Learning-based current estimation for power converters operating in continuous and discontinuous conduction modes

The problem of current estimation of switched power converters operating in continuous and discontinuous conduction modes is considered. A method is presented for the direct design of an estimator without exact knowledge of the mathematical model of the system. The structure of the proposed method is simpler than other approaches found in the literature, which use hybrid or averaged models to represent the dynamics of the power converter in each operating mode. An algorithm implementation using parallel computation and dimensionality reduction techniques for improving the execution performance is described. The method is demonstrated in the case of the pulse-width modulated SEPIC DC–DC converter, where simulation and experimental results are discussed. The proposed method shows better estimation results with respect to other well-known model-based and data-based approaches.