Use of Artificial Neural Network to Predict Duty Ratio in Two Sub-Modes of CI-SIDO Boost Converter

This paper focuses on the coupled inductor single input dual output (CI-SIDO) boost converter, a variant of SIMO topology. The CI-SIDO boost converter operates in continuous conduction mode (CCM) and discontinuous conduction mode (DCM). In CCM, the duty ratio depends only on input and output voltages and is sector-independent. In DCM, however, the duty ratio involves complex equations with load current, input and output voltages, inductances, and coupling coefficient, varying across sectors and sub-modes. There are 21 sub-modes in 9 sectors when one boost operates in CCM and the other in DCM. The narrow load range for each sub-mode in DCM makes duty ratio determination challenging using conventional methods. This paper proposes to determine the duty ratio in DCM for CI-SIDO boost converter using an artificial neural network (ANN). Existing literature has limited applications of ANN in power electronic converters and none for DCM operation in CI-SIDO converters. This study fills that gap by proposing an ANN model to determine duty ratios, considering frequent sub-mode and sector changes under varying loads. Using the feed-forward and back-propagation algorithm, the ANN is trained for two sub-modes in one sector by varying the duty ratio, output voltage, and current in the converter in DCM while keeping the parameters of the converter in CCM fixed. After training, the ANN predicts the duty ratio for the converter in DCM based on reference output voltage and current. The predicted duty ratio is applied to a MATLAB/Simulink model of the CI-SIDO boost converter to verify output voltage and current. The predicted duty ratio is also compared with theoretical values for the same sub-mode and circuit conditions.