Artificial Neural Networks based Four-Leg Matrix Converter for Linear and Non-Linear Load

Abstract

The 4-leg matrix converter, which it is capable of providing a 3-phase plus neutral supply from a standard balanced 3-phase source . Tralatitious 3x3 matrix converters deliver bounded use in this application as they are only capable of supplying a poised three-phase payload. It would be suitable to be able to power unbalanced and nonlinear loads ,requiring that the converter provides a neutral connection .This goal can be achieved by extending the number of output legs in the matrix converter to four .The fourth leg of MC is required in order to provide an electrical path for the circulation of the zero -sequence current in the load. The space vector modulation technique is use to provide a switching pattern for matrix converter using ANN. The SVM technique is an extension method , which is currently used in 3x3 matrix converters, and then it accords the dissociated restraint of both the input and output .The matrix converter is a nonlinear controller because it uses nonlinear components. ANN controller is an approach to overcome the waveform quality problem. A simulation in the MATLAB / SIMULINK platform has validated the proposed idea.