Design and Simulation of a Predictive System to Determine the Basic Factors of Solar Cells Using Neural Networks

There are different sustainable energy sources in the word. The solar energy is the most global source that contains an enormous of energy. The Photovoltaics (PV) is a tool that converts the sunlight into electrical power to supply different types of buildings. In this paper, a proposed Neural Networks (NN) tool is used to determine the main parameters and required equipment of the PV system to produce the power for a given building or city. Fives input are chosen for NN which are: the load of the building, sunlight intensity, mean time of available sunlight per day, dust ratio and the air humidity. The input parameters were chosen for its effectiveness on the PV capabilities, the numerical values of the input are normalized to unity number and general use of the proposed method. The response of the NN will determine the required PV no. and its connection type in addition to the size of batteries that will save the produced power and finally the main mechanical variables of the PV installation. The error ratio of the work was 7*10^-3 mean about 100 Epoch of each input pattern. The test and validation were held by using MATLAB 2012a which reached the validity of "0.034", zero is optimum value, after nearly 16 Epoch. The flexible treatment and simple construction are the main things of choosing NN in the design and control systems.