Yu-Chi Wu
a,⇑, Meng-Jen Chen
b, Sih-Hao Huang
a, Ming-Tsung Tsai
c, Chia-Huang Li
a aDepartment of Electrical Engineering, National United University, Miao-Li, Taiwan
b
Department of Electrical Engineering, National Kaohsiung University of Applied Sciences, Kaohsiung, Taiwan
c
Department of Electrical Engineering, Southern Taiwan University of Science and Technology, Tainan, Taiwan
a r t i c l e
i n f o
Article history: Received 6 August 2012
Received in revised form 3 March 2013 Accepted 13 March 2013
Available online 12 April 2013 Keywords:
Three-point-weighting method Stand-alone solar power system Maximum power point tracking Modified three-point-weighting method
a b s t r a c t
This paper proposes a three-point-weighting method that incorporates mid-point tracking to improve the drawback of the perturbation and observation method and to enhance the efficiency of the three-point-weighting method. A design was simulated with PSIM, followed by hardware tests of a stand-alone solar power system using real-time Matlab/Simulink hardware-in-the-loop, for observing the efficiency of the perturbation and observation method, the three-point-weighting method, and the proposed method. It was found that the proposed method tracked better than the three-point-weighting method, and it was capable of improving the deficiency of perturbation and observation method that has difficulty to track from the open-circuit voltage (on the right hand side of the P–V curve) as well as enhancing the pre-cision of the three-point-weighting method in the case of zero-weight.
Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction
In order to elevate the availability of solar energy to uplift the system’s total efficiency, solar power systems generally are equipped with the function of maximum power point tracking (MPPT). The existing algorithms of MPPT include the voltage feed-back method[1], the power feedback method[2], the incremental conductance method[3,4], the perturbation and observation (P&O) method [4,5], artificial neural network method [6], fuzzy logic methods [7,8], and the three-point-weighting method [9]. The voltage feedback method compares the PV output voltage or cur-rent with the voltage or curcur-rent on the maximum power point, MPP, based on a given I–V or P–V curve of the PV cell, using the sin-gle voltage or current feedback to make the system operate at the MPP by the rule of voltage control. This method features less com-putation and ease of design, but is subject to the deviation of MPP resulting from its open-circuit voltage and short-circuit current being changed by environmental factors, e.g. temperature and irra-diance[1]. The incremental conductance method requires calculat-ing the slope of I–V curve for the PV module, dI/dV, followed by approximating MPP by PI control[3], whereas it becomes a power feedback method if the control process is realized by P–V curve[2]. As for the P&O method, the direction for tracking is set by compar-ing the power and the voltage before and after perturbation. This method has two variables to compare, while the voltage feedback method has only one. The P&O method also features easy designing
and structuring, however, swing tends to happen with similar power but at different voltages, which compromises the efficiency of tracking[5,10]. The three-point-weighting method[9], which decides whether it is operating at MPP according to the position weightings of three points, thus avoids such unnecessary swing near MPP as P&O method has. This paper investigates further the incorporation of mid-point tracking in the three-point-weighting method to enhance the efficiency of MPPT.
2. System configuration and control process
The stand-alone solar power system is an energy system that can store the electricity generated by solar panel in batteries and supply energy to loads at night or when sunshine is not enough
[11].Fig. 1shows a typical stand-alone system of interest herein. Because the PV output voltage in this system is higher than the battery voltage, a DC/DC buck converter is used. However, some other types of converters can be chosen to meet different voltage requirements, such as boost converter or buck–boost converter. Through the voltage controller, it is possible, when the maximum power point is reached, to operate the converter switch to dispatch power from battery to the load economically. The reference voltage to the voltage controller is from the MPPT. The control process in
Fig. 1is described as follows. 2.1. Control process of MPPT controller
As shown inFig. 2, a voltage sensor and a current sensor detect the output voltage and current from the solar panel; the detected 0142-0615/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijepes.2013.03.008
⇑Corresponding author. Tel.: +886 939967722; fax: +886 37327887. E-mail address:[email protected](Y.-C. Wu).
voltage and current are sent to an MPPT controller, whereby to output the reference control voltage, Vpc, which is set according to the solar panel MPP, such that the output power of solar panel operates at the maximum point.
2.2. Switching signal control process
AsFig. 3shows, a voltage Vpfis obtained after Vpis amplified by a feedback gain; it makes the solar panel operate at MPP through the voltage controller and outputs correct control signal Vcon to actuate the switch.
