Maximum Power Point Tracking with Constant Power Generation of Photovoltaic Systems Based on Artificial Neural Network Under Partial Shading Condition

Abstract

Solar panels are widely used by the community to support reducing the adverse effects of change. Installation of solar panels is mostly done in large buildings in urban areas. There are several obstacles to installing solar panels. One of the obstacles to installing solar panels is the shadow of objects around installation of solar panels. Many buildings now use solar panels but the installation of rooftop solar panels does not get maximum energy because the solar panels are covered by the shadows of other tall buildings and shadows from trees. One of the problems in the solar power generation system is the instability of the power generated by the solar panels caused by the influence of shadows on the PV. To overcome this problem, the Maximum Power Point Tracker (MPPT) method is used which can make PV operate at the GMPP point so that the power generated is maximized. However, the maximum power in the PV can cause an overvoltage in the power grid. Therefore, the Constant Power Generation (CPG) method is used when it reaches the rated voltage to avoid overvoltage in the load. MPPT works when the output power of the PV is less than 100 watts. While CPG works when the power exceeds 100 watts so that the power remains constant at 100 watts. The power that exceeds the load power causes damage to the load so that the power needs to be maintained according to the power required by the load. The method used is the Artificial Neural Network (ANN) method where the SEPIC Converter duty cycle can be adjusted continuously to get the maximum solar panel output power. The results showed that the Artificial Neural Network was able to track GMPP in partial shading conditions. The partial shading condition causes the power to be trapped in the LMPP so that the use of the ANN algorithm helps to get the GMPP value. From the results of data collection, the output power obtained is limited to a value of 100 watts. The use of the Artificial Neural Network method can reach a maximum power of 5.31% greater than without the method.