Processing pipelines for simulated data (left) and observation data (right). The simulated data pipeline simulates visibilities, adds system temperature and RFI of varying intensities, and detects outliers from gain-calibration solutions using winsorizing, which is evaluated using FP, TP, and F1 scores. The observation pipeline performs calibration, outlier detection using the gain-calibration solutions, additional flagging, and recalibration. Then it images the data with WSClean to construct power spectra for evaluation.