck-boost converter For the sepic converter the system information is as follows: U, = 15 V, Uorcf= 20 V, Ircf= 3.6 A, Load changes from 20 R to 200 R and back to 2022. Figs. 6-8 display the output voltage, the inductor current, and the duty cycle as functions of the time.

I85

10

15

20

25

Time, ms

Fig. 6 Output voltage for the sepic converter

0
0015
0025

I
0035
0045 0055

0065

Time, Sec

Fig. 9 Output voltage for buck-boost converter subject to step change in the load resistance

Time. ms

Fig. 7 Inductor current for the sepic converter

15

6 $
P)

1 0

5
0

> ,

Fig. I O Duty cycle for buck-boost converter subject to step change in the load resistance

0
p

-5
-10

-I5

1-

_ I

Time, ms

Fig. 8 Duty cycle for the sepic converter B. Control Topology I1 This control topology is tested using three types of DCDC converters, namely, buck converter, boost converter, and buck-boost converter subject to input voltage change and load impedance change. So, we will have six cases, namely:

Fig. 11 Input voltage variation for Buck-Boost Converter

48 1

4.

H. Sira-Ramirez, Sliding motions in bilinear switched networks, IEEE Trans. Circuits Systems, vol. CAS-34, pp. 919-933, Aug. 1987.

1
-

0 0.015

,
0.035
Time, sec

,
0.055

5.

H. Sira-Ramirez, Design of P-I controllers for DC-to-DC

power supplies via extended linearization, Int. J. Control, vol. 51, no. 3, pp. 601-620, 1990.

6. Fig. 12 Output voltage for buck-boost converter subject to step change in the input voltage
3 1.........................................................................................................................
?

Kugi and K. Schlacher, Nonlinear &-controller design for a DC-to-DC power converter, IEEE Trans. Contr. System Technology, vol. 7, pp. 230-237, Mar. 1999. G. Escobar, R. Ortega, H. Sira-Ramirez, J.P. Vilain and I. Zein, An experimental comparison of several nonlinear controllers for power converters, IEEE Control Systems, vol. 19, no. 1, pp. 66-82, Febr. 1999. L.A. Zadeh, Outline of a new approach to the analysis of complex systems and decision processes, IEEE Trans. System, Man & Cybernetics,vol. SMC-3, pp. 2 8 4 , 1973. T. Gupta, R.R. Boudreaux, R.M. Nelms and J. Hung, Implementation of a fuzzy controller for DC-DC converters using an inexpensive 8-b microcontroller, IEEE Trans. Industrial Electronics, vol. 44, no.5, pp. 661-669, Oct. 1997.

7.

8.

Time, ms

9.

Fig. 13 Duty cycle for buck-boost converter subject to step change in the input voltage

CONCLUSIONS Two hzzy control topologies are designed and implemented; the differences between the two structures are basically in the input variables and in the number of fuzzy logic rules. Many test cases via simulation demonstrate that the two fuzzy topologies are capably in reducing the effect of different disturbances such as load changes and input voltage changes, on different types of DC/DC converters. Simulation results show the ease of applying fuzzy control to dc/dc converters, as an interesting alternative to conventional techniques. Also, test results illustrate that the fuzzy logic approach can provide considerable control performances. Fuzzy logic appears to be a valid element for generalization to many control applications. The control topologies designed and simulated in this paper is a relatively easy to implement in the laboratory.

IO. P. Carbonell and J.L. Navarro, Local model-based fuzzy control of switch-mode DCDC converters, in Proc. 14 IFAC Triennal World Congress, pp. 237-242, 1999.
1I. P.P. Bonissone, P.S. Khedkar and M. Schutten, Fuzzy logic control of resonant converters for power supplies, in Proc. of the 4th IEEE Conference on Control Applications, pp. 323-328, 1995.

REFERENCES
R. D. Middlebrook and S. Cuk, Advances in Switched Mode Power Conversion, vol. 1 and 2, TESLAco, Pasadena, CA, 1981. S.R. Sanders, J.M. Noworolsky, X.Z. Liu and G.C. Verghese, Generalized averaging method for power conversion circuits, IEEE Trans. Power Electronics, vol. 6, pp. 251-259, Apr. 1991. J. Sun and H. Grotstollen, Symbolic analysis methods for averaged modeling of switching power converters, IEEE Trans. Power Electronics, vol. 12, pp. 537-546, May 1997.

482