摘要：AbstractThe magnitude of soft error rate (SER) of integrated circuits (ICs) utilized in space missions is jeopardized due to the inconsistent intensity of radiation exposure. To protect critical electronic elements and ensure desired system performance, it is necessary to establish the real-time detection of space particle events (SPE). This research study assesses eight supervised machine learning algorithms by varying history data length (3 to 24 hours) to predict the occurrence of SPE one hour ahead. Customized SPE hourly predictor based on logistic regression is chosen for hardware implementation owing to high prediction accuracy (96.35%) as well as simplicity. After that, the optimal prototype design of the logistic regression algorithm is implemented on Field Programmable Gate Array (FPGA) with affordable hardware footprint. Finally, the digital design tested on FPGA is simulated to generate an application-specific integrated circuit (ASIC) chip layout (industrial 130 nm) integrated with SPE hourly predictor.
