An improved control for MPPT based on FL-PSo to minimize oscillation in photovoltaic system

Received Sep 4, 2019 Revised Nov 22, 2019 Accepted Jan 13, 2020 Photovoltaic (PV) is a source of electrical energy derived from solar energy and has a poor level of efficiency. This efficiency is influenced by PV condition, weather, and equipments like Maximum Power Point Tracking (MPPT). MPPT control is widely used to improve PV efficiency because MPPT can produce optimal power in various weather conditions. In this paper, MPPT control is performed using the Fuzzy Logic-Particle Swarm Optimization (FL-PSO) method. This FL-PSO is used to get the Maximum Power Point (MPP) and minimize the output power oscillation from PV. From the simulation results using FL-PSO, the values of voltage, and output power from the boost converter are 183.6 V, and 637.7 W, respectively. The ripple of output power from PV with FL-PSO is 69.5 W. Then, the time required by FL-PSO reaches MPP is 0.354 s. Compared with MPPT control based on the PSO method, the MPPT technique using FL-PSO indicates better performance and faster than the PSO.


INTRODUCTION
There are various types of renewable energy that coming to replace fossil-fueled plants such as photovoltaic (PV) [1] and wind turbine [2]. Lately, PV gets much attention because PV uses solar energy that holds very abundant availability to generate electricity while having environmental benefits and low maintenance costs [3,4]. However, the advantages of PV still cannot be highlighted compared to its shortcomings that the investment costs of PV are expensive and the PV power efficiency is low. The efficiency of PV is influenced by many causes such as PV condition, and weather like solar radiation, and temperature [4]. Besides the reasons above, the presence of MPPT technique also affects the power generated of the PV system. Research currently developing MPPT techniques can be implemented to optimize the power generated by PV systems such as Incremental Conductance (IC) [5], Perturb and Observe (P&O) [6][7][8], and Hill Climbing (HC) [9]. However, besides being able to produce better output, P & O and HC can produce bad oscillations and speed for the equipment while IC method can reduce oscillation, besides that, this method is susceptible to significant irradiance changes. One of the ways that can be used to overcome the problem of oscillation is using artificial intelligence.
Several artificial intelligence methods were developed to overcome oscillation problems and tracking efficiency. and Grey Wolf Optimization (GWO). NN constraints [10,11] are large amounts of data when training and flexibility of Fuzzy Logic Controller (FLC) [12] are limited due to the characteristics of non-linear solar modules. PSO [13], Firefly [14] and GWO [15] have widely used in the engineering field. In this paper, the Fuzzy Logic-Particle Swarm Optimization (FL-PSO) method is implemented to reduce ripple and oscillation with optimal Maximum Power Point (MPP) from PV. FL-PSO is a modification between Fuzzy Logic and Particle Swarm Optimization (PSO). PSO results are used as input from Fuzzy Logic. The addition of Fuzzy Logic after the PSO is to solve the problem of the value of duty cycle. So, the system can be more stable and have a fast response

RESEARCH METHOD 2.1. Photovoltaic model
In this case, several PV modules like Figure 1 are installed in parallel and in series form PV arrays to get high voltage and high current [16]. The current and voltage characteristics are generated from PV with ideal conditions (temperature 25 o C and irradiance 1000 W/m 2 ) shown in Equation 1. (1) Where, I, IL and I0 are output current, the current produced by photovoltaics and saturation current, respectively. q is element load, V is voltage between output terminals, RSH and RS are shunt resistance and series resistance, severally. n is ideal diode factor. Then, K is Boltzmann constant and T is temperature. Finally, NP and NS are the numbers of PV connected in parallel and series, respectively.

Boost converter model
The proposed boost converter is regulated by the high-frequency switch that controls the duty cycle to produce output voltage higher than the input voltage with the help of inductor and diode. like Figure 2 [17,18].

FL-PSO MPPT Algorithm
PSO is one of the artificial intelligence designed from adopting the intelligence of bird and fish colonies [19]. Equation 2 and Equation 3 show standard formulations of PSO commonly used. Where, vi and xi are the particles velocity and the particles position, respectively. c1 and c2 are the constants of positive, w is the inertia weight which affects particle velocity. Then, φ1 and φ2 are random variables between 0 and 1. Finally, pi and pg are duty cycles obtained from the particles best position and the populations best position [20]. ( 1) . ( ) ( ( )) ( ( )) Evaluation of the fitness value to update the value of pi and pg  Compare the value of each candidate pi to get the best pi value, then the best pi value compared with the pg value to get the best pg value  After the pg is obtained, the pg value is entered into the Fuzzy Logic which contains the membership function as shown in Figure 3.  If the best pg value is not obtained, then return to step 2  Repeat the iteration until it reaches the limit to get pg with the highest Maximum Power Point (MPP) or the best value. The implemented Fuzzy Logic control the duty cycle of the boost converter like Figure 3 and Table 1 [21,22]. After pg is obtained, the best pg value is made as input from Fuzzy Logic. There is one input and one output that resulted from the Fuzzy Logic.   Figure 4 is PV system which has the PV array, boost converter, MPPT, and load. The duty cycle of the boost converter is controlled by FL-PSO to increase the output voltage, minimize oscillation and reach the Maximum Power Point (MPP). Figure 5 and 6 show the characteristics of the PV module with an irradiance level of 1000 W/m 2 and a temperature of 25 o C [23]. The parameters used in this paper are four populations and one dimension. The symbols w, c1, and c2 are 0.4, 1.2 and 2, respectively. From the simulation using FL-PSO, the power of PV is 637.65 W and the power ripple is 69.5 W like Figure 7. In Figure 8, the voltage generated by the PV is 91.8 V. Figure 9, Figure 10 and Figure 11 are the duty cycle, voltage and power from the boost converter using FL-PSO, respectively. The voltage and power from the boost converter are 183.6 V, and 637.7 W. Stable conditions are reached within 0.354 s like Figure 11.  Figure 11 The boost converter power using FL-PSO Figure 12. The power of PV using PSO From the simulation based on PSO in Figure 12, the power of PV is 635.3 W and the power ripple is 72.2 W. The voltage generated by PV is 91.77 V like Figure 13. Figure 14, Figure 15, and Figure 16 are are the duty cycle, voltage and power from the boost converter based on PSO, respectively. The voltage value of the boost converter is 183.6 V. The power from the boost converter is 637.7 W and Stable conditions are reached within 1.085 s as shown in Figure 16. To compare the performance of FL-PSO, this PV system is also simulated with a standard PSO. Different results are obtained from FL-PSO and PSO. The average output power of PV with FL-PSO is greater than PSO of 2.35 W. Ripple output power of PV with FL-PSO is smaller than PSO of 2.7 W. The output voltage of PV with FL-PSO and PSO is 91.8 V and 91.77 V, respectively. Finally, FL-PSO is faster and more stable than PSO because FL-PSO reaches MPP at 0.354 s. The results of the duty cycle with FL-PSO logic are more stable than PSO as shown in Figure 9 and Figure 14.

CONCLUSION
The MPPT technique is successfully executed by FL-PSO by its faster reaching MPP, a more stable system, and smaller oscillation the output power of PV. The output power ripple of PV is reduced to 2.7 W while the average output power is 2.35 W. FL-PSO touches MPP faster at 0.354 s. The reduced oscillation before reaching steady state results in a more stable system.