2.3. Design of voltage controller
The battery herein operates at 12.2 V and the MPP of PV module is 20.52 V under irradiance 1000 W/m2hence a buck converter is needed to convert the voltage to keep the battery from being dam-aged. The equivalent circuits when buck converter switch Q is on and off are shown asFig. 4. The equations for switch Q on and off can be written as(1) and (2), respectively.
v
L¼ LdidtL¼ Vp Vb iC¼ Cddtv¼ Ip iL ReCddtvþv
¼v
p 8 > < > : ð1Þv
L¼ LdidtL¼ Vb iC¼ Cddtv¼ Ip ReCddtvþv
¼v
p 8 > < > : ð2ÞGiven a switching duty cycle, d, the average inductance voltage,
v
L, capacitance current, iC, and solar cell output voltage, Vp, are as follows: hv
LiTs¼ dv
p Vb hiCiTs¼ Ip diL hVpiTs¼ ReC _v
þv
8 > < > : ð3ÞBy the State-Space Averaging Method[12,13], the small signal transfer function can be obtained as:
~
v
pðsÞ ~ dðsÞ ¼ IpRe D s þ 1 ReC s þDVb Lip s2þReD2 L s þ D2 LC ð4ÞThe value of L can be determined by the boundary condition be-tween continuous conduction mode (CCM) and discontinuous con-duction mode (DCM), i.e., the critical concon-duction mode (CRM); C can be determined by the ripple peak, output power of PV panel, and battery rated voltage[14]. It is thus possible to design the con-troller ideal for adjusting the PV voltage. The solar panel chosen herein has an MPP at 20.52 V, 20 W under irradiance 1000 W/m2.
Fig. 1. Diagram of detailed mechanism of stand-alone solar power control system.
Fig. 2. Control process of MPPT controller.
The purpose of MPPT is to enable the PV module to provide maximum power, increasing the energy use rate of the system. Of the MPPT methods, the P&O method is a process of dynamic finding of MPP and is commonly adopted in solar power systems. It is generally realized by observing the output voltage and power before and after perturbation and spotting the MPP from the P–V curve[10,15]. In order to test the efficiency of P&O, the open-cir-cuit voltage of the PV module is used as the starting-point voltage (SPV), and the tracking begins from the right hand side of MPP. To speed up the tracking, a response regulator is designed. This
regu-over 1 K is applied with a limiter, then an integrator is used to de-crease the steady state error and to adjust the response speed (at about 0.09 times). The output of PI controller then is added to SPV to enable the MPPT within the adequate range (0 SPV0.1). AsFig. 5shows, in MPPT by P&O method simulated by PSIM, when SPV is very close to the open-circuit voltage feedback (2.17 V), Vpc and Vpfwill be unable to track to the feedback voltage value (17.9 V under irradiance 820 W/m2) of MPP in real time, which compro-mises the agility of MPPT. This paper thus considers a three-point-weighting method incorporating mid-point tracking to improve such restriction and enhance the efficiency of solar power system.
3.1. Three-point-weighting method [9]
Fig. 6 shows the three-point-weighting method for MPPT, which comprises a weight generator (Fig. 7) and the tracking re-sponse regulator.
P
V
V
pcV
pfP
V
pcV
pfFig. 5. P&O tracking result when SPV is on right hand side of MPP.
Fig. 6. Three-point-weighting method for MPPT.
from PV is not enough to provide the load; therefore, the battery discharges to provide the required power to the load.Fig. 15shows the Simulink model for the battery charging control. InFig. 15b, the ‘‘Current of lead-acid battery’’1 block takes the measured battery current signal after amplified and offset and passes it through low pass filter to prevent the control signal from noise. The ‘‘Voltage of lead-acid battery’’ block takes the measured battery voltage signal after amplified and offset and passes it through low pass filter to pre-vent the control signal from noise. The ‘‘Solar panel output voltage’’ block takes the measured PV voltage signal after amplified and offset and passes it through low pass filter to prevent the control signal from noise. The ‘‘Constant-current charge method’’ block is for constant-current charging control. The switch is to perform con-stant-current charging first when ‘‘Charge Voltage’’ (filtered battery voltage) is lower than VB_set (charge voltage set point inFig. 15a). The charging current may be smaller than the pre-set maximum charging current if the MPP of PV is not enough to supply such current. When charging, the battery voltage increases. When ‘‘Charge Voltage’’ is higher than or equal to VB_set, charging strategy is switched to constant-voltage charging to avoid over charging. The ‘‘Constant-voltage charge method’’ block is for constant-voltage charging control.
