High performance adaptive maximum power point tracking technique for off-grid photovoltaic systems

Solar photovoltaic (PV) energy has met great attention in the electrical power generation field for its many advantages in both on and off-grid applications. The requirement for higher proficiency from the PV system to reap the energy requires maximum power point tracking techniques (MPPT). This paper presents an adaptive MPPT of a stand-alone PV system using an updated PI controller optimized by harmony search (HS). A lookup table is formed for the temperature and irradiance with the corresponding voltage at MPP (VMPP). This voltage is considered as the updated reference voltage required for MPP at each temperature and irradiance. The difference between this updated reference voltage at MPP and the variable PV voltage due to changing the environmental conditions is used to stimulate PI controller optimized by HS to update the duty cycle (D) of the DC–DC converter. The temperature, irradiance, and corresponding duty cycle at MPP are utilized to convert this MPP technique into an adaptive one without the PI controllers' need. An experimental implementation of the proposed adaptive MPPT is introduced to test the simulation results' validity at different irradiance and temperature levels.


Problem formulation
The proposed system, Fig. 1, comprises a PV module connected to a DC-DC converter feeding a load. The DC-DC converter's duty cycle ratio is controlled using an optimized PI controller to maintain MPP of the PV system. A lookup table (Table 1) consists of the temperature, irradiance and the corresponding voltage at MPP obtained from the characteristics is formed. This table is used for updating the reference voltage (VMPP_ref) for MPP at each temperature and irradiance. The PV voltage (Vpv) is sensed and compared to the updated reference voltage. This error is minimized by HS optimization technique and used to stimulate the DC-DC chopper to track MPP 36-39 . PV modeling. The PV equivalent circuit is given in Fig. 2. This circuit comprises a light-dependent current source with a shunted diode 6,7 . The current produced by this current source is directly proportional to the light dropped on the PV. R S and R p are series and shunt resistances of PV, respectively. In this paper, the EBS solar module is used and the specifications of this module are given in Table 1 at standard conditions of 25 °C temperature and irradiance of 1000 W/m 2 .
PI controller design. The updated reference voltage, V MPP_ref represents the voltage at which MPP is attained is used and compared to the current PV voltage, V pv . The difference between these two voltages ev rep-  www.nature.com/scientificreports/ resents the error used to drive PI controller. An objective function, J is formed based on the integral square of error of ev that can be defined as: The HS optimization technique is used to minimize this objective function and determine the optimal PI control parameters at each irradiance and temperature.
Harmony search optimization. The first step in HS is the initialization of the two PI control parameters by supposing random values of them. Then the fitness function J, given in (1) is determined with these random values. Then the Harmony Memory (HM) is set as in (2) The HM matrix contains two columns for the PI controller parameters (K p , K i ). Each row in this matrix is the enhanced harmony vector based on the harmony memory considering rate (HMCR) and pitch adjusting rate (PAR). The updated harmony vector x n+1 i of the current harmony vector x n i is calculated as: where rand is a random value between (0, 1) and BW is an arbitrary distance bandwidth. The updating process is done again till reaching the minimum possible value for the objective function J or reaching the maximum search iteration numbers, Fig. 3 36 .

Simulation results
Harmony search optimization technique. Due to the temperature and irradiance variations, the PV output voltage will change to new values different from the required values for MPP. These variables PV voltages will be compared to the voltage required for MPPT at these environmental conditions as in Table 2. The error between the PV voltage and the updated reference voltage based on Table 2 is used to drive PI controller to produce the duty cycle required for the converter. The HS optimization algorithm is used to minimize the objective function given in (1) and then determine the optimal PI controller parameters. HS optimization is used 30 times to obtain the optimal PI controller parameters for each case. The convergences of the first three cases are given in Fig. 4.
The optimized PI controller parameters using HS for the 30 cases proposed are given in Table 2.
Adaptive maximum power technique proposed. The proposed technique PI-HS for MPPT will be compared to P&O and IC methods to study the proposed method's effectiveness. At the standard conditions of 25 °C and 1000 W/m 2 (point 26 in the table), the MPP with P&O, IC and PI-HS are plotted in Fig. 6. The performance of the PV system with the proposed technique surplus the performance with P&O and IC methods. There are fluctuations between 33.4 and 34.9 W in power generated from the PV were obtained when using P&O; slight improvement in these fluctuations was attained when using IC as in Fig. 5. The proposed HS-PI reached to the maximum power value of 35 W without any fluctuations. The performance of the proposed system will be investigated with some changes in environmental conditions. A sudden change in the temperature from 37 to 31 °C and a sudden change in irradiance from 1000 to 800 W/m 2 (change from point 30 to point 22 in Table 2) occurs at 1.17 s. The proposed technique gives better performance than P&O and IC techniques in both steady-state and transient intervals, Fig. 6.
The PI controller introduced in this article succeeded in achieving the MPPT with better performance. Another test case will be presented to check if the optimal PI controller parameters determine at certain conditions will satisfy the optimality at other operating conditions or not.
The standard test case (point 26) will be tested with two different values for the optimal PI controllers. The first PI controller, PI-HS-1 is the optimal controller parameters determined at the standard condition (2.497, 13.986), while the second controller PI-HS-2 is the PI control parameters at another point, point 24 as an example, with PI controller parameters of (8.486, 12.548). The PI controller parameters obtained for this case showed better performance and tracking the MPP to a higher level when using PI-HS-1 at which the operating conditions are considered than using PI-HS-2. This calls for changing the PI control parameters for each operating condition Fig. 7. In other words, convert the fixed PI controller parameters into variable ones based on the environmental conditions. This transformation is difficult to be applied practically. This issue had been addressed by adding another technique to the system and implement it practically.  www.nature.com/scientificreports/

