You are on page 1of 8

Electrical Power and Energy Systems 33 (2011) 16231630

Contents lists available at SciVerse ScienceDirect

Electrical Power and Energy Systems

journal homepage:

Control of a DC/DC converter by fuzzy controller for a solar pumping system

N. Mazouz , A. Midoun
University of Sciences and Technology of Oran, Electrical Engineering Faculty, LEPES Laboratory, P.O. Box 1505, El Mnaouer USTO, Oran, Algeria

a r t i c l e

i n f o

a b s t r a c t
The exploitation of the solar energy is very signicant for the very sunny countries. Moreover the dryness phenomenon in these country is imposes more and more the use of pumping plants. The storage of the water in insulated basins from the wells has a double advantage. On the one hand, it is a technical storage solution of the solar energy collected by the photovoltaic panels. On the other hand, it is a hygienic way out to supply water for the rural population. In our work, we propose a technique for the identication of the maximum power point (MPP) based on fuzzy logic. This method is used to generate the cyclic ratio to operate the switcher within the maximum power of a photovoltaic array (PVA). For simulation purpose we made a complete modeling of the entire system. The system carried out consists of a photovoltaic array supplying, through a DC converter, a direct current (DC) engine coupled to a centrifugal pump. Our experimental bench consists of two principal units. A DC converter module composed of IGBT power transistors. And a processing module connected to a PC serial port, handling the input signals delivered by photovoltaic generator and controlling the power unit. The obtained experimental results conrm the simulation result which is very satisfactory and show the utility of the fuzzy controller for the optimization of the system. 2011 Elsevier Ltd. All rights reserved.

Article history: Received 11 January 2010 Received in revised form 2 February 2011 Accepted 3 June 2011 Available online 7 October 2011 Keywords: Fuzzy logic Maximum power point Photovoltaic array

1. Introduction The PVA is a system providing a non-linear power. It is required a real time identication and the tracking of the maximum operation point. This maximum power point varies largely in time according to the climatic conditions such as the sunning and the temperature. When we connect a load at the outputs of a PVA, this load imposes a point of operation which is not necessarily the point of maximum power. To ensure an optimal adaptation of energy between the PVA and the load it should be introduced an adapter so that the PVA operates at its maximum power point. In our case the adapter is a DC/DC converter whose we control the cyclic ratio by regulation. The rst used method of regulation is a traditional technique. It has disadvantages, such as the oscillations around the point of operation, and the choice of the step of the tracking of the maximum point. To go beyond these problems we chose a technique more powerful, the fuzzy logic. In this technique, we used two method of regulation, the rst one that of the variation of the input voltage by xing the optimal voltage, and the second one that of the variation of power per the current (dP/dI), the two methods

led to good results (i.e. good adaptation). The simulation of the system (Fig. 1) was carried out in [1]. 2. Followed process 2.1. Structure of the developed set up The system consists of a PVA supplying a DC engine coupled to a centrifugal pump, through a DC converter, allowing the tracking of the optimum operation point. The developed controller is a microcontroller based board connected to a PC through the serial port for monitoring purpose (Fig. 3). 2.2. Presentation of the simulated system The block diagram of the maximum power point tracking (MPPT) system is composed of a PVA, a DC converter and a load represented by an engine coupled to a pump. The point of optimum power is controlled by the cyclic ratio generated by the fuzzy controller whose the inputs are the voltage and the current of the PVA (Fig. 1). Before any synthesis of a control law, it is necessary to analyze the process to be controlled and establish an appropriate model. Regarding the PVA, we considered as an inputs, the current and the voltage measured experimentally (I1, V1) with specic climatic conditions of operation, sunning and temperature, respectively.

Corresponding author.
E-mail address: (N. Mazouz). 0142-0615/$ - see front matter 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijepes.2011.06.016


N. Mazouz, A. Midoun / Electrical Power and Energy Systems 33 (2011) 16231630

Fig. 1. Block diagram of the MPPT.

The PVA includes 20 modules (LA361K51, SI-polycrystalline, Vopen circuit = 21.2 V, Ishort circuit = 3.25A, PMaximum = 51 Wp), each 10 are connected in series, and each module contains 36 cells. The characteristic (I1, V1) obtained experimentally does not ensure a nominal operation at the optimum power point for our load, from where the need for the dimensioning of the PVA. The dimensioning of the characteristic by simulation consisted in multiplying the currents and voltages of the characteristic raised by suitably selected coefcients Ki and Kv. With Ki = 2 and Kv = 10. Starting from a characteristic of reference experimentally obtained at the temperature T1 and the E1 sunning. The translation of formulas (1), (2) would provide the characteristics at the T2 temperature and the E2 sunning.

