ANN Based Model-Free Sliding Mode Control for Grid-Connected Compact Multilevel Converters: An Experimental Validation

Sensitivity to parametric uncertainties and mismatches is an unpleasant fact that significantly debilitates the reliability and the efficiency of model based controllers in dealing with power converters known as variable structure systems. The parametric sensitivity is adversely escalated in the case of multilevel converters since more active and passive components are used to increase the voltage levels. Conventional first order Sliding Mode Control (CSMC) is a simple but robust model based technique, which can dominate the parameters variations effects on the control loop stability by considering the parameters as bounded variables instead of constants. However, designing CSMC becomes difficult and chattering as a damaging phenomenon rapidly grows up in the presence of the parameters variations and disturbances. Accordingly, this paper is meant to investigate the possibility of modifying the CSMC equations using artificial neural network so that it becomes insensitive to the model uncertainties and mismatches. Several convincing simulations and experiments are also considered to assess the proposed model-free SMC technique in practice.