An EMC Study for Renewable Energy Applications Using AI Algorithms

In this paper, it has been established that hybrid artificial intelligence (AI) is an efficient tool in performing electromagnetic compatibility (EMC) studies for power electronics systems in renewable energy applications. Indeed, a three-dimensional electromagnetic (EM) inverse method is presented for modeling and characterizing radiating sources using the near-field technique. The proposed method is based on time-domain (TD) analysis. AI-based techniques allow performing sources identification and classification with a guaranteed convergence and an adequate computing time. Indeed, the TD EM inverse method is coupled with a hybrid AI algorithm based on the implementation of a particle swarm optimization (PSO) algorithm along with a Pseudo Zernike Moment Invariant (PZMI) descriptor and an artificial neural network (ANN). The proposed method is applied to a simulation of a structure with a transient radiation behavior, such as those used in renewable energy systems. Configuration parameters are fully mastered to avoid cumulative errors related to experimental measurements. A comparative study has been carried out between reference and estimated EM field distributions using TD inverse method based on the hybrid AI method on the one hand and on genetic algorithms (GA) on the other. Obtained results prove the efficiency of the proposed algorithm, in particular in terms of convergence speed and accuracy.