Fig. 3. Test bench set up.


dX Cc dt

These parameters are determined from information recorded on the motor nameplate or from experimental tests made on the engine. The values of the plate are given for nominal operating conditions:  Va = 180 V.  Ia = 4.9 A.  X = 1750 r/mn. The counter electromotive force:

I2 I1 I E2 =E1 1 a T 2 T 1 CC V 2 V 1 b T 2 T 1 RS I2 I1 K I2 T 2 T 1

1 2

  P NnU a

with a = 1.6e3 A/C (coefcient which takes into account the inuence of the temperature on the current), b = 7.8e2 V/C (coefcient which takes account of the effect of the temperature on the voltage), Rs = 0.4 5 X (serial resistor), K = 5.5e3 X/C (factor of correction of the curve), and ICC (A) is the current of short-circuit measured at the outputs of the solar panel, (I1, V1) is a characteristic of reference at E1 = 60% and T1 = 30 C. The used engine is a kind of DC motor with permanent magnet. His model is dened by the following two equations: The equation of the electrical circuit of armature.

N is the number of conductors, n is the rotational speed of the motor shaft (r/s), U the ow from a pole, a is the number of pair of way of rolling up (way of rolling up: the circuits that are parallel to the armature brushes), P the number of pole pairs, the voltage constant is expressed by K.




We have:

V a Ra Ia K UX La

dIa dt

e K UX 3
 The useful power:

7 8 9

The mechanical equation

Pu C X eI
 The magnetic torque:


The used pump is centrifugal type. Its torque is a function of speed X, and is expressed by the following equation:

C c a bXn


The moment of inertia is approximated to JP = 5 Jm, where Jm is the moment of inertia of the engine. The parameters a, b and n must be chosen so that at the speed is equal to the nominal speed Xn, and the torque of the pump is equal to the nominal torque CmN [7,8].

C c X Xn C mN


C mN K mN XN 6 N m


Fig. 2. Diagram of the buck converter.

We have used a step down converter, the model is dened in Fig. 2. It symbolizes our converter by the switch S.

N. Mazouz, A. Midoun / Electrical Power and Energy Systems 33 (2011) 16231630


First case: 0 < t < Ton: The switch S is closed, The voltage is given by the following equation:


diL V L t dt


Second case: Ton < t < T: The switch S is open, The voltage is given by the following equation:

RL i L L

diL V L t 0 dt 1 T Z
0 Ton


V Lmoy



V Lmoy

  Ton Ea T


E is the voltage of the PVA, a is the cyclic ratio or hash rate ranging from 0 to 1. The frequency of the converter was set to 20 kHz. The variation in cyclic ratio will be made such that VLmoy equal the optimum voltage Vopt of the PVA. 2.2.1. Digital simulation of a direct coupling (without regulation) During the launching phase (Fig. 4) the engine/load assembly need a large amount of current, which pushes the point of operation towards the area of the short-circuit of the PVA. This situation disappears in the established mode, in which the load is some were in the area of the slope of the PVA (not necessarily at the point of maximum operation). Exceptional cases can appear, for example an operation with weak sunning or high load. In this case the load operates in the area of short-circuit and thus practically at almost zero voltage, which does not make it possible to have a signicant couple to drive the load. A regulation trying to bring back the point of operation to the optimal point is strongly recommended to optimize the effectiveness of the system (operation in point of maximum power) and thus to ensure a sufcient couple to drive the load. The change of sunning causes the displacement of the point of operation on another characteristic I = f (V), which involves an increase of the load voltage and thus an increase of the speed of the engine, the current does not vary considerably. For the high loads, the drawn current is large, which brings back the current of the PVA towards the area of short-circuit. When the load is decreased, the call of current decreases also, which causes

an increase in the output voltage of the PVA, and consequently an increase of the engine speed. It is also noticed that for signicant sunning the point of optimum capacity approaches the boundary point of operation to a = 1. To bring a good margin between these two points of the characteristic, we slightly increased the engine load and we will show it later in the results that the increase in load extends the range values of the operation points of the system. We note that the optimum capacity varies with the sunning, but this power must always be close to the nominal power of the engine. By considering that the sunning is most of the day ranging between 60% and 80%, we can evaluated the range of the electric parameters of operation (power, current, voltage) see (Table 2). The simulations show the need for a tracking to exploit to the maximum the PVA. In the following, the detailed study of the regulation techniques used will be presented. 2.2.2. Experimental and digital simulation system with regulation Several techniques are used for the MPPT. The Hill Climbing technique is based on the derivative of the power to the voltage which is equal to zero within the MPP [911]. The double feedback control technique which uses the fact that the voltage of the MPP tends to being sufciently close with a certain xed percentage to the voltage of the open circuit of the PVA. There are also, sophisticated methods like the use of fuzzy logic, the use of DSP and neurons arrays. In our work, we propose some of methods of identication of the MPP. Who will be used to generate the cyclic ratio to operate the chopper for a maximum power of the PVA. Traditional technique. This method allows the optimal point tracking of a PVA by analyzing the difference in power between two points of characteristic IV of a PVA [3]. We take two measurements of the power, P1 and P2. We calculate the power difference DP. If DP is positive amount we decrease the step of displacement. If DP is negative amount, we increase this step. We carried out simulation tests, the obtained results, are presented in the gures below. Figs. 5a and 5b show the variation of

