Speaker
Description
In 2021, the Chinese ADS Front-end demo superconducting radio-frequency (SRF) linac, known as CAFe, successfully conducted a commissioning of a 10 mA, 200 kW continuous wave proton beam. During this commissioning, it was observed that the SRF faults are the leading causes of short machine downtime trips, contributing to approximately 70% of total beam trips. Analyzing fault data and identifying fault types is a time-consuming and laborious process, especially for large modern accelerators with hundreds of RF cavities. Here, we propose a machine learning (ML) based model for automating SRF cavity fault recognition. First, a comprehensive study of the cavity fault mechanisms was conducted, leading to the identification of several distinct fault patterns. Next, we converted the “expert reasoning” process into a “model inference” process though feature engineering. We demonstrate the feasibility of this method using the CAFE2 (CAFe had been upgraded to CAFE2 after 2021) facility, achieving an accuracy of over 90%. By combining ML mdoel with big data analysis, we identified three different mechanisms of quench at CAFE2 and developed an automatic detection algorithm to distinguish them, which is crucial for the prevention of quench. Although the specific faults in SRF cavities may vary across different accelerators, similarities exist in the RF signals. Therefore, this study provides valuable guidance for the fault analysis of the whole SRF community.