A Gramian Angular Summation Field-Convolutional Neural Network-Based Openswitch Fault Detection Technique for Interfacing Inverters in Microgrids

This paper proposes a novel adaptive open-switch fault detection technique using Gramian Angular Summation Field (GASF) and Convolutional Neural Network (CNN) for renewable distributed generation (DG)'s interfacing inverters in microgrids. Since the inverter control is essential in microgrids, three different inverter control schemes (droop control, Virtual Synchronous Generator (VSG) control, and VSG with a Fuzzy secondary controller) are evaluated regarding their impacts on open-switch fault detection. This paper uses a novel dataset measured through Opal-RT real-time simulator in our lab for a three-phase voltage source inverter (VSI) in a microgrid under various healthy, single- and multi-open-switch fault conditions, where the three control schemes and different inverter loadings are also implemented simultaneously. The measured time series inverter three-phase currents serve as signals. These current signals are first converted into images using GASF, the images are then fed into the CNN model for fault classification. The proposed method can accurately detect the type and location of inverter open-switch faults under various control schemes and inverter loadings.