Table 2shows the test results for different irradiances (820 W/ m2and 640 W/m2), different temperatures (25 °C and 55 °C), and different SPVs (21.7 V and 14 V). The expected power inTable 2
was measured by using a solar module analyzer (PROVA 200).
Fig. 16shows the test result with Light 1 set at 10 V, irradiance at 820 W/m2(22,400 lx, blue light, luminous efficacy 27.32 lm/w), temperature at 25 °C, and SPV at 21.7 V (tracking from the right hand side of P–V curve). It reveals that the three-point-weighting method did improve the deficiency of poor efficiency of perturba-tion and observaperturba-tion method when tracking from the right hand side of P–V curve. The proposed method even effectively increases the PV output power and the tracking speed (about twice faster than three-point-weighting method).
Fig. 17shows the test result with Light 1 set at 10 V, irradiance at 820 W/m2, temperature at 25 °C, and SPV at 14 V (tracking from the left hand side of P–V curve). It shows that the three-point-weighting method tracks MPP faster than P&O method. P&O method tracking MPP from the left hand side of P–V curve per-forms better than tracking from right hand side of P–V curve. Among these three methods, the proposed method has the best
PV output power, although its tracking speed is slightly slower than three-point-weighting method.
By adjusting the light regulator, we can change the irradiance of SPT-2C.Fig. 18shows the test result with irradiance at 640 W/m2 (17,500 lx) temperature at 25 °C, and SPV at 21.7 V (tracking from the right hand side of P–V curve). It shows that the proposed meth-od outperforms the other two methmeth-ods in terms of tracking speed and PV output power.Fig. 19shows the test result with irradiance at 820 W/m2, temperature at 55 °C, and SPV at 14 V (tracking from the left hand side of P–V curve).Fig. 20shows the test result with irradiance at 820 W/m2, temperature at 55 °C, and SPV at 21.7 V (tracking from the right hand side of P–V curve). They show that the proposed method still tracks better than the other two.
5. Conclusions
This paper investigates the improvement of efficiency of MPPT algorithm by incorporating the mid-point tracking in the three-point-weighting method. It begins with designing the simulation by PSIM, followed by real-time hardware test using Matlab/Simu-link to observe the results of the P&O method, three-point-weight-ing method and the proposed method. It was found that the proposed method has higher and faster tracking efficiency than the three-point-weighting method, not only improving the defi-ciency of the P&O method being unable to start tracking from the open-circuit voltage from the right hand side of MPP but also rein-forcing the accuracy of three-point-weighting method in determin-ing MPP at zero-weightdetermin-ing condition. This proposed method is hoped to be applied to systems of higher power in the future to achieve, in combination with the power grid, practical MPPT con-trollers with even higher efficiency.
References
[1] Maheshappa HD, Nagaraju J, Krishna Murthy MV. An improved maximum power point tracker using a step-up converter with current locked loop. Renewable Energy 1998;13(2):195–201.
[2] Nafeh AA, Fahmy FH, Mahgoub OA, El-Zahab EM. Developed algorithm of maximum power tracking for stand-alone photovoltaic system. Energy Sources Jan. 1998;20:45–53.
[3] Harada K, Zhao G. Controlled power interface between solar cell and ac source. IEEE Trans Power Electron 1993;8:654–62.
[4] Houssamo Issam, Locment Fabrice, Sechilariu Manuela. Experimental analysis of impact of MPPT methods on energy efficiency for photovoltaic power systems. Int J Electr Power Energy Syst 2013;46:98–107.
[5] Lium Fangrui, Kangm Yong, Zhang Yu, Duan Shanxu. Comparison of P&O and hill climbing MPPT methods for grid-connected PV converter. In: Proc. 3rd IEEE conference on industrial electronics and applications; June 2008. p. 804–7. Fig. 20. Test results by different tracking methods with left-hand-side SPV = 21.7 V. (Remarks: under the condition of irradiance = 820 W/m2
, temperature = 55 °C, and PV converter with direct load of 30X).
1
For interpretation of color in Fig. 15, the reader is referred to the web version of this article.
