Enhanced-NN Digital Twin for Parameters Identification of a DC-DC Boost Converter

DC-DC boost converters are widely used in power electronics due to their high efficiency and versatility in numerous applications. Their reliability is a critical aspect and accurate parameters identification is essential for monitoring the health and performance of these converters, particularly under aging or stress conditions. In this paper, a novel methodology is presented for identifying the parameters in a DC-DC boost converter, combining neural networks (NNs) with the digital twin (DT) concept. The state equations of the converter are reformulated to separate constant parameters, represented as NN weights and biases, from time-dependent variables, which serve as inputs and targets. Once trained, the NN enables precise estimation of the converter's parameters. The obtained results confirm the effectiveness of the proposed method, with the identified parameters closely matching the expected values under different operating conditions. This approach highlights the potential of NN-based parameters identification for real-time monitoring and diagnostics of power electronic converters (PECs), offering a valuable tool for ensuring reliability and maintainability in power conversion systems.