Testing the proposed MPPT algorithm Experimental verification of P-V and I-V characteristics. Experimental test rig outdoor atrium in is
shown Fig. 8. A solar panel is mounted at an optimum angle to get the maximum light at peak hours. An irradiance sensor is mounted at the same angle as the solar panel to get an accurate response. A temperature sensor is placed on the solar panel to get the panel's current temperature and find out its dependency on the temperature. The experimental verification of P-V and I-V characteristics of the solar cell used is given in Fig. 9a,b at 25 °C and 1000 W/m 2 . This experimental verification is used to verify the simulation and will be used many times at different environmental conditions to determine the voltage at MPP, which is, in turn, the updating the reference voltage (V MPP_ref ).
The experimental set-up contains a solar panel, sensors for temperature and irradiance, buck converter, PWM generation and lab-view front panel. The updated duty cycle at MPP and the corresponding temperatures and irradiances are used and added to the Lab-view. At any values of irradiances and temperatures, the optimized duty cycle at MPP is the output from the PWM generation to make the PV system operates at the MPP conditions. Adaptive MPPT verification. The model is simulated for 30 cases, and each case has a certain temperature and irradiance to get the voltage at MPP from the P-V characteristics. The temperature is taken from 22 to 37 °C with a step of 3 °C while the irradiance is taken from 200 to 1000 W/m 2 with a step of 200 W/m 2 , Table 2. From these data, the reference voltage is updated at each irradiance and temperature.
The modification technique is applied practically by determining the duty cycle required for MPPT corresponding to each operating condition. In other words, the output of PI-HS is determined at each point in Table 2. A new table is formed with the temperature, irradiance and the corresponding duty cycle. This technique is working without using PI controllers, which simplify the system and make it easy to implement.   www.nature.com/scientificreports/ In order to test the proposed technique for adaptive MPPT, a test setup was developed that was capable of data acquisition in real-time with parameter recording functionality. The block diagram of the system is shown the Fig. 10.
The block diagram shows the logical connections of different blocks of the system. Si-420TC irradiance sensor is used to sense light, which provides 4-20 mA current output, and is compensated by a built-in thermal compensator. Moreover, it has a wide range of spectral responses from 0 to 1200 W/m 2 . Besides, three-wire RTD is used to monitor solar panels' temperature in a wide range with high accuracy. Furthermore, National Instruments robust and precise current and voltage sensing modules NI9217 and NI9225 are used to monitor input and output parameters. PWM National Instrument X-series card is used for high frequency, which has 100 MHz on board time base for accurate duty cycle control and frequency generation.
The switching circuit is based on a single MOSFET DC-DC buck topology. N-Channel MOSFET is used on high side. Since the gate is at a higher voltage level, a gate driver circuit is used to drive the MOSFET with respect to floating ground. Line fuses are used on both sides of the circuit for protection. A voltage sensor was placed to put recording the input and output values. The schematic is shown in Fig. 11.
The list of components used along with their values is given in Table 3.
LabVIEW is used to develop the main software to test the proposed adaptive algorithm. LabVIEW Front panel shows input and output voltage; temperature, irradiance and power.   www.nature.com/scientificreports/ The test ring is left for 1 day for about 3 min outdoor, the irradiance and the temperature are recorded as in Fig. 12 using the temperature and irradiances sensors used in the experimental setup shown in Fig. 8.
The system is tested with the proposed adaptive MPPT with an adaptive duty cycle and compared to a fixed duty cycle for driving the DC-DC chopper.
The input and output power for the same environmental conditions, Fig. 13 are given in Fig. 13. The system efficiency with this adaptive duty cycle is improved with values greater than 90%.
Another practical test case for fixed duty cycles is applied and the efficiency of the system will be investigated. For a certain period, the input, output power of PV and the efficiency are recorded. The duty cycle in this case is kept at fixed values not adapted according to the proposed technique. Five values of duty cycles of 0.5, 0.6, 0.7 0.8 and 0.9 are adjusted at each interval, Fig. 14. The system's maximum efficiency is 84% compared to the range of efficiencies between 91 and 93% when using the adaptive duty cycle as in Fig. 13.

Conclusions
This paper introduced a new adaptive MPP technique for a standalone PV system. This technique is based on updating the reference voltage for MPPT based on the environmental changes including irradiance and temperature. This updated voltage is compared to the PV output voltage and the error is used to drive the DC-DC converter. This error is minimized by harmony search optimization technique. The proposed technique updated the PI controller parameters and consequently the duty cycle required for the converter. The proposed adaptive technique gave a better performance than P&O and IC techniques in terms of system efficiency. An experimental setup is used to simplify the controller by lookup table consisting of the temperature, irradiance, and corresponding duty cycle required for each operating condition to achieve MPPT. Compassion between the proposed adaptive and some fixed values for the duty cycles necessary for MPPT was introduced at every operating point. This comparison proved the effectiveness of the proposed method in MPPT with higher efficiencies.  www.nature.com/scientificreports/