Multilayer Artificial Approach for Estimating Optimal Solar PV System Power Using the MPPT Technique

Artificial neural networks (ANNs) are widely recognized as technologies for solving complex problems and they have been applied effectively in several fields. Today, ANNs go beyond the limits of the conventional approaches by extracting the desired information directly from the available data. The prediction and estimation performance for this technique in the context of PV systems is essential for choosing it because it has a significant impact on the quality of the production of electrical energy and therefore on the efficiency of the PV system. After analyzing the DC/DC control loop which includes the maximum power point (MPP) controller, this work focuses on the analysis of the estimation and the prediction of errors through the new "Multi-Input Multi-Layer One-Output (MIMLOO)" approach using the ANN technique. It was developed in a model based on either a single hidden layer or two hidden layers by using a scheme of multiple inputs to one output (a "many-to-one" relationship). Its goal is to estimate the performance of the employed method by modifying the number of hidden layers and the types of algorithms that were implemented in the context of the analysed PV system. For this purpose, the performance of the proposed approach was estimated based on a real database acquired from the "SOLON 55W" PV panel. The simulation results obtained in MATLAB / Simulink show the efficiency and robustness of this approach for the "SOLON 55W" PV system.