Fig. 4. Transient answers of the direct coupling system at E = 60%.


N. Mazouz, A. Midoun / Electrical Power and Energy Systems 33 (2011) 16231630

Fig. 5a. Small step tracking.

ator, And also by its amplitude about the degree of proximity to this point(Figs. 6). From these observations a procedure rises for the tracking of the optimum capacity which consists in varying the cyclic ratio in the opposite direction of the sign of dP/dI. This variation will be weaker as the point of operation will be more close to the point of optimum capacity. The position of a point of operation on characteristic (IP) compared to the point of maximum power can be given according to the slope dP/dI in this point. The sign of this slope informs us of from which side of the optimal point is located the point of operation on the left or on the right of the optimal point, its amplitude indicates the degree of proximity of this point to the optimal point. In addition the variation of the cyclic ratio moves the point of operation on the curve (IP). The direction of the displacement of the operation point depends on the sign of the cyclic ratio variation, and the importance of these displacements is proportional to the amplitude of the cyclic ratio variations. The regulation is thus done by changing the cyclic ratio according to the slope dP/dI in order to bring back the point of operation on the optimal point where the slope is zero.

Fig. 6a. Characteristic (IV) relocated at T = 30, Es = 60%. Fig. 5b. Large step tracking.

the operation power during the tracking regarding small and large steps. From these gures, we can deduce that for a large tracking step, the response time of the system decreases and the oscillations in static mode increase. In the case of a small tracking step, the response time increases and the oscillations of the power around the optimal point decrease. To improve this tracking a variable step using a fuzzy logic control technique is adopted. Fuzzy control technique. Since the temperature has only a small effect on operation of the system, the only external disturbance that we consider in the following study is the variation of sunning [6]. Information on the place of the point of operation compared to the point of optimum capacity is necessary and can be known through the parameter dP/dI (slope of the curve (IP)) at the point of operation. This parameter is very useful since it informs us by his sign about the place of the point of operation compared to the point of optimum capacity. Zone 1 : operation in constant voltage generator or zone 2: operation in constant current gener-

Fig. 6b. Characteristic (PV) relocated at T = 30, Es = 60%.

N. Mazouz, A. Midoun / Electrical Power and Energy Systems 33 (2011) 16231630


This method is based on a ratio calculation between a power variation and a current variation. The block diagram of the regulator is as follows (Fig. 7): Where:


e dP=dI pk pk 1=Ik Ik 1: de ek ek 1

17 18

P is the measured power on the PVA. For the inputs variables, the error and its derivative, we chose ve subsets of triangular forms being spread out over the [1, 1] interval (Fig. 8). However the output variable, which is the result of a deduction between the two input values, representing in our controller the cyclic ratio, the room values are spread out between 0.25 and + 0.5, with seven subsets for more precision [4] (Fig. 9). In our regulation, we have used the SEGENO logic [4] whose rule is: IF the error is PM AND derivative is NG THEN the cyclic ratio is NG . Table 1 gather the whole fuzzy rules. To measure the slope dP/dI we must take two measurements of I and V brought closer in time since we seek the slope in a point. Thereafter, the regulator determines the output corresponding to the measured slope, that is to say the variation of cyclic ratio. This variation is then added to the preceding value. The variation of cyclic ratio is accompanied by a change of state which results in a displacement of the point of operation. After a time during which the point of operation will have sufciently progressed we take again the measurement of dP/dI. The cycle measurement and regulation is continuously started again in order to bring back the point of operation to the optimal point.

NL (Negative large), NA (Negative Average), AZ (Approximately Zero), PS (Positive Small), PA (Positive Average), PL (Positive Large) is the total of the subsets [2].

Table 2 The range of the electric parameters. Nominal electric quantities Power, Pm (W) Voltage of armature, Vm (V) Armature current, Im (A) 1000 180 4.9 E = 80% 650 150 4 E = 60% 450 140 3

Fig. 7. Structure of the fuzzy logic control.

We dene for the inputs and outputs variables 5 and 7 functions of membership respectively. These functions of standard membership have the same symmetrical form, of width equal and each function has an overlapping of 50% with the closest functions. Also, we distribute these functions on the elds of speech so that the condition has a zero slope corresponds to a zero increase in cyclic ratio is veried. For that, the functions of membership Z of the inputs and outputs variables must be centered. The results of the simulation of this fuzzy method are represented in Figs. 10. To highlight this technique we carried out simulations tests whose obtained results are indicated in Figs. 10 for an initial cyclic ratio set to 0.7. Both curves show the results of simulation in the case of a starting at a sunning of 600 W/m2 of sunning. It is noted that the optimum capacity is well reached. For an increase in the cyclic ratio the point of operation moves on characteristic (IP) in the direction of the decreasing voltages by exceeding the point, then returns slowly towards this point. Thus more the variation of cyclic ratio is large more the going beyond is important, and thus more the response time is large. This behavior justies why the going beyond, and thus the response time, are more important when the initial cyclic ratio is more important.

Fig. 8. Membership functions.

Fig. 9. Fuzzy singletons of the output.

Fig. 10a. Simulation results of optimal operation voltage.


N. Mazouz, A. Midoun / Electrical Power and Energy Systems 33 (2011) 16231630

Fig. 12. PIC control card and DC/DC converter. Fig. 10b. Simulation results of optimal operation power.

We noted that in addition to the tracking of the point of optimum capacity the regulator also makes it possible to optimize the response time and to decrease the oscillations of the power around the point optimal. The fuzzy control method of (dP/dI) converges quickly because of the computation precision of the error and its derivative. Comparatively to the traditional method witch need the choosing of the tracking step (small and precise) Despite of the compatibility of the performances, the tracking obtained results from these methods are satisfactory. The experimental implementation of the algorithm of the fuzzy controller on a Micro chip family microcontroller based board is represented in Fig. 11. To highlight this technique we carried out an experimental test with an initial cyclic ratio of 0.7. To interface the PV array output to DC motor driven centrifugal pump, a microcontroller-based DC/DC buck converter was designed and built (Fig. 3). By measuring PV array voltage and current, voltage generated by an analog tachometer, and voltage produced by a digital manometer, we can easily determine PV array output power, rotational speed of the DC motor-pump, and pump outlet pressure respectively. After that the controller on chip 10-bit PWM generator output drives the DC/DC buck converter according to each algorithm. The buck converter comprises:

MOSFET switch IRF740, diode BYT08, coil (L = 100 lH) and PV array voltage ltering capacity (C = 1000 lF) [12]. The switching frequency (20 kHz) is designed to obtain low output ripple. PIC microcontroller can send data to PC with a line driver/receiver chip (MAX232) and a null modem cable. The test of MPPT algorithms was conducted on sunny day (Average insolation: 600 W/m2, average temperature: 30 C).

Fig. 13. DC motor-pump.

Fig. 11. Experimental test bench.

Fig. 14. Optimal operation voltage.

N. Mazouz, A. Midoun / Electrical Power and Energy Systems 33 (2011) 16231630


According to the PV array output power and voltage information, the microcontroller computes the output and generates a command representing the duty cycle given by the microcontroller PWM pin which is isolated by an optocoupler (6N135), amplied by a hex buffer-inverter converter (HEF4049) and applied to MOSFET switch. The converter duty cycle is adjusted such that maximum PV array output power is extracted under all operating conditions and transferred to DC motor-pump which in turn draws water from a storage tank in a closed hydraulic system (Figs. 3 and 1113). Fig. 14 illustrates the operation of our regulator. Initially, the assembly is in open circuit, this corresponds to a voltage VOC of about 195 V. Once the program launched with a cyclic ratio of one, the voltage falls to 50 V which corresponds to the point of operation without regulator, then converges towards the optimal tension Vopt around 145 V. What shows the good performance of the tracking.

Fig. 15a, 15b, 15c show the shape of the current of the PVA and the load following an increase in the cyclic ratio. It is noticed that to provide current to the load, we should increase the cyclic ratio, which mean that we should not boost the current beyond the optimal current. More the cyclic ratio increases more the current of charge and discharge of the engine decreases. It is noticed that the consumed current by the load is almost equal to the current of short-circuit. What justies the call of current at the starting. As soon as ow is established, the current decreases and the point of operation positions on an area. In the case of a weak load (see our case), we observe the current, the voltage and the power consumption, this point is in the operation region of constant voltage generator. In the case of a strong load, we work in the operation area of constant current generator. In both cases the power consumption of the load does not match with the maximum power which can provide the photovoltaic

Fig. 15a. Engine current, PVA current for cyclic ratio = 10%.

Fig. 15b. Engine current, PVA current for cyclic ratio = 50%.

Fig. 15c. Engine current, PVA current for cyclic ratio = 80%.


N. Mazouz, A. Midoun / Electrical Power and Energy Systems 33 (2011) 16231630

generator, implying that the efciency of the system is not optimum. The introduction of the regulation loop resolves this problem and contributes to the optimization of the efciency. Before launching the operation of tracking, knowledge of the system is essential in order to control its behavior during the variation of these parameters to predict the operation of the regulator. For that we took samples of the current and voltage of the PVA, the current of the engine thus the delivery and the speed of the assembly (engine-pump) by applying various values of the cyclic ratio to the whole system carried out to a sunning of 60%. We noted that more we increase the cyclic ratio, the voltage of the PVA decreases, and the current of the PVA increases. Figs. 15a, 15b and 15c show us that after a certain value of the cyclic ratio which is about 0.8 the voltage and the current of the PVA are stabilized what carries out us to conclude that the point of operation of our load (engine and pump) is not far from the optimum point (what was proven and shown in simulation), this conclusion is much more justied during sampling of the curves of the currents of the PVA and the load. 3. Conclusion This work is a contribution to the integration of Soft-Computing and the articial intelligence in the eld of exploitation of energies with the aim to improve the performances and the optimizing of the efciency of photovoltaic array, by making them working with their maximum power, our contribution efforts are xed on the development of a linguistic tracking system based on fuzzy logic, ensuring a good adaptation of the load. One of the specicity of the fuzzy regulator proposed is that it does not require a preliminary knowledge of the sunning or the optimum power since the slope dP/dI at the point of operation is only function of this point position compared to the optimal operation point. We noted that in addition to the optimum power tracking the regulator also allows the optimization of the response time and the reduction of the power oscillations around the optimal point. The obtained experimental results show on the one hand the usefulness of the fuzzy controller for the system optimization, and on

the other hand the match with the simulation results what is very satisfactory. Technological advances always renovated in the eld of power electronics. New technologies more efcient in quality and response time are developed. We are currently designing a new structure Superbuck step down choppers whose literature is still limited, with operating mode CCM (continuous conduction mode) [5], commanded by the PCM (Peak Current Mode). References
[1] Mazouz N, Midoun A. Tracking of the optimal point of PV array through a DC/ DC buck converter in a pumping solar system by fuzzy logic. In: Proceedings of the IV international conference on engineering, 2223 November, 2005, Oran, Algeria; 2005. [2] Walker G. Evaluating MPPT converter topologies using a Matlab PV model. J Electr Electron Eng 2001;21:4955. [3] Singer S. Maximum poxer transfer from a non linear energy source to an arbitrar load. IEEE Proc 1987;134:2817. [4] Mazouz N. Tracking of the optimal point of PV array through a DC/DC buck converter in a pumping solar system by fuzzy logic. Master Thesis, July 2005, Electrical Engineering Faculty, Oran, Algeria; 2005. [5] Erickson RW, Maksimovic D. Fundamentals of power electronics. 2nd ed. University of Colorado; 2006. [6] Abd El-Shafy Nafeh A, Fahmy FH, Abou El-Zahab EM. Maximum-power operation of a stand-alone PV system using fuzzy logic control. Int J Numer Model Electron Networks Dev Fields 2002;15:38598. [7] Jafar M. A model for small-scale photovoltaic solar water pumping. Renew Energy 2000;19:8590. [8] Koutroulis E, Klaitzakis K, Voulgaris N. Development of a microcontrollerbased photovoltaic maximum power point tracking control system. IEEE Power Electron 2001;16:4654. [9] Esposito F, Isastia V, Meo S, Piegari L. An improved perturbe and observe algorithm for tracking maximum power points of photovoltaic power systems. Int Rev Model Simul 2008;0(0):106. [10] Esposito F, Isastia V, Meo S, Piegari L. A maximum power point tracking algorithm for photovoltaic power systems with the control of the output voltage. In: Proceedings of the XVII international conference on electrical machines, 2006, Chania, Crete Island, Greece; 2006. [11] Dasgupta N, Pandey A, Mukerjee AK. Voltage-sensing-based photovoltaic MPPT with improved tracking and drift avoidance capabilities. Sol Energy Mater Sol Cells 2008;92:15528. [12] Antunes F, Torres AM. A three phase grid connected PV system. IEEE Trans Ind Electron 2000;5:7238.