考虑传感器误差的DC-DC降压变换器数字孪生参数辨识

The digital twin is becoming a promising technology for power electronic converters. Digital twin performance strongly relies on parameter identification. The accuracy of the measurements used for the commissioning and operation of the digital twin will be key for parameter identification. The use of ultrahigh-accuracy sensors and/or frequent calibration can be feasible in laboratory environments but unfeasible for most industrial environments. This article analyzes the effect of sensor errors on parameter identification and eventually on digital twin performance. It will be shown that including sensor gains and offsets as unknowns in the parameter identification algorithm significantly improves the accuracy of the estimates of critical parameters. The interior point method (IPM) and particle swarm optimization (PSO) will be considered for parameter fitting, the first being found to be superior. The proposed methodology is applied to a dc/dc buck converter. The simulation and experimental results are provided for validation.