Operational data for fault prognosis in particle accelerators with machine learning.

This paper presents real operational data collected from the power systems of the Spallation Neutron Source facility, which provides the most intense neutron beam in the world. The authors have used a radio-frequency test facility (RFTF) and simulated system failures in the lab without causing a catastrophic system failure. Waveform signals have been collected from the RFTF normal operation as well as during fault induction efforts. The dataset provides a significant amount of normal and faulty signals for the training of statistical or machine learning models. Then, the authors performed 21 test experiments, where the faults are slowly induced into the RFTF system for the purpose of testing the models in fault prognosis to detect and prevent impending faults. The test experiments include interesting combinations of magnetic flux compensation and start pulse width adjustments, which cause gradual deterioration in the waveforms (e.g., system output voltage, system output current, insulated-gate bipolar transistor currents, magnetic fluxes), which mimic the fault scenarios. Accordingly, this dataset can be valuable for developing models to predict impending fault scenarios in power systems in general and in particle accelerators in specific. All experiments occurred in the Spallation Neutron Source facility of Oak Ridge National Laboratory in Oak Ridge, Tennessee of the United States in July 2022.

Real electronic signal data from particle accelerator power systems for machine learning anomaly detection.

This article describes real time series datasets collected from the high voltage converter modulators (HVCM) of the Spallation Neutron Source facility. HVCMs are used to power the linear accelerator klystrons, which in turn produce the high-power radio frequency to accelerate the negative hydrogen ions (H-). Waveform signals have been collected from the operation of more than 15 HVCM systems categorized into four major subsystems during the years 2020-2022. The data collection process occurred in the Spallation Neutron Source facility of Oak Ridge, Tennessee in the United States. For each of the four subsystems, there are two datasets. The first one contains the waveform signals, while the second contains the label of the waveform, whether it has a normal or faulty signal. A variety of waveforms are included in the datasets including insulated-gate bipolar transistor (IGBT) currents in three phases, magnetic flux in the three phases, modulator current and voltage, cap bank current and voltage, and time derivative change of the modulator voltage. The datasets provided are useful to test and develop machine learning and statistical algorithms for applications related to anomaly detection, system fault detection and classification, and